17 Episode results for "Daniel Fidel"
Streamlining Financial Reporting with AI and NLG - with Emmanuel Walckenaer of YSEOP
"This is daniel fidel with emerge artificial intelligence research. And you're listening to the a in business podcast over the years. You've been a loyal listener with us. You've heard many use cases and trend explorations for the future of a in banking future. Vi insurance and wealth management we covered how compliance workflows or changing. How customer service is changing. How fraud and anti money laundering or changing. Today we're talking about reporting our guest. This week is manual. welcome year. Who is the ceo at easy peasy op is based in paris and a focus on natural language generation. They've been at it for a while. We actually had another member of their team on. I think it was some four long years ago. Something crazy like that and our focus in this episode is getting into detail as to the specific workflows where automating reporting really delivers value. We talk about two in particular one of which is credit risk understanding in assessing credit risk. What does that process look like today. What kind of data and insights required. Where does that fit into the mix and then we also look at improving report quality for folks who are controllers or analysts. We've talked from the perspective of compliance leaders but never really from the perspective of controllers. So a little bit of uniqueness here some of the use cases that were covering and i appreciate emmanuel's detail in exploring how the use cases work and also how the underpinning data infrastructure comes together to actually allow a natural language generated report to come to be a lot of detail in this episode of excited to get started. This episode is sponsored by easy. Op if you're interested in reaching emerges on go to e. m. e. r. j. dot com slash a. d. one that's eighty like add and then the number one m. e. r. j. dot com slash eighty-one and learn more about our creative services. What it is we offer in terms of content and reach and also how brands work with emerged. They wanna learn more about that process yemi rj dot com slash eight one without further. Ado this is emanuel decio easy. Opt here on the business. Podcast so emmanuel. I know we're going to be focused on some unique use cases in the financial services face in terms of where a i is starting to find its fit and add value or to talk first about the business units that you tend to work most closely with which again are somewhat unique. We cover a lot of different areas of banking but you guys hang out around the risk and compliance world as low as the finance and controlling world. I imagined kovic is change a lot of things for these folks the banking ecosystem insurance ecosystem or undergoing a lot of change. What do you see as some of those big kind of structural shifts that are changing and imposing new demands on on those business leaders. Well thanks well. It's a it's a great question so we have been working at easy up with at twenty five now tijuana large fashion institutions both in in europe. And the us. It's true that these guys are always intrigued by You know i could help us on so you know. I've been doing this project. Are you know. A lot of them are in production and pretty happy. Now with could be done. you'll think did accelerate. I think these guys are under massive pressure. You know they have given you know. Unprecedented amount of loan to help the economy a dramatic way. I mean they have. They have to be super giant. You know the certa tastes pretty high so they clearly have a lot of customers asking me. Okay how do you know how i could help me. I've let my team. How can i do more with the team. Have and you know. Could we be way more. If he shouldn't have to jail did a prediction that twenty five percent. You know of any companies with adopts on form of language a in g union natural language generation by twenty twenty two. We plan to double our install base this year so we actually way way way bigger. Yeah two things. I think some of the benefit of these kinds of applications that we're going to get into use cases in a moment is that there are some sorts of ai. Applications that require. I think a lot more data infrastructure overhauling than some of the stuff where we can kind of take what exists so we can actually just make it more simple. It's it's a little bit. More of a a circus. Accessible application of ai i take the jackhammer to the data. Infra overhaul what we're doing and by the way there's a place for that i think banks are going to have to evolve in a bigger way than surface but some of the accessible stuff right away. I think really great opportunities for you. These increased demands. Is it just due to. How much has changed about what they're reporting on is due to how many loans they have now out in the world. How much risk they have to manage. What are the things that covance brought onto them. That's made them forced to be more agile. Course being forced to work digitally. I think is probably part of it or do you see those being forces pressure from while you know. These guys have pressure from Everywhere yeah poor. Poor regulation england that now got into customers. Yup and because some say you know you need to help us to really get more data much more. Precise is more timely. And so on the buses who wants as well. You know what he's going to have any surprise this month. There's going to be you know a big change in your lab. thomas shoulders. Got the complaints you know. The governments you know compliance is is super era is not going to be lighter in the coming you know in the company and the sometimes you know these guys are. I'm talking about finance but even risk organization you know. They would love to become way more business driven organizations. Of course they can you know. Have the right data in your get things straight and i think they are very good at that but helped transform these huge that They have on hand how to use that to make a better business. How to debuts intial customers to actually drive their business more efficient way. This is super so lots of pressure from all directions. It sounds like a little bit. Glad i'm not in that business but obviously it's very important. Facet of the financial services world is make sure our reporting is working when we get the insights. We need to move on make smarter decisions so we can talk about some of those decisions. One category that. I know we wanted to dive into here. Was looking at risk reporting specifically around of credit risk tons to think about you know so many more loans out there in the world walk us through a little bit of kind of what the business workflow is in where is starting to find. Its fit into that particular workflow. That's To use case which is getting more and more popular so just before i stopped what we do in a nutshell. We do transform based on algorithms And the i. We do transfrom data into narrative that anybody can understand the letter in this type of ad. Kimmy trying to transform that. I in so the credit crate-raised departments for retail culprits fashion institutions. You know these are huge departments within the banks of course and the year every time you do what you have to grant a new loan and then afterwards every year have to reassess the risk based on the data fashion statements to the customer based on the cash flow based junior old type of financial information. They have to reassessment and so that they can. You know do some direct provision and so on. so where are we jump in. We say guys you know we can help you to actually do sixty to seventy percent of this reports automatically. We will produce a draft for the enemies so that all the section which are data driven will be actually drafted by the machine and the ice can actually dialogue with the machine. Say okay guys. Thank you for these highlights and you know this paragraph now a little more detail so yeah got slider you know and these interface and against say okay. I so to the machine was good morning until i want less detail the east you know dialogue who is a machine and makes the report the way it wants. The reports look like super efficient. You know have been deployed that three four banks. Now they are one of the banks you know. They told us well. We cut the the amount of time to do this report by half so it's a fifty percent efficiency gain. Which is you know. Pretty nice super high volume you the typically you have got two thousand two thousand five thousand and that is involved you know and you have to do in corporate banking you can. Do you know small and bean sized business. You can do so. It's you know super away again. The volume of our super high. So the arrow i on this type of solution are you know in millions of dollars a year and so i want to talk a little bit. We'll go a little bit more into the nitty gritty of the reporting in the use case itself in just a second but i want to kind of pull up to the business value side of things. Obviously you folks have experienced deploying these so i think there's lessons learned here you. You mentioned saving fifty percent on what it takes to do this report. I think maybe not everybody listening in will understand how much has to go into news reports together. Maybe if we could if you wouldn't mind painting a bit of a picture of i need a certain kind of report on my small business loans in the state of the risk profile of my small business loans kind of a high level. What kind of data has to get pulled in. How how did that process used to work before. Because i think people might not realize how much goes into this context helpful you know. It is of the time pretty tedious tasks to warn the updated fashion formation near the band the piano cash flow. You have to compare this year. Versus last year with us changed it's super tedious. And it takes you know two three hours to do that. You know more if it is you know. Of course there'd be other companies this and then the you know. This is a data crunching and this is really putting together. Okay let us change. What is important and so on. And this is a task that i can actually do automatic. You clever way. Intelligently he's not just describing. What's us change trying to say okay. This has changed because you know finding the underlying causes and sometimes guys you know. This is very strange. I mean this move or this. Change is very unusual in the you know in this type of cell so it went you know whatever so what is key. Is that the adoption by the end. User is absolutely oh yeah has be left to You know we take them with these type of solution out of their comfort zone. We appreciate report. Which is tedious. And so they not the most important part of the job that used to do it. Yeah and they're used to using what Microsoft excel and you know that the old bunch of tools they used to use right. oh uneven. Whoever out of tools and zone are used so it's a bit of a challenge for sure netease and the way you write has to be perfect the my god you know i told you you. This doesn't work you know. This sentence does a meanest So a perfect. It's definitely part of the challenge for us guys right. I mean if i'm not mistaken you have to design and interface where the feeding of the data and then the structure of the output on some level is clean and that kind of people put things in in that cleaning consistent way to get that clean consists now quote. You've got to also make that user friendly and that's not an easy job. I think that's one of the reasons a lot of a. I have a hard time. Yeah always spend enough time to actually find you on that by getting a lot of feedback from you know from the customers. Some seven we got some rates. We ask the customer cain the deployment phase as the. You know the enemies to rate from one to ten. You know they happy again. Yeah that that feedback is essential as a central. We can turn of another feature introduced to put some mesh. Earning somebody's may you know may be very comfortable with their some tone. You know some would love to have some Some seem some you know sentence structure so basically we gave them a few options you know. Would you like to prefer to rights like this or like that and we time we learn the style of the nfl so that we can actually something that you know. Grieve rights like-minded likes dan. Like shenault and that helps a lot in the option. Yeah i can imagine what. Because if if i recall correctly i think we interviewed folks three or four years ago or something like that. There is a lot of not necessarily machine learning in getting some elements of energy right. Because you don't want statistically we think this will work reporting what you want. Is you want like a reliable output which is a little bit more of an if then if then and structuring those things properly so rule fronts engine but then style in the way you ride this is where it can use machine and mission our mission so g is really getting better in that respect right i mean just look at the last three years you know i mean think about opening i. It's like it's incredible the kinds of things that are now possible even open source versus just a few years ago. Okay so that's the. Nfl fits in in sort of the the style of the output in terms of how the user would like to have it design. Okay so this is one example. We're talking about riskier and this is where we can take. Risks reports about credit risk and are these often very bespoke manual in other words. I can imagine there's almost an unlimited number of things to do reporting on do you have are there. Are there almost out of the box. Everybody has to do this kind of analysis in probably this kind of way boom baked into your product or is it more. Hey they're going to do an infinite number of different bespoke things based on what their internal and external stakeholders are needing and we just need a platform. That's more general if there are some regular ones i'd love to know what they are but if it's all the spoke that'd be interesting to us. This twist report is clearly an out of the box in a matter of days now every banks have their own rules of their own risk rules. You know rivers will allow no coach to joe so that none technician at the bank level can actually do the musician station themselves. They can actually fine-tuned the rules. They can actually build the reports. They want an every banks you know. Even though the same report you know every single banks of their own way for sure that we don't have to have a incredibly clever engineer to all the the whole the the business guy. Say okay guys. This is a standard way to do not based on our experience and were built and song now we completely understand you may do things slightly differently and that the risk level you want to you know whatever you want so all the parents. It's a three hour training and somebody can actually built his own report based on the mayor. We have prepared in sweet hours. They can actually choon these modern and have something that can really work. Yeah that's that's one of the things you only get from experience working with customers writing mean there's so many assumptions you'd have about what kind of reports would be needed a bank and how could a nontechnical person build it. It's like i imagine you learn so many things first couple of years in business learning about how to make this non tactically accessible right. That's that's a real a really important element of building products now every single of our customers once we build the first report hundred percent of them. Say okay guys. This is cool. This is working great. You know we have other ideas. Yeah yeah yeah well. And that's that's where you wanna get one. This ngo have other use cases in mind on progressive. You are building libraries of different reports that we can actually automate. Let's get one hundred percent. That's that's a tough number to hold onto. I feel the closer you can get better and certainly being able to have nontechnical people using an ai oriented tool at all is not an achievement. Anybody should should under. So that's risk. Now there's other spaces where you folks play as well. And i know that one of these is around improving quality of according for financial analysts in controllers. I think again the workflow for some of our listeners might be familiar if they work inside of that area of kind of the the controllers business unit. If you will but for some folks might not so. Maybe you could walk us through. What are these folks trying to report on what insights they looking for and then what is the process. Look like for them where we bring. Ai into the picture be an interesting project on many other projects folk controller or dna organization many reports echo than performance reports. They've got all at one. One project who have done for large Very large retail network in france. That they guys we would love to understand the performance or branches and they came up with twelve different. Kpi's you know what type of product y type of channel using which region and which branch which specific retail network and so on. So long. i. It's it's so got all these data what does does provide them a very. Are resisting tactic. Analyses of okay. What is the best branch where the customer customer satisfaction being battled out where the customer satisfied and they put some contest. You know the stat on the weekly and monthly basis. Though that every branch the branches are were actually waiting for that. He's not just competition for the sake of competition. The explaining guys you are now in the ba- five in your region or a national vesey's this she's trying. This is your key area of improvement This is the highlights of the mounts. You you know it's automated. It's absolutely you know we were talking about quality the information. Yeah the same rules for everybody not depend so fanatics restarted. Yeah this person uses. Microsoft salvage This other thing it's very well say explains why you know and that's as well very potter to explain okay. This is a big why you are unknown number one in customer satisfaction and this is what made the changes this you can count this product and saw happen to interview some of the brench measure. They saw that as a very powerful team even to motivate their teams guys. You these what we are you know and so it's pretty. You know very interesting. Quantity of the information timeline formation. Other banks exactly the same executives abdication. The made them to analyze their beck society. Spendings to be very specific. And so. it's you know. Less of a competition here is more to understand. Okay why did spendings. How does that compare to the last year or how compared to at what is forecast. You you can do that on. Sales know bbc on that you'd be was while the dodgers french banks are their production. Your mortgage production sale gate. Let's do this. That had a team that you know and they did. One is for the whole group. This was before zero and these guys super-busy they say okay we can just do. One is for the whole for the whole group. They've got tool major networks. They have got to. You know a lot of different departments and so on. They may not remorse for the branches for the retail thirteen different reports. Same team and they plan to do hundreds of different reports and the head of the You know the team told me. I want to be a business partner. I want to really bring to my internal customers detailed information how their business work just to colorado of this stuff. When i think controller i almost immediately and this is probably because i'm not a controller. I immediately think finance you mentioned customer service. I'm interested you know we talk about looking at the performance of a whole bunch of different branches and being able to compare them have healthy competition so that we can really drive business outcomes. I think anybody listening can understand. You can only manage what you can measure if you just have a bunch of spreadsheets that might be measurement. But it's not making it clear if you have really simple reporting that people can understand it compares things properly. You can now really everybody on the same page that makes sense for what a controller cares about. You mentioned mortgage production. I would imagine you know the. Pnl on the balance sheet would be where they're focusing. But it sounds like there's a lot more elements in there. What are some of the things that controllers need to pulse jerk raleigh. Nl and balance. She is as well. You know a very hot tub. 'age here that's time critical at real. The summit that any control is spending fifty percent obese time. Fifty one works on report. And this'll be porty pence fifty percent of his time building the report and fifty percent of. It's done understanding. And billy action plan discussing with his until stakeholders their sense on. So what he does is to just suppress fifty percent of i fifty percent who this but domestically don few minutes a can just focus on the more quantity work. I would essentially bet. Let's say my mother's life on the fact that it's not a few minutes for everybody. In all cases like this probably has got banks that ripping in data source xyz q r s. But you wanna make simple as you possibly can. Is this mostly from. Api's piping this stuff in is it from having the right places for humans to just manually input things what helps bring that time down. Ninety five percent of the time it is okay. Yeah you're you're fine the way to pipe on matica of information it can be from any different sources who have that can come from a c. p. can from your erp can comes from excel spreadsheet. Don't really care and fifteen different sources. We information's store that you know what we call a cuban which is a multi dimensional way to present the data and then our sta. That's that's a real challenge. What you're doing there. Because i think that i suppose we call it. The data engineering part of this problem. Right where you folks are working in different languages. Obviously you know your franchir. He english whatever you have different formats you know whether it's excel or csv or getting pipe from this database that database and there's gonna be a way where we take those through the right combs right filters to align the columns pool in the right data to be able to streamline that is really a non trivial task especially because there's probably some variance from bank to bank or even department to department. What are some things. Those are the big challenge. One of our biggest feels really significant to walk a very impressive these targeted use case his. Now i've got the data we are. We know what we're looking for if you want. So we got a targeted data structure. So we know we. Won't you know these. You know what we call an anthology prebuilt until g where we know initiative of all these different. Kpi's unsawn okay. Then what is left. I was says to do this quote. Unquote that mapping and discussing with a customer. Okay where's your arm. And then having data mapping so that we can match dinner specific environments with the targeted. You know easy ab- way of looking at things as men. There's a ton of challenge there. But i imagine you work with enough customers. You build on policy. That's pretty complete. And then you know customer number twenty a lot of the things you can just find the match right you can say oh well this lines up here and okay. We've used accel before. And he's kinda got something built where it can pipe band. But i imagine building. That system has been credit-risk. Is something like five hundred. Kpi's now to experience a lot. Yeah i would say well. So the customer say yo. You don't have that they are. I just do an upgrade and you know but most of the time we are. Yeah i would say hi. Runners want truck. I think one of the things that's important for people to understand listening in is that you know getting that data to make sense in a coherent way where we can start reporting on top of. It is absolutely not an easy in easy part of the job. Right for you to be able to put together paragraphs of what's happening in all these different branches that ontological that structure in the back end is really what makes what makes the magic happen and it sounds like for you to be able to pull in customer service or mortgage production. It's because there's an oncology that can support those differences again the customers can do his own. You know things in and then change the. And that's the magic. And we produce a texted medically now. We can say okay. I want very verbose. A you know this in french and english show on his paragraphs to be away lighter. I mean that's you know. Snow could studio super helpful so that they can. You know it will become their report. Yep we just provided a tool. Yeah like you said cutting down on that. I fifty percent right bringing that stuff down so they can focus on understanding and conveying those insights so final question will wrap up here but i think got to the media things. I think we really started to explain. The some of the the under girding technologies that are allowing to make this happen which is really instructor audience. The use case here quality improvement is the overall surname use case. It seems like the way quality improvement. The way that you folks facilitate that is basically by having a unified way to look at the data. It feels like in the data matching what makes quality happened because otherwise this person's got a pie chart they built off of google sheets. This person's got a pie chart from twenty day old data that was on their desktop. And it feels like that back end is what makes quality happened is that is that right or there other elements to thinking about quality improvement that we don't want to leave out here. Could he comes from quantity of the as you say don zone until the algorithm who plays quality comes from consistency so even language may be different. You know that the fundamentally were very consistent as we can go down to fifteen different levels you know doing some scroll down impossible for human to fifteen dollars. You know a fifteen levels downs. We can actually detect that. So it's a very fine and that is even you know. The narrative may look super clean. Always the you know different and edges and so on so critical from realtime disease very system or deductible so you know it's a lot of elements that come together here. But i like this general idea of cutting down on that i fifty percent and the importance of data engineering and even the place of machine. Learning in terms of the style. I think is all interesting for opening the eyes of the listeners. For this this use case so manual. I know that that's all we had for time. But i'm glad you could join us all the way from paris for this episode. Thanks so much for being able to be here on the show thank you. That's all for this episode of a in business podcast. I hope you got some good ideas from this episode. I appreciate emmanuel being able to join us and put some detail on where. Ai fits into reporting and also what makes it happen. Certainly a lot goes on in the background to make these reports. Come to life. And i want to appreciate you for listening all the way through to the end of this episode. If you're not already following on social please do been a real pleasure to see more folks engaging with us on lincoln and twitter and as our audience for the podcast has grown our social audiences grown in kind or creating a lot more content this year including some video kicking off in the coming months and i look forward to making sure that you see all of that. So stay abreast of all the latest trends in business. Our new best practices for our ally in ai. Adoption and other great use case episodes like this one by following us on social is just the m. e. r. j. on twitter very easy to find us or more at emerge artificial intelligence research on linked in or on facebook. If you wanna get all of our latest material whether it's infographics articles interviews and more stay plugged in on social. We appreciate you being here and if you want to show following us on social certainly helps so that's all for this episode. Look forward to catching the next one. You're on the a and business podcast.
Challenges and Opportunities of Deploying Enterprise AI - with Derek Choy of Aktana
"This is daniel fidel. And you're listening to a and business podcast into thursday. So this is our making the business case episode. Every tuesday we cover use cases every thursday. We talk about the arwa. They i and the realistic considerations for deploying ai. In the enterprise. We speak this week with derek choy. Who's the co founder and ceo of a company called Mcdonagh offers Sales and marketing. Ai based solution into the pharmaceutical space this is a company that has developed a solution in a part of the farmer ecosystem. That frankly doesn't have that much attention rather complicated use case rather different than the way. The companies are running their marketing and sales operations. Now so how do you make an ai. Product work when you're not just one of many but you really are changing the game. What are some of those hurdles you have to overcome and are the some opportunities you can kind of grab hold of bringing a brand new type of product to market. If you're a vendor you're gonna wanna listen to this episode and if you're an enterprise leader that's may be considering spinning out in a product or even picking what vendor you might wanna work with. These realistic challenges are things. You're gonna want to bear in mind and making those decisions and making the hopefully with a greater degree of context and a greater likelihood of success. I really do appreciate derek. Sharing some of his journey some of his hard lessons learned in making ai. Work for novel new use cases. If you're getting started with the pulling in your business or in your enterprise you wanna help your clients. Deploy artificial intelligence. You can download our beginning with a. I guide it's three critical insights for non technical professionals. If you don't write code but you really want to understand core fundamentals about what it takes to apply in an enterprise context go to emc rj dot com slash biji. One and you can download our free. Pdf brief on exactly that topic. It's beginning with three key insights for non technical professionals and again that's e. m. e. r. j. dot com slash b e g and in the number. One that's being g like beginning and then the number one if you're looking for more context on exactly this topic and taking some of these insights the next level. Hopefully that resources will be helpful for you without further ado. We're gonna fly into this episode with garrick choi of tana. You're on a in business podcast. Derek we're gonna talk about making the business case and you guys are in the life sciences domain obviously a world that's that's certainly not not nude investing in l. Now when you kinda go about explaining to a customer client what's different about deploying in a system like the one that you folks offer versus of traditional it. What are those key distinctions. That are really important to be frank about an honest about that. Make the the adoption deployment considering a little bit different. Yeah that's a great question. I mean i guess the first thing. I think something everyone probably already knows the importance of data and what that means is the ability to. It's not just the importance of having access to data that's available one off but having ready for it to be used for scalable aiops locations in this means having it in a the rightful matt and kind of unified but also having Updates it's updated frequently enough to be able to be used to update models that you create so i think that's a really big thing and the readiness of accompanying. Does i need to understand that. That's a heavy list one of the things that you know. We've been focused on company. Help customers move. This is understanding that. There's a possibility that the more you try to have your data the ready to not only build machine learning on but actually have that same data set be the same data and structure. You're going to test deploy this model in production on the easier it gets. I think something that we really help customers with and we we try to push them towards which thinking about rather than taking dada that might lead you to Theoretically optimal betta prediction but is harder to integrate in full hodeida to a maintain trying to focus on getting the subset of data that you can actually pull into a data warehouse and be able to run your eye on top of and do that in production that is actually gonna be more valuable than focusing too much on the ideal lonzo just more spots and maybe not easy integrate. That's one aspect. Yeah yet in. That might even be a little bit different customer to customer but certainly as the data considerations are going to be a big part of the mix up. there's so much hands on guidance from vendor companies like yourselves to take that into account. What else is a little bit different. Yeah they gotta have their data house in order and obviously the usually needed out of guidance from you. What are some of these other ingredients. Many definitely need. I'm the right kind of focus on people had change management. This is something that i think we've learned over. The years is so critical that so much complexity and so much going here that you actually need to make sure that the organization is ready for the change. That's happened and so what this means. What we found is the companies on most successful are the ones where all the way from the right kind of from the executive level that dedicated to and they're they're willing to undergo the change that's required in the life. Science industry means a commercial organization that brand strategy to change the way they managed content is going to change the content needs to be more real time needs to be tagged in the right way Greater needs to be willing maybe to move in a different way so they done slowdown kind of what's happening rations in the chinese mocking operations training the sales representatives Trained on different things and the sales reps themselves as well as the management will need to adopt different tools. At if that's gonna happen. You need that biden all the way from the top but also need that buying from the lowest level as well so it at the top down support the local buy-in so that changes not just something that is talked about but it's actually something that people adopt and continue to engage with ties back to another aspect of why think is different with a solution set up once and then done in a what we've found what we do in terms of recommendations is you have the scattered e that you set up within your engine and that's setup phase but then over time you're learning all the time the strategy is changing. The market is changing. You're looking at things out working and you ought to get things done not working. You're adjusting kind of what the engine is doing. And that requires kind of an ongoing kind of Maintenance as well as support evolution which might not be as and you have to plan for that you have to get like the role of the right people involved at it might not be as prevalent. We looking at just standard. It solutions yeah mean. All three of those. I will say are relevant in every single industry. And i'm sure you can imagine that tackling oil and gas to financial services. The data considerations the change management kind of cultural willingness considerations in especially you mentioned regulatory obviously finances plenty of that and then the served the considerations around ongoing maintenance upkeep iteration you know prevention algorithm drift Adjusting to reality. These are new. These are different than i feel four executives who are not aware of this that that sort of presumed that. Ai is just it. We're just gonna plug it in just like we did with salesforce just like we did with hubs just like we did with whatever all of this comes as a surprise and a big unpleasant surprise as opposed to a natural part of kind of beginning to get to the next level when it comes to solutions for their company. Obviously i'm not saying is always the right move to adopt but i think with the right expectations. People would be in a better place to make a smart decision. When you think about how to frame these will we could. We could call new challenges new considerations how do we frame these new challenges and new considerations to make sure that you know we're we're sort of educating these buyers as we're moving in with them because they may not know that this is what i requires but we may need to say look. This is what it is. But here's why it's actually a good thing. We kind of make that argument for folks that don't quite get it yet. Yeah it's really important aspects of the obviously being able to convince all the stakeholders you need to have that buying as i mentioned that by critical away from the top and yet the focus on what they care about. I think what we've learned is we have to focus on what the business outcomes in the impact. You're hoping to have and baking sheet a defining that all the way from the start but then also making sure that you break that down to kind of some different indicators that you can use to help drive towards that and i think you know if in the will of commercial operations in kind of trying to optimize. That was a marketing. The ultimate outcome. You're trying to help his sales you would focus on. How can we show that. It's gonna be sales impact in the area but then also being very clear about why you're going to see it and being cleared some of it is gonna come directly from the use of you know the intelligence itself by providing better decisions that someone's going to act on some of us actually gonna come more indirectly because you're gonna be having decision makers think and act in a different way than they did the full they're gonna be more data driven even if the recommendation that you make someone is to do x but they look at it and they say you know what. I'm gonna do why. That thoughtful decision is actually different than what i might have done. That may lead to them. And by the way if you close the loop may also lead to your recommendations getting beat up because you see someone took a different action you recommend it and then you learn that actually leads a bit outcome because i think when you think about impact from hey i do have the direct impact which is when you have a data driven recommendations be adopted. Obviously that is one one place where you're going to be driving different behavioral type of indirect impact on the fact that people paying attention to the questioning things. They're kind of engaging with systems rather than just having them be something that they input tool off things to actually is indirect benefit that you qualify for someone as you're thinking about executives We think about that. Derek is will we have breakdown. Roi into three ways. There's measurable roi which is often tied to something financial or something that proxies to something maniacal strategic roi and then also capability roi so the new capabilities. We need to build i. E cross functional teams. I e better data infrastructure. Those things. can we want to be able to frame our deployment so that the necessary capability building has spin out benefits and prepares us better for the future in general and it sounds like you're trying to do that with a cross functional team thing and i think that that's it's very smart. I think that you know these are seen as hurdles by executives unless you can frame it in a better way. Are you also able to do that. With data those you folks have to get your hands on data and infrastructure. And make sure that's clean are you also finding. Is it possible in your case. Anyway to find a way to frame that overhauling rethinking of of how are data is sorted cleaned harmonized as having kind of spin out future positive enablement for the company beyond just the expense of getting your your system started absolutely and i think the way that we try to do this is number one. Recognize that when you are making the investment in the data to be able to pal like what we all this the benefit directly get when it comes to using it to drive s disfigures case but then there's also the benefit that you can start using that data. Full father use cases and really. It's recognizing that somebody think about this as a second order data. One of the things that happens is you leverage dada announced into recommendation engine. And then you ends up providing you with a new set of data which is how with data used drive a use case or particular type of a strategy and that strategy effective with that customer. In that second type. Dodd created in terms of the customer responsiveness to a particular tactic or particular hyper strategy that dada is potentially even more valuable than initial if customers not realizing companies not realizing that the initial data it builds on itself inland second order things from it and then they start capturing that any integrating back. There's a whole another set of values that can kind of cool. Yeah and again. We've got to take the things that are that are hurdles that are seen as challenges and frame how they're actually modernizing improving. Nah helping the company moved forward any last closing notes. Derek things for you when it comes to really making the business case you know getting to yes for lack of better terms when it comes to an ai solution. That really for you is important. You've talked about some great points. I love this capability focused. And i think it's awesome that you folks are are really conscious about making that part of your presentation. Because i think it has to be. I think very mature thing. Any other little notes about what's critical for for sort of framing the right way and getting you know getting to a pilot or deployment think setting the expectations. The right way away from the stock that when when you gonna a you're gonna wanna leverage the best when it comes to technology with the best when it comes to human intelligence and on there's always a human factor to it. I think that's been pretty critical for us as a business and how we position ourselves and have credibility without customers. That when you're gonna liberty is solution. It's it's great that you can up with the best algorithms and machine learning models and optimization but you gotta need to associate that as well with the best input for the rules and constraints that you gotta set from the experts that you have but also you're going to need the focus on that change management in adoption from the news and those people that will be managing the program. But also you're gonna need to focus on the element of what the or the play is. What the company needs to State cone is need to be able to do to get a managed to scale think. She's set those expectations upfront. In you also associate that with the impact they get in. You can see the very clear and transparent about the hour ride. Not just being about the cost of the the data kostov the technology but the awry in the cost element needs to include kind of the human aspects. Such managing that change and itching still demonstrate. That win those things. All add up. You still gotta get the impact that you're looking for and you still have a huge win that transparency really helps people have credibility that this can actually transform an organization. I like it. i think that's that's really the only way to to grow one of these. Very hands on survey is solution firms at at scale. I think that companies like yours. Derek whether it be the the bespoke ai consultants who know what they're doing or the vendors that have a lot of experience and learned a lot of very hard lessons as i'm sure you have on. It's it's really you folks in again. The people selling these services to some degree who are going to. I guess i could call it. Your pitch is actually part of what's necessary it's necessary education for the c. Suite two to think about and like you said set the right expectations about what is so. It's sort of part of leveling up their smarts at the same time and i. I think that's a pretty cool win win. I like the way you worded it derek. I know we're up on time for this one. But i sincerely appreciate you jumping on and getting into one of these making the business case episodes with us. It's been great to have you on absolutely thanks for having me. That's all for this episode of the ai. in business. podcast. I appreciate you joining us if you like what. You're hearing here. Be sure to stay tuned on social. You can find us at emc. R. j. on twitter or at emerge official intelligence research on linked in or on facebook. And you can get all of our latest infographics latest articles and our latest episodes of this podcast. The i'm business podcast. And our other show called the a in financial services podcast. If you're interested in banking insurance financial services. Broadly before to stay tuned for that podcast as well. If you're not a resubscribe you can search i in financial services on itunes or apple podcasts or on spotify or soundcloud and you can subscribe to us there as well and stay tuned for that show but otherwise do hope you stay engaged on social or twitter following has really grown since i started mentioning social at the end of the podcast. It's been great to hear some of the comments and see some of the engagement from some of our longtime listeners. On social as well. I really do appreciate you. Great part of the community. We'd love to put into the broader conversation on social to. I'm often creating lengthy threads on twitter and linked in where we have some of our listeners and subscribers chime in with some of their ideas and maybe you could be part of that conversation so follow us on lincoln. Twitter facebook otherwise. You'll catch us here. Next tuesday for our use case episode and the airline business gas and i look forward to catching them.
What AI Readiness Really Means - with Tim Estes of Digital Reasoning
"This is Daniel Fidel on, you're listening to a in business podcast. It's Thursday that means it's are making the business case episode making the business case we talk about Ai Deployment and the realistic challenges in the enterprise and opportunities and measuring Roi of Ai. This episode, we're going to talk about what a I. Actually means and we've got a guest with some excellent perspective on just that Tim Estes as the Executive Director and co CEO of digital reasoning. Digital reasoning has raised an awful lot of money to apply artificial intelligence to various industry sectors including financial services, federal and more tim speaks to us about what artificial intelligence readiness means from his vantage point. So digital reason applies mostly in the natural language processing. Space, but many of the will transfer no matter what you're aiming to do within your business. What is it that companies have to have kind of ground rules as sort of a baseline reality of their data of their enterprise expertise of their in house talent to truly be served ready to adopt in deploy artificial intelligence with nearly twenty years doing exactly Tim's got a perspective that I think is worth. Tuning into if you're just getting started with a in your own journey of a high readiness or you're helping your clients do death, then be sure to download our beginning with a free pdf guide you can find that at EMC RJ dot com slash be e. g. one that's bg like beginning the Edgy One. You can download the free pdf brief, which is going to be basically grounding concepts on. Early deployment that's GONNA help you get more out of your podcast listening and also help you help either your clients or yourself with the early phases of AI deployment the areas where critically a lot of companies get wrong. So hopefully, that resource, you'll find valuable and I'm certain you'll find Tim's interview today valuable without further ado. Let's hop into it says Tim Estes with digital reasoning on the in business podcast. So, Tim will kick things off and get your perspective on what Ai Reading this means. When an enterprise says walked, we want to become a I ready. We want to start using a I what kind of components have to go into that? Yeah, well, I think the first thing is you had to have infrastructure that sounds so basic especially with the cloud bud, the larger enterprises a requires a good functioning process to allocate infrastructure with their on premise or cloud. And then data governance of data can be used for training and validation around any process it's going to be tested. So it's all too often that you know one group in enterprise wants to try something. The aren't really the owners of the data that is required. To validate what they want to try. And they are not the suppliers of the infrastructure. So you might run into a substantial gap. The could take you know a sixty day or thirty day pilot. Or PSE and make it a nine month process because you're waiting on them to sort out data governance and infrastructure availability. So those are two pieces you know something about the education side of it. In terms of you know this this dictation you want to build and educate yourself to understand the difference between certain techniques, but it's always overalled because in the end of the day I I'm a little bit more pragmatic I think there's certain techniques which are better for. Some things and others, but obviously, the most sexy technique that talked about the time or different variations of deep learning. Yeah and we could go into braces but the phillies he'll is in most cases, the customer doesn't have the data sets available to train a really good deep learning class fire and so or an engine of some kind. So I I think that what you find actually is it's not just that they have data general had dated is prepared a certain way. Often to teach a machine that the machine can perform the task and that's really the that's the area. So these maybe the this question you elite in some other things but you know basics, infrastructure data governance I can pull they need to run the test fast and as a as a vendor or someone the outside I mean I would becoming in asking these questions now because I lived through being wishful thinking and. This is really exciting CTO and they want to do this and they have a business stakeholder that wants to do this kind of application. They think we're the answer I've been through that whole dance, and then you find out that of the whole dance that dance might take months four and then you wait nine months for data and infrastructure to be available in the large bank. Yeah. Well, not surprising at all within a large bank you're lucky it's not a eighteen months or something. So you bring up infrastructure you're bringing up data does this mean in the process of speaking to whoever your initial? Champion as your your initial kind of point of contact who you think is GonNa either signed the Checker help sign the check the you really have to be clear that sort of what infrastructure you need to access of what kinds of data you'll need to access of the state of that data with that person and or with whoever they need to rope in serve as part of the process of working to a pilot. So like doing that diagnostic I, guess as you go as you progress forward. Yeah that's right. So I mean I'm naturally gonNA give it more from the vendor sides, of course. But if I flipped the hats and I'm I'm actually in the buyers persona what I wouldn't WanNa do is the last thing I want to do is to put a lot of energy into something that could create real value get excited marketed internally, and then find out that getting infrastructure having data governance process in place where we can get the data necessary to test the system is not really well figured out or is figuring out but the restrictions that make this not work. So I, think that there's a good upfront investment in that but there's a difference between that and sort of what I might call the the Data Lake Panacea. We're everyone wants to have this. Highly Organized Library of data with the Dewey. Decimal system in their enterprise. And that's not gonNA. Prize is unfortunately function. So many it efforts in an enterprise are responsive the business as a higher priority to eating across business lines. That you'll almost never find as you will a pristine data infrastructure. So you really WanNa make sure the process to pull data, put it in the compute environment do that safely and would security sign all's that should be enough to get moving, and so I think if you try to go four steps beyond that. You have much bigger challenge and essentially trying to boil the ocean and I think a lot of people went down that road with all the WHO do vendors to be blood. You know the idea that just got to spend all this money on that and then from that. You end up having all this application. These applications become so easy and here we are five years later it must. We're seeing what applications besides restoring my you know might my loan scores or some other batch structure process? You could probably done some other way. Yeah well, yeah. People talk about the data swamp as opposed to the data lake that was sold or what have you cloudera still made a lot of money but but yeah, I think that's that's often the the gripe. So what you're saying is maybe be more modest with initial goals if we're a buyer for assessing our own readiness, do we at least? Have the stuff organized enough to harmonized organized whatever enough to train some kind of model on it, and can we get enough of a handful of even run a pilot and those are hard nose than maybe we're just not ready for this particular business function altogether, and we need to either focus somewhere else or or focus on getting our infrastructure up to speed. Exactly, and let me tell you the cautionary tale, which is there are people that don't do that work in enterprises. They don't do that assessment of front they get the vendor very far into the process. They may even think that it's going to be easy to get the they need and what happens is this they get toward the end they done all this work and the vendor says, well, okay let's do this. Here's the data I need and they said, well, we check on, we really get that data. We have this other toyed over here. Can you do this and the toy data in Hungary startup or small or oh yeah we'll do that. But then what you actually prove isn't what you meant to prove. and. You find out that they'll be objections and come back. Well, that's not really on our data. I don't know if it's really going to work, and so you're not really that much further along than when you started. That happens more than you would expect that people don't do that upfront work in it sets up essentially or worse economy works in that other data but data. So not exemplary of what the real business process data is that it goes to production. You almost are back to square one and you disappoint because it's like this is the kind of data that we saw will. Yeah we didn't have access to the data. So I'm just pointing US I I. Think this is a real threats much bigger issue than people think of through readiness even though in theory it sounds simple. It's not because the politics of who owns data. Either risk version of disclosing it to more people than have to or the you know the politics of this data has value and I wanNA control who gets value out of the data. Yeah that's unfortunate. But you know reality that in my opinion Tim, every vendor learns that with a bat to the face. I. Don't really I don't really know of any vendor that just sort of bypassed you know stepping on the rake there because it feels like if you spin out of Amazon, you just think data's accessible and easy. If you spin out of university, you just think that your science is good enough and you'll just make it happen but everybody that. Moves into a space gets hit in the face with that and then and then has to really dial back their sales process needs to be really white glove upfront handle all this kind of gunk. Do you think that in I don't know four years from now tim that we will have less of that upfront gunk to get a POC to work to get access to the data we need or is it a way longer ballgame than that? What are your thoughts? I mean it could be I tend to think it is a longer ballgame. I think he's got to use that I mean at the end of the day the I wanNA covers always gonNA invest in. You is the time and them investing time as long as think they're getting something out of it is not always a bad thing. Now it cost you money time just call them but generally the the buyers lot bigger than you are. So that's not a thing you can run indefinitely so insured, I, mean I. There is a long window here where now just thinking about a little bit back to the face I was wondering if my nose is actually genetic or not. No No? No represent reference no no no. I'm just having fun with him because I think that it's a really great lesson learned that it's almost like no matter how someone tells you during a learn because you just want to believe it's easier and it's not the area I think people are now coming around to the piff handed pivot a little bit in this it's Okay let's say the infrastructure let's say they can get the data staged and you can run whatever you're gonNA run on the process you want to validate. Most the time what you come in with has to be taught or adapted to the customer's data. I e the out of the box it just works Hell No. In a non-consumer area just in the enterprise space, it almost never happens. So the next barrier becomes, do they have the data organized in any way to teach your machine or not? That is the thing that has probably broken more a I projects in the bigger. Project the harder in the bigger that problem I just described is. So, if you go in on their very public projects, it's been tens of millions of dollars and the customer will say we get the outcome we fall and generalists because the expectation for set high and they were set high by radically underestimating the availability of the data to educate the machine. To solve that you have to really coach the customer through that. Oriented real answer. We actually don't think that's the image you coach the customer on it. We don't think the customers are going to move fast enough to organize your data. So we actually had admit technology around that. So we we found that the biggest bottleneck for us was, could we have our machine learning models be taught from customer data really fast by having the I find the data to teach the I. So, we actually went one level further, which was, could we actually use? To build a curriculum and teach the I a customer data customer data so. I I don't actually know a really good answer to that. But I am confident there's a good process answer I. think There's a technology answering that problem but I think that's what's going to bottle up a lot of these projects is this problem of not having the training set available to educate the machine and then having to educate the client on on how teach machine we had a product is that into literally something that was almost like a game for our area and I think probably that's going to be what happens I think just sulphur technology in the end actually so Yes if you If you don't WanNa, be a service business and ucla out of the enterprise There are some news flashes that you will learn sooner rather than later hopefully but maybe more bats the face for some of the folks tuned in regardless with that being said, we've talked about infrastructure and this really pivotal concern. The you've talked about around wise initial projects fail and how the bigger and more ambitious oftentimes the more. Tragically they flop couple of things that we haven't touched on yet I'd like to get your thoughts on is around sort of talent and culture. So you talk about like well Geez, we couldn't. Get our hands on the data. We hear of a lot of pilots failing in just doing having so many conversations because enterprise leaders expect that it's plugging play or because enterprise leaders don't realize what kinds of use cases are realistic or not realistic. Let's call that like executive understanding. There's also people don't have sufficient house talent I've talked to the enders who actually help their customers hire in house data scientists to help them work with the vendor I mean like in that didn't even really surprised me that much because of how intense this stuff is. So there's internal talent as well. What are those other components? Infra? Okay. What are the other things that make it work what what else goes into readiness? Well. I think that the talent question really comes down to what kind of buyer a re talking about I believe there are either. Or aspirational enterprises that one have deep competency in ai to the level that they could build their own applications and and make it work. That doesn't mean they're going to build other applications by any means. But it means that you're actually overcoming almost an internal competitor to most of your value propositions. And whether competitor is real or not meaning, do they ever launch project and fund ten to one hundred developers to replicate what you build? It. May never happen. But what happens is you're fighting business case inside. which is a week to do this, and it would be one half the cost. So I think that what I see more often is you have that extreme, which is the you know we can build versus buy it. And Open source in the large tech companies open sourcing such substantial advancements in technology. Have made that less challenging than it used to be meaning. It's it's actually far more tree that you could build a lot but having said that it's almost like, why would you wanna Fab your own chips and building motherboards? It's more hobbyist thing to do I think we still have large institutions that are doing it and so you really want is. If, you have a lane where you can make a very distinctive value proposition. What you end up doing is you end up bringing on those people as champions because you're enabling them to show value. You're saying why spin the energy in this area? Do you really want to build a system like this and based? We have your internal products. Or instead you know, do you want to focus on what comes next? And by focusing what comes next year allowing them to get ahead of their peers or match their peers so far left if you will spectrum is internally I shop that's their next spectrum is the ones that think it's there but it's not and. That's really challenging because. Bad News to him. Yeah that that's that's where you have a handful of experts but not really believed implementer deliver, and so the expectations are said Hi and then a lot of the load comes on to you for delivery because the talent isn't deep enough or broad, enough enough volume to actually make themselves sufficient. And then you go to the next level over, which is you know they they have curiosity that may have one or two experts that are there for vetting only but they know they gotta buy they're looking for a solution and they recognize though that it's not fully turnkey and those tend to be some of your other customers to because I think the that they respect for this area had to go through and cost out what it would do to actually build a team to do this and make it function, deliver multiple products, and finally heal the far. Right. You have the you know what? I really want to check a box but I liked God this word somewhere like. Man Happens all the time and so so I think that you have like those extremes you've got kind of the almost like Saudi matter experts that are I'd Hansen the middle and then you've got check box checkers on the right that essentially one a little bit of of Ai Decoration because the idea they didn't take something when I was available makes them look like behind and on the far left you have the weekend bill all of it probably better than you and we pay our people more. Than you do and so you have that full spectrum I'm not trying to be negative. Need us now the talent the talent basically is my point is the talent issue is not one of having it or not he it's if you have the talent and you can have other false if you don't have the talent, you can have a of pitfalls and so that the truth is it's like it's all about humility of the enterprise and humility of the vendor to get to the truth, and if you don't have that then I think you end up with misalignment and it creates tension until you get to an answer either you have large enterprises try to. Build it and then find they're holding onto a masterpiece of technology and have to keep all this talent. Happy because when that talent NCFEI's it win, that talent leaves to go start a company. You can't make the system he built right this happens large solutions additions all the time. It's why you know generally you have commercial off the shelf software because they know that risk is catastrophic and then the next over is we're GONNA go pick something that's kind of demos well, and has ai, and we're going to elevate that even though we don't have any capacity to assess internally because what this stuff is not even well enterprise hard. What if it is doesn't More basic requirements, but it's just sexy new. So I think in the talent has to align to the mission of that you set the team won. But more importantly, it's the humility. Of whatever you really have. You could be you know J. P. Morgan have amazing people in talent or you can have nowhere near that budget in another industry that has like the one person that's got the to masters level in Ai that have been on teams it done in. L. P. twice. Like. You have that spread in the enterprise space, right? It would that spread just know where you're at. And then optimize appropriately and basically ask for humility, and then expected of the vendor to vendor over sells you or doesn't show the same then maybe not good partner. Got It. So little takeaway Tim, we'll nutshell this as we wrap up here it sounds like if I'm a leader in, we've got plenty of sea level folks tune into the show directors, etc who were looking at their own organization. There are asking the question are we ready? Where do we stand? You know we talked about data you talked about infrastructure you talk about what kind of projects to pick it sounds like another big maybe take home. Point in maybe reword this but another take on point is understand really frankly I guess number one where you are on the date in the infra. But also where are you with your talent and expertise as well to not sort of have to maybe feel bigger and stronger about what you have in house in you actually do but be able to say, Hey, look, this is what I think. We're good at what we're frankly not good at we have experienced where I think we're not and be able to guess readiness. It sounds like really involves a readiness to assess that very objectively if you are a leader. Itself Assessment I. Use a it's a cultural thing. Right? I guess I'm saying that like in many things if we were talking about in a different area than a even the still is true. A incorrect self-assessment leading to a misalignment is going to create issues in a I. It can create really substantial issues because. It's not very well understood yet even by the people that are the experts. In fact, what you find with some of the deepest experts the ones that have you cadillac some of the deep learning revolution. For instance, you'll find humility you'll find them talking about, yeah. We're about to hit a wall here really gonNA pattern recognition. We can do stuff and getting dictating years ago but you know when it comes to generalization and different kinds of signals choosing together and unsupervised learning we're just really the basics they're like that's what you'll hear like he actually been. been on the show five years ago. When you have those caliber or people who deserve the credit or You. Come at it from the perspective like that's all I need to know right that tells me that if they don't. Have a lot of swagger about it. Then who has right to? Yeah Nice I think that's a good point to end on. If if if COON INVENT GIO are swag about this, being a done deal in terms of making this stuff work in the lab, what makes anybody think that they should carry that into a business conversation where money's on the line? It's not. It's not like something where you turn into a skeptic money means it's just it takes humility and once again focus on the things like there's all these tasks that humans spend time. All right now that shouldn't have to because we're a today. You can make effective classification of a lot of things and you can triage and you can get a lot of multiple human time like that's that's an area that's highly under-exploited. The ARP guys are doing a tiny fraction of that and I mean there's a as a credit. They tapped into it at a lot of success. All these other kinds of signals language vision like we're just the beginning of that automation, and in the fusing of those different triage is into more complex ranking of what's important with a risk score that's areas which just need to be baked into these enterprises. Right and harnessing the feedback from people working on those tasks like always learning having a learning loop. Those are things that are right in the thick of the most competitive enterprises right now. So I don't WanNa sound like a ski at all. I'm just saying that I think people you know probably set. An expectation because they're trying to market themselves, which is human thing to do, and I think in the end people better served by you know humility I actually I really when I was younger in this I've been almost twenty years as a twenty years old boy al I've made all these mistakes. So I say this not from any variance like I've already done these mistakes and so I come at this and say, yeah, I remember when I thought we would get from here to here and it would take you know two years. Now I'm your tandem got. Not all of it yes. Problem big time in like you said, it doesn't mean that we can't solve meaningful promises with Ai. It just means we shouldn't take for granted where we stand or maybe over underestimate things but we need to look things frankly worthwhile lesson for literally anybody tune Tim I. Know That's all we have for time in the readiness episode year. Thank you very much for joining us on the PODCAST. Christian. So that's all for this episode of the AI, in business podcast if you like what you're hearing, if you did what we bring to the table ear in our use case episodes are making the business case episodes. You should know that those two different formats. PODCAST subscribers like you folks who've message me on Lincoln or who've left us nice reviews on apple podcasts and told us what they like about the program what they really want to see more of I'd love to hear your thoughts. If you've enjoyed the program, drop us a five star review on Apple podcasts and let us know what you want to hear more of what you liked most and what kind of value you particularly get out of the show and that's GonNa help me and my team continued to improve things for you. To make this program better than ever in the years ahead. So breezy define us on apple podcasts every now and again also I'll share those views in our email newsletter or on my social feed and give a nice big shout out to the folks that kind enough to share their thoughts on apple podcast. So again, a in business on Apple podcasts also sure to check out the I and financial services podcast. If you're not already a subscriber to that one as well if you love use cases on in, you're interested in how those. is making its way into banking insurance and wealth management than subscribe to that program as well on Apple podcasts from check us out there too. That's all for this episode will be catching next Tuesday for our. Use case episode again will be getting back into logistics and supply chain of don't miss it I'll see you soon.
Conversational Interfaces for Busy Salespeople - with Gilad Turbahn of Conga
"This is daniel fidel in. It's tuesday so you know what that means needs. We're covering use cases again on the eye and business podcast every single tuesday. That's what we do one of the most fun parts of my job and we cover this week. Something we haven't covered before conversational interfaces by that. Of course i'm joking we've covered conversational interfaces before but we haven't covered is conversational interfaces for sales people those of you who've ever managed salespeople or maybe noah salesperson in your family or friends circle know that they really like to be bogged down inside software systems. It's notoriously tough to get sales people to enter the right information. The right places and file away things right. They want to do what they're good at. Which is selling they want to be face to face with customers. Not in their techsystems. So how do we streamline what they're doing with jackpots. Virtual agents with conversational interfaces. Well that's a focus of this episode and we interview this week to bonn of this who's recently merged with conga after a leading quote to cash accompanied big enterprises on counts receivable and they've recently merged with a firm called conga heads. Up number of their ai initiatives and he explains this week. What they're working on for conversational interfaces with sales folks. The idea of better sales software has been around for a couple decades now but the idea of layering on. Ai is still the wild west to some degree. there's a lot of interesting potential use cases. It'll be neat to see which ones evolve and move forward but this is a use case. I thought you all should hear about so. We're gonna dive into this episode. If you are interested in our full library of use cases. I know many of our listeners. Are here because you really want to be able to have a library of use cases to plug into whatever your business situation in some of you are innovation leaders and you need to be able to have the stuff on deck to build a using leverage to potentially add business value and we have a lot of consultants and service providers who tune it really want a breadth of cases. They can use with their customers. They can find an instance in the workflow and say hey. I know what to fit in there. If a use case library is of value to you then be sure to check out emerge plus the podcast is really just the tip of the iceberg. All the use cases that we have in emerged dot com slash plus. You can also go to just e. r. j. dot com slash p. One that's plus and then the number one and you can learn more about emerged. Plus this is all of our use cases. This is our ai. Whitepaper library as well as our full list of ai. Best practices which infographics and frameworks for a wide adoption. And anything else. You'd need to make the business case for i if you want more on use cases and also more tools to put this stuff in action checkout yemi. Rj dot com slash p. One that's where you'll learn more about emerge plus you'll be able to get a sense of whether that might be fit. We've got hundreds of folks now joined the program. I know many of you are listening in right now. If you're not already a member at least learn more. J. dot com slash p. One without further ado this is key lod with which is recently merged with conga. You're on the ai. In business podcast. So we're gonna talk about a use case for kind of conversational interfaces but within a very A very interesting and bounded sort of domain of people on the go. Were looking for information aiming to move forward. Can you describe the current process as it is. Sales people sending quotes looking for info. Kind of what that general business problem in processes today at to start off. Yeah for sure so think of a sales person that goes into the first meeting with a customer and the meeting went great and the customer says all right in order for precede. We need to sign an nda time. This takes today right because usually you can't really go into the system get one yourself. You'd usually reach out your sales officer legal off in you'll do some email back and forth or call them and then sometimes a weekly usually get that documents in your inbox or let's assume that the customer says. Hey you know what we said. Fifteen licenses quote with twenty. Yeah think about the time that it takes to go. Open your laptop. Make the change go through the approvals all of that type of stuff and then regenerate and send it out. That's where the pain is at. especially since. Most salespeople are on the go now somebody would say covid right. Yeah it turns out that people still use mobile even when they're at home. Think about you waking up in the morning the first thing that most people do guilty as charged is the wake up and they look at their phone to see what happened overnight and was my contract finally finally approved so i can send it out. Yeah yeah in doing that through. Your phone is still extremely relevant. Even during covert and we won't be here forever anyway and probably. Some sales people are so used to being on their phone. They're like pacing back and forth in their yard just to get the old the old buyback feel like they're out there back at the office translators. So okay. So people who sales folks on the go they need to change contracts the to find information about a certain client that they might be are acting with what is this kind of bounded set of twos. Because i know. We're going to talk about where i fit into. Yeah updated contract. Find information about a kleiner. A contract are there are a couple other quick ones. That are really important here that often to do it turns out there actually so people refer to those as the crud operations. Lemme not use the technical term because it just sounds weird the ability to quickly create or at least get started with specific types of records so new agreements or requesting them or new quotes or at least getting started the ability to find information. That's the ours. That's the read so hey was contract ex approved. What's the status. Or what the date on quote y when does it expire. So that's the read the updates making those quick tweaks to the system so the number licenses from x. to y. Or you know what. Push the expiration date on this quotes to the end of the quarter. All all of that type of stuff. That's the update. And then deleting we haven't seen as many of those. But i can say just crust. So that's the full acronym. Yeah okay well the this. I've never heard this acronym before. Is this internal for you guys or is this just a term. It's actually a now at developer originally. It's a it's a software development term. How you work with databases and other systems. Yeah okay so this kind of task. These tend to be the tedious stuff that has happened on the fly that eats up. Time might slow our communication with the client. And we don't want that to happen exactly okay. So currently the way it happens is. They've got a flip up their laptop to export dock. Or they've got a call somebody at the office to get that information that maybe they don't have these sorts of things. Yeah exactly it's you know. A lot of systems that exists today revolve around getting so even if you look at salesforce which by the way amazing company in amazing products right but if you open their mobile app you get a wealth of information. But it's all geared towards showing you the world join you all the information that exists as opposed to you as the salesperson especially on the gold are interested in getting just the right amount of information it just right time in not before not after in being able to quickly act on it and that's the paradigm that we actually went with finding. What's the right of information to show. When's the right time to show it in being able to take just the relevant next action. That's how we're thinking about it. That's how we can find that world in specifically made it relevant and doable using a and other pieces on top of it especially around user experience. Okay so we can talk now about where. Ai fits into this. Because there's so much here right in some in some magical future world we have some amazing siri on ridiculous steroids who i just talked something to pump sat a new contract. I wanna book new meeting. I wanna check my flight. My you know whatever the case may be some some jeannie right and at some point that genie actually makes the damn sales But we're luckily you. And i probably have forty years before that happens. I think if i'm estimating correctly so in terms of what's realistic today for conversational interface to help with those crud tasks leave articulated. Where does it come in where cannot use case actually deliver value for these folks on the go. It's all about speed. it's all about making it possible for you. As the sales rep move the ball one step forward or two steps forward the way we keep on thinking about it is first of all. We need to address the user where they're at on mobile in whichever system. They're used so if you think about. If you look at sales people today you see a split. The one thing that we're doing is meeting these irs where they're at and if you think about it there two options. Today one is use whichever tools your company tells you in terms of actual system tools. So if the company tells you use salesforce one mobile app great. You'll just use that. Other option is use whichever communication. Channels your company. Has you have slack. You're gonna use that because that's where a lot of sales people live. Same applies for microsoft teams. That's low. We saw what happened in the last couple of months around and explosion of teams so first thing as for us to meet the users where they're at when they're on the go teams slacks force one. We're there that's where people interact. The second piece is okay. Let's give you just the right amount of information. Notify you when something happened in the system. That requires your attention so okay. It's the end of the quarter. It's let's say it's june thirtieth and you're waiting to close this really really big deal. But it's still pending on the vp for approvals. Second the epa approved that quote that. Okay finally i can send it out so we will notify you done. Hey personnel approved quote. Would you like to generate that quote. Send it out to the customer. So it's always about notifying and then letting you take the logical next step right here from the system without needing to open your laptop and again since we know the user we know their preferences and we know the system we already know all the relevant setup so no learning curve. You don't need to say use this template that template and set this setting. You don't need to know all that yellow. let's let's through. Maybe two instances of what this could do again and because before we started recording. I let you know my overt scepticism of all like sort of like asking a and open language and it'll take an action for you. It's like i done seen a lot of companies bigger than you. Not do so good with that so i always i always look for bounded realities realistic use cases that i actually could could believe are working in the world. You guys are working on this for quite some time. So what which one example of something where this sort of flowing conversational question within a slacker. Microsoft teams could lead to the kind of action. Where we're looking for. what's what's maybe one. Good one absolutely. So let's take that quote related one that we talked about her perfect okay so i don't even need to look at the whole flow. Hey i've been working on this quote for a while. It's now in approvals waiting to see what's going on with with the approvals in the system. So i think i the system will notify me saying. Hey it's been three days in person x. Has not yet done anything about this. Do you wanna poke them. You want to email them reach out. Do whatever you're on your phone so he can do whatever but the fact that i am reminding you that that's what then finally the processes finished then. The system will again notify you and tell you. Hey now the process finished. I know what the next logical step is. You can always ask for more information or see what they said so we will always give you both that natural language peace and also the quick action buttons the ones that will just quickly you can go through a navigate and will enable you to generate the relevant quote and then send it out for signatures for the customer using whatever. It is nobis dockside congress. Sign all those is a power the real power behind this is natural language understanding which is part of what we do is very broad and as you said. You can't really rely on you boiling the ocean and being able to do it well but when you constrain it to our dooming to contract life cycle management when you can stream it to quote management and within that domain you build additional layers on top of it that are all about understanding business speak understanding how sales people speak in specifically how they talk about contract. Lifecycle management in about quotes. That's where things become powerful in death. Words feasible yes. You know when i think about alexa for example seems really open ended a Honestly it's not conversing with me. I'm not going to be able to have a conversation about a wreck. That joker movie. I never watched still haven't seen it. But i am not gonna be only go into something like that with with alexa but but you know it will order me batteries play. It'll play music. Whatever but that's because it just detecting intent but they're bounded reality is obviously very very broad for a b. two c. famed breadth nauta depth. Yeah not a lot of death. The teamer exactly the opposite. Yeah and even depth though is really hard. I mean you'd no matter. How are you go man. I mean it's it's death is really really tough but regardless you're able to constrain. How many initial messages could come in and then interpret intent. The the something something quote for the something something account and it'll sort of assume okay who sending this message. What accounts they managing. Which one are they talking about. And then the old kind of know what that is so intense interpretation on on some level the taking of actions though the like replace this with that or whatever that feels really really tough for a machine to know where in this word doc to remove accent This stuff feels kind of space man level far out compared to where we're where we are today. How does that stuff get done. The taking action feels harder than interpreting and then suggesting options cool. Here's what i heard. Here's what you can do. That's an lp possible. You're talking about going a little bit farther so here a couple of things we learned. I felt that i layer. That you were talking about actually was originally harder than the taking action in the recent. Was that the natural language. Engine steady exist out there and by the way most of them are awesome were using several different ones. Were good at identifying intent entities but being able to take to the next level and understand the business speak. That's where we came in. It's only when we introduced our own proprietary layers. Where if a business person says. Hey i need to see all my quotes that are expiring before e o q okay. I did not name okay which actual field i'm talking about. I just said expiring trust me. There's no field in salesforce called expiring and did not mention the date. I just said e o q but since our own layer knows. Okay what is today. We know the business beacon. We know the configuration for that specific company that they're fiscal quarter in. Let's say it ends. July thirty first ours does as opposed to the typical june thirtieth. So we know all that and we understand it. We know how to translate your query into a national thing that you can run against the system and get the results back. So that's how did that. Firstly the actions that you were talking about is all about containing the type of actions. That are doable in telling the user. Hey i understand what you're asking for. That one i don't support but here's the reference to the manual on how to do this system. This is a game about trust. If people can trust that. I can do one thing. Well they'll keep on coming back as long as if i try something else system won't crash. It won't give me a bad error or something. It'll tell me i understood. I just don't support it yet. And i'll add that feature in the in the future but that's how we update so we started by basic dates. Then we focused on making quick changes in the cart specifically around quote cart that the term. We're using so specific number of licenses for specific product in a deal or a specific discount that you're applying for one of the line items so it's very specific. It's the typical use cases. That are the ones that people do quickly. And on the go. Think of the customer they would tell you ashley. We don't need twenty licenses. We need twenty five. That's the type of thing you can do quickly in. That's the ones we've invested in the rest. We explicitly said we cannot support these right now but maybe in the future when we need for even to do that step. I think it's important to for me to validate this. You can tell me if. I'm wrong because i i'd love to know but i would presume you fell heading up. The product understands as well as other data scientists. A lot of the subject matter experts meaning sales people in the field had to come together and build out those trees. Okay under these circumstances. What realistically are they going to ask. Are they gonna need. And then. what's the best resource under this. What's the realistically are they. Gonna machines not gonna comb your faq and know all that stuff right correct. You're going to have to have some kind of structuring for this these prompts and that was a lot of strategic thinking up up front i would assume absolutely and also a lot of Partnering with actual customers who try this out and help us train the system. We needed to understand how people speak. We needed to understand what type of things people want to do with this system. And what would they trust in. What would they not trust to the end of the day. The the real problem is you're asking something that you don't actually. You're not seeing it in front of your eyes but you're asking it to make changes to a deal you're working on and if numbers are off my god that's going to be bad so in order for you to trust it. A talked about this concept before is never good enough as a stand alone. It always has to be married to a wonderful user experience. That gives you the ability to trust at whatever's happening is assisting you it's not gonna fully take the action until you safe so but it's assisting you in just removing the tediousness of you having to know the system in allowing you to focus on what you do best which is sell if i know that the system's going to choose the right template for me to use when i'm generating the new quote in all i need to do is say your looks good. Let's go that saves me time in. That's you x married with. That's where we're successful. got it okay. Yup and again. Yeah you need to. You need The i you also need. I think the strategic forethought around. What are the trees this thing will handle. And that's not even necessarily ui because it's an ugly old spreadsheet or the line diagram somewhere but it's but it's so important right. I mean absolutely in that journey that we went through so i. I was lucky enough to be there at the beginning of working on this product. There aren't a lot of best practices around those out there especially not in the b. two b. space but it was an amazing journey with amazing team where we got to think about all the different types of problems. You're facing what happens when somebody asks a yes. No question as opposed to ask for information. How do you handle it differently. In stunningly we spun up a feature. That knows how to handle question answering today. People expect that as you go to google and you type in. What's the weather in. San francisco and google is going to come back with an actual answer. It's not gonna show you the search results so we built a layer for it in inside our tool. We built layers to understand that parsing dates in how different business people talk about it. We built layers to understand. What's called implicit references. Show me my quotes from last year. I didn't specify which field i'm talking about. But in the context of my company when we say quotes from last year we mean quotes that they're valid until date was last year or their creation. Data's last year we've built those layers because we understand in congo. We understand how people talk about their quotes. How people talk about their contract. We just married it where they i in. Great user experience to surface that to users. Hence you're not boiling deal. Yeah context. Specificity is it's gonna help you not boil the ocean make things possible. It's also gonna be your differentiator and enjoy the hard work that goes into these projects that you've done a great job of highlighting that i know that's all we have time but thank you again so much for being able to be here with us for sure. Thank you very much. That's all for this episode. A big thanks to god for being able to join us for this show and it big thanks to you for listening all the way through. We certainly appreciate it. If you want to be able to support the show then be sure to follow us on social. It means a lot. We've seen our twitter handle grow a lot in the last six months since we started mentioning social at the end of the podcast. We've seen are linked in following an engagement. Start to grow since we've mentioned it. I don't know why i didn't do it earlier. But it's a lot of fun to your ideas from you all and to keep the conversation. Rolling also definitely helps us to support show and to learn more about what you're interested in so finds just e. m. e. r. j. on twitter or emerge artificial intelligence research on linked in or on facebook. You can follow us there and get all of our latest episodes. Also engage with us there. It's a lot of fun to talk to. Merge plus members were subscribers people all around the world with different kinds of job titles interested including ai in action building. That community is a big focus for us at twenty twenty one and we'd like for you to be part of so go ahead and follow us on social if you haven't already in otherwise thanks again for being a listener florida catching you in our next episode. Coming up on thursday. You're on the ai. In business podcast.
AI and the Evolution and Automation of Live Chat - with LivePerson CEO Robert LoCascio
How Retailers can Get Started with Personalization - Debjani Deb of ZineOne
"This is daniel fidel of emerge artificial intelligence research. And you're listening to the ai in business. Podcast when you think about industries that have changed radically due to covid. Nineteen none probably come to faster than retail. The transition to e commerce has been swift has been unprecedented. And there's a lot of changes in terms of how retailers are having to approach interaction with their customer yes moving to digital yes moving to online but also how do we keep loyalty. We don't get to see them in person. How do we compete with the big players like amazon who stepped up to such a massive extent during covid and be able to compete realistically in our own niches in with our own businesses. We speak this week with someone who has an awfully sharp take on that space. Deb johnny depp. She's the ceo of zion. Ones one is a a i based personalization firm in the bay area. They've raised some fifteen million dollars. And if you go on their homepage you'll see brands like men's warehouse and even some financial services firms that they've worked with this week. Johnny speaks to us about two topics number one. How is she seeing the retail world reeling and responding to cove nineteen. What are the changes that are happening. What's the perspective boots on the ground in retail. Secondly i think this is a really interesting. Take what is sort of the scale of maturity of personalization where can companies start. And how should they think about first projects for building recommendations in personalization into their experience. She lays out a number of potential use cases. Companies could apply as well as some rules of thumb about sifting and sorting through those projects and thought that they were awfully helpful. If you're interested in learning more use cases in terminology for a i in retail be sure to download our ai in retail chichi you can go to emc. Rj dot com slash r. e. t. one that's not like retail than the number one. You can simply download our ai and retail chichi for free. It's a pdf guide. That hopefully give you some additional use cases on top of what you'll learn in this episode that's r. rj dot com slash. Our eighteen one without further. Ado this an awful fun episode and a lot out of it. This is deb johnny. Deb with zayn one here on the a in business podcast. So did johnny. We're gonna be talking about the transitions forward in e commerce in retail world. You guys are in the cutting edge of recommendations personalization. But i think the space is fascinating by itself. But we're now in this whole shift of the cove era that we're not out of yet. Maybe you and i both would have thought by now. We would be but we're not. How do you see the e commerce world bending in shifting to accommodate customers in this new system. What's the zeitgeist changed from your perspective. One question daniel. It's been fascinating to see this. You know as covid hits you know everything to set up a bit of a lockdown shock and all right. I think it has evolved into fascinating said those aspects. You know we see that. Governor actually publish says certain parts of the commerce spectrum sporting goods a groceries electronics exercise equipment really is at the forefront like really sort of popping in regards to their performance had there is the brick and mortar stores that are suffering in having trouble because people are not as much in the malls. One thing that is common through across all of this really is that the move to digital what would have happened in three years really has been in a condensed time frame of six seven months. If you would where the direct to consumer revenue through ecommerce have really sort of evolved very fast has in fact. Put a lot of pressure on these folks to do a lot more very fast. I have conversations with you know very large grocers things that they were thinking with three years out in regards to buy online. Pick up at store area functionality. Things they've had to do it overnight so there are things in technology that got in amazing push via day at m l. Just pure automation. That would have been much further out happening in a month timeline. Yes so we're seeing. I think this common idea of this being an accelerator to existing digital trends. I think is true as true in retail ecommerce says anywhere else in the if not more so clearly. You're seeing that. One thing that i think is easy for a layperson to observe or even a market researcher like myself to observe. Is that many of the firms in the mid market or even smaller large firms mostly or in integrate many cases seem to be kind of modeling the functionality of the big players who are a little farther along you mentioned pickup app store or whatever. I don't know how long ago was that walmart. Put that stuff out. I imagine they were somewhere on the cutting edge there. there's all kinds of maybe online promotions. It folks are doing. What are the kinds of new ideas that are that are entering this space. And what percentage of those are directly inspired from whoever they looked to as bigger competitors versus drawing from other industries. What are you seeing. Start to populate the new capabilities space of digital so. I think you know this has been to a certain extent a little bit of a if you would the covid every for mid market to the lighthouse customers. If you will because now the common languages everybody got to be digital ride. So the pressure is equal in regards to the sort of evolution. So you know the things that we are seeing. You know commerce right. It's a big one you know the adoption off. Cd b.'s customer data platform so that they understand the customers in a common way has become very part of the conversation the deployment of a and m. l. technologies that allow them to react to these customers. Sort of better. In a more competitively you would has become part of that conversation. All of them sort of leading to so the what does it take to get that edge in very pressurized environment which sort of covid has enabled so to a certain extent. We are seeing the tips off a new things that are coming along such a stash less commerce such as the rise of data the rise of aml technologies and so on so forth. Yeah i guess this kinda leads us in the direction of the next question which is what the the maturity in life cycle of those technologies actually looks like because a lot of firms are just making this wild swing into digital. Everything's moving so much faster than maybe they would have presumed and they're thinking about where to get started. I think a lot of people admire what amazon does but there's so many things amazon doesn't mean we can leave the warehousing and logistics out of it. Just talk about customization terms of your email prompts in promotions customization terms of. What's on your homepage. What's on individual product ages on your add to cart pages a bunch of suggestions before you check all the way out. All of these things are different functionalities types of personalization. And i think if the small folks are going to compete upstairs with the big folks you better hope you have loyalty because if it's just who has more money to throw at the tech you're going to lose so having a better relationship feels so important but where to start there's an unlimited number of places where recommendation can get going. You walk us through how you see companies maturing from nothing on the personal side maybe sending out some you having a loyalty program or something to really being more on the cutting edge. Look like yes. And i think that's a great question. Ill you know. The personalization immature to are really has come into focus in these times reading. What will mean why that is that one of my big sort of bad needs have been you know that the way to reach the customer you know who lost ten fifteen years really have been okay. I'm going to send them a lot of email. I'm gonna send them a lot of push notifications sms and sort of a spray rates. Hurry but as the market has evolved become more wetted. Already had had pressure in this code framework. As i call it. You know people have been compelled to do more. So what i mean by that. What is the trajectory. Understand the customer so that you can be value added. that's the fundamental data is plenty. There is a lot of data. How you use that data to add value to your customer really is the question. So if you think about personalization majority curve from left to right at the very bottom left you think about things like doing better email where you know your customer so i do want to get an email about an abandoned guard. I would welcome that. But i don't want get thirty thousand emails about new products. It's not specific to me. So and then you go sort of sort of the next step right. I m walking into a store to get my both order. Know that on their geofencing yesterday are plenty. Know that i'm there at the moment. I'm they're sort of uber has made this common. You need to know the location and if you know the location you can help them in that moment. Telling them as the order ready. Who's coming out to get it. Rather than make them wonder sitting there. Going is go further up ahead thinking about okay. They just bought x. number of things. What else goes with it. is there a repetitive nature. To what they just bought it nice from five days from them saying are you running low on x. maybe is welcome instead off. Here is a long list of coupons that everybody's getting the same list and the only saving grace is how fast i can find the garbage can throw it out at so really about. I bought x. Tell me about why at the right time predict when i would need it predict the time between the visits when i would needed and let me have it when i come back. No one little bit about me or a lot about me. Because i've been shopping there for a while either online or in the store and give me the items that i most by so put them on top. The island of things on ish grocery shelf has been assigned twenty years. What do you that i level. Should we not do that with online as well. Should we not put time into. I buy petite i a petite woman but i never see the petite material when i log into a women's clothing store. Wyatt's things so bad nature you know when we talk about machine learning it's not about just very deep signs than just what scientists think about. It really is really fundamental features that add value that can be powered by the signs. Such as the ones we've been talking about that add value makes my experience better makes me feel like amion est or even though i'm shopping online. Yeah yeah yeah and so in. Our audience will be certainly familiar that you know hiring people from carnegie mellon just for the sake of putting data out rhythms for fun is not even remotely on the radar for our audience. Luckily where we're here to get some business goals done certainly build. Ai functionality and infrastructure and the things that the technology needs we can't pretend like it's just a like a hammer or mallet. There is some technical stuff. You're getting value is where we're headed. And you've given us some great initial portfolio items. I'm just gonna shake out a couple of these and get some more than you thought. So you mentioned the general lens is how can i add value to the customer great general lens and to some degree some of that has occurred even with the old school or stuff even before email. There were some direct marketing. Mailers that were. Let's say a little. You know there were a little aggressive. But they were still calibrated personalized. Whatever and they were split tested in different ways and whatever. So there's some there's some mentality of that in marketers that exists but we talked even about more than marketing you mentioned. Hey for my pickup. When i show up within you know a certain geo region. What's the prompt that should be in front of me about my order about related orders and about whatever else. What should that be okay. Let's think about it. What would be the follow-up order at how much longer later and that will come by sms or email k. We think about it when i log in my homepage could be altered based on historical purchases if i purchase clothing for small folks again. Can you show me the new stuff for small folks instead of having me. Click six times to find the things that are relevant for me. All of these fit the bill of the lands. You asked us to put on. I wonder are there places where people typically start. When he frame this question. A little bit differently. I say men. I sure would like to get into personalization. I'm gonna come up with a plethora of ideas. I mean our our research on the ecosystem of ecommerce vendor firms is something like sixty seventy percent of the venture. Money's going things that we could probably could. Under the umbrella. Personalization recommendation almost analysts amazingly big so hiccups in store email. Sms related purchases detection of a different user. Maybe i shouldn't show the small clothing to. I should pursue all of these things right. Do we generally have guidance on where to start. Is it as simple as what is technically most feasible. What data's most accessible and what's most valuable to our customer. We we kind of score things along those three criteria to figure out what projects to begin with. You probably have a more sophisticated way to think about it but really thinking from a company whose coming at this fresh. There's so much where do they get going. Yeah i really think about this as crawl walk run yeah certainly certainly sort of based on the majority of the company it place to start and typically it is about start with something small where you're not boiling the ocean people talk about omni channel and the ubiquity of data etc cetera. But nobody but nobody not even walmart gets there. Overnight takes years for the week about stock with a single channel and start with a use case. That doesn't require a lot of data. I'll give you an example danielle. The one that is very interesting. is that broad. We call it product. The lost city. How many other people are buying this in this moment in time. That supports the motivation feature. That really drives revenue and you can get started with it on a single channel with no data except for the data coming from your website. Homepage personalization simple easy. The other one is offers. You know the one thing that everybody who's offers driven company does is give the same offer to everybody than percent off today. Take well. I might just be there to buy some toothpaste. I don't need ten percent off by it anyways so there is a lot of signs of good signs that's available around offers an offer optimization. That's fairly easy to do. Makes a huge difference in regards to revenue and margins. That can be done really out of the box you know. There is a bar a promo bar in on most sites the same one that stays it easy piece of real estate. It's on brady much every website and you can use that to talk to the individual rather than everybody so instead of twenty percent off you can say hello to me and you can say something about me. That would make me feel on special. I can see what they're talking to me so there are a few things like that. That doesn't require boiling the ocean. You want to set your sights towards that longer term treasury. You do one a single view of your customer. You do want to know the customer as the drivers different channels as you have them. But that's not day one. This more newman task doing everything at the same time. I think it's really important for people to hear that from you. Because i think that boiling the ocean is common so is treating. I like. it's plug and play. You know it's it's obviously somewhere in the middle we. We do have a whole to get to this omni channel future world. We've got to have a better data infrastructure. We've gotta have team leaders understand the value of data and can integrated with it and can make this a fluid part of how we do business and move and shake those are real foundation level changes however we can begin with specific areas where we can see impact on marginally to begin with specific projects. That don't involve the wild infrastructure changes. We can get some momentum. See some dollars. Come in a question to run by you just because you know we get. See these amount of this from the enterprise side. Is it often. When people begin with personalization they find ways that are yes technically feasible maybe not super date intensive but also that if they bet for lack of better way of saying it if they flip up a lot of customers aren't gonna really notice it in some way so there's some kind of personalization around. I don't know maybe it's like you mentioned offers right. Maybe it's like an email offer where a backseat goes out. That's a little bit off not the end of the world versus if your homepage for every user for the whole twenty four hours is really messed up on. Of course you would only do a cohort anyway. But we've seen some folks experimental personalization ways that are a little bit less jarring the experience because enterprises are a little skeptical in. Because it's new for them. Do you see that being something that people have to balance in think about when they go for first projects yes. It's interesting actually asked that question. You know one of the things that i have seen across a customers as they have delved into personalization is that the mentality is destined. Learn this day. I will cut over to echo experienced that the final right so for folks that are supplying personalization to the market. They have to gear up for that. Same mentality of fast experimentation could troon experimentation and the ability to give the enterprise fahd to success. So i think it is. It is about control roofs. It is about starting where you're comfortable it is about starting with one use case and with. Aml you know. The idea really is about do things. That are easily explainable. Because you don't want to treat any algorithm as a black box because that doesn't really help the outcome you're trying to achieve you. Want this to be very explainable deliverable roi generating and ultimate satisfying to consumer. You're trying to serve so there is a baby step again crawl and do it with testing testing and testing more testing to get your full deployment big time yet. I think marketing personalization. I think naturally lead to this kind of iterative approach. Ai needs that. Anyway so i think that's kind of a in my that's good news it's great we have a mindset that's maybe fitting a that's a beautiful thing and you guys have seen these kinds of experiences with big brands like north face a new working financial services so those of you listening in these are words in the wise. You've seen this stuff Hit the ground running and depth johnny. And that's we had time. But i appreciate you sharing some of that wisdom with us here on the show. Thanks so much. thank you for having me. This was a great conversation. So that's all for this episode of the business. Podcast a big. Thank you to deb. Johnny for being able to join us on this episode. And thank you to you for listening all the way through if you want to support the shell and you like what you're hearing here. Be sure to drop us a review on apple podcasts. It really does mean the world when we get reviews from you folks not only do we use that information to build a better editorial calendar put together better interviews for you but also it helps other people know about the show so if you want to support us go to apple podcast find a and business leave us a five star of you and let us know what you want to hear more of or any episodes of really made an impact for you. It doesn't mean the world to me and my team to read those from listeners. Like yourself otherwise. I'm glad you were able to join us for this episode. And i look forward to seeing you next episode here on. The airline business podcast.
Applying NLP in FinServ, Practical Considerations - with Luca Scagliarini of Expert.ai (formerly Expert System)
"This is daniel fidel. And you're listening to the financial services podcast. There are a lot of potential use cases for a financial services. Let me see if you can tell what these three habit common. The first is a conversational interface. The second is the summer ization of call center. Data calls from our customer support operations and the third is documents search and discovery. But they all have in. Common is reliance on natural language processing applying. Nlp can happen in all different dark corners of banking and financial services. And this interview we speak with lucas scaglia. Rini luca is with expert system expert system for those of you who are enterprise customers here with emerge. You're aware of expert system from our ai. In banking opportunity landscape anybody who's ads access to that data as part of their work with us here to merge will be aware of the various products of experts system one of the over one hundred companies covered in that research and a big part of what they do. Lucas speaks to us about. Serve the differences between machine learning and other. Ai approaches and some of the practical considerations for applying natural language processing inefficiential services context. When is it the right move. When is it the wrong move. And what are some of the things that companies often get wrong. They could do better to have more success with their. Nlp applications if you're interested in bringing nlp to life in your business be sure to download our unlocking the business value of nlp pdf guide. It's short pdf brief. You can download it e. m. e. r. j. dot com slash. Nlp one that's nlp like natural language processing and then just the number one yemi rj dot com slash p. One again that's unlocking the business value of nlp if you're interested in more use cases going deeper with actual application checkout that guide otherwise. Enjoy this interview with luca. Glad he could join us on this episode with expert system. you're on the a infinity services. Podcast so luca. I want to kick things off by talking to you about serve the as you framed it the war between symbolic and machine learning techniques for nlp. I know there's kind of a wrestling match year as to which is better which is worse which is going to win. You've got a little bit of a different take on this. Can you talk a little bit about how you see that kind of wrestling match on. What i think is that i think trying to fight for with is the perfect solution for any possible applications kind of waste of time. I think it's because we have been dealing with this now for years. It should be clear that one technique like machine learning is pretty good in doing a certain kinds of things under certain scenarios for sure the best way to apply to business use cases and then it's also obvious that You know that in certain circumstances especially when you don't have a lot of training You know sad or you don't want to spend the hundreds of thousands of dollars just to train a system that then you need to retrain to improve the performance in approach is more symbolic my symbolic i mean. An approach is more similar to the way. The human brain works in terms of You know understanding concepts relationship between concepts. I think these kind of techniques can work much better in certain circumstances so i think that what we would see in the future and referring to that in the future in general application to nlp to business but obviously being the financing of being one of the first kind of adapter. It's probably way we'd see these. I at a kind of a makes What we started referring and an a person is referring hybrid approach so the possibility to actually understand the case. Understand what are the what is the specific situation and then based on those contextual information really peak models are mainly based on machine learning mail based on on symbolic. I really think this is kind of Appearing and i think it's going to be probably even more common in in the future. Maybe we could even talk about instances where you know as you had said a big overall. Who's going to win. The whole game is kind of a silly way to think about it. There's going to be different approaches for different kinds of applications and based on performance and based on maybe energy use in cycles or whatever we're trying to optimize for there's going to be other ways you in different ways to skin the cat for you. Are there instances where you know maybe representative instances in financial services where we could talk about the symbolic really being what we wanna lean on and maybe not making as much sense to machine learning and maybe situations where machine learning tends to be what we wanna lean on for. Nfl as as we can kind of. Put those in buckets and talk about when each might have their place without oversimplifying things. But i think that there are pretty obvious situations in which you know machine learning requires even if obviously the training requirements are not as demanding as they wear in the past but still machine learning techniques need to learn from examples right so he situations where use cases present naturally for an enterprise a lot of existing samples so a lot of ways where you to start from. They are a perfect candidate for For machine learning. I'm referring for example. The chat area is an area where machine learning fines very media. It's obviously superior because you'll have the samples the variety then the variance of the questions tend to be always in eighty twenty. So with the you know the the twenty percent of questions to actually answer eighty percent of their parents. It fits pretty well. There are other situations in which these kind of reliability of A big set of data to train the system is not there and so you know having a system that is based the knowledge and existing knowledge is knowledge graph and that has an understanding of language these independent from the training so it's sexually based on On general sani of the language need to deal with finds a much more natural application. I'm using some examples. That are maybe not so. Common depends on vary from country to country but for example the area of managing automatically garnishments for example that requires a lot of going after reading in deep in depth documents extracting and for. You don't have samples that can really be representative all the cases and now the situation in field that is linked to the financial institution is for example claims management insurance or even areas like Contract contract is a contract comparison comparing a contract with a new version of the contract. You make have to make sure that all the exclusions are over represented correctly. Can the inclusion are represented correctly. All of these are situations that usually are. He does that from a major train. Sat. and so those situations where naturally symbolic a person to be. Better and let me add. One thing is that is not that The situation loves to change change over time. And that's the example of boats that have to deal suddenly after kobe with a lot of queries from customers about things that will never mentioned before is a good example. Also maybe pick the machine during this structure change and then suddenly you untapped to train. The system is so again i am. I'm here to say that it's really a question of not being is not like a religious war you know. It's not the is just. That is just a question of being pragmatic impractical. Yup and that will depend on the circumstance circumstance. Being the use case circumstance. Also being the time like you said maybe if you're leveraging machine learning system but the the world changes radically maybe that's no longer the best suited use case so hopefully some of these distinctions. You've called out are useful for people listening in now in wondering sort of. How do they want to approach things in hopefully for the folks listening in if you hear somebody really making this into you. Know as luca put a religious war as opposed to a practical consideration. You'll know that that's not exactly a strong footing to be standing on. If you're really trying to solve business problems speaking of business problems one facet that you and i chatted about before the interview here other wanted to get into was around the kind of acquisition and cleaning of data when people think about nlp. They think about often what it can do. Oh you know. I want to have a chat bot customers. Oh i want to search for documents or find certain clauses in in my legal contracts or whatever the case may be but they often aren't thinking quite as much about what it takes to acquire and clean data so that. Nlp can actually do. Its job a wanted to know maybe for you if you could kind of t up for the audience really. What if some of the big concerns there as business leaders who might not be writing the code. What do we need to understand. And think about when we're thinking about using nlp in this respect for position cleaning. Yeah i'm i'm making some some simple examples because obviously this can become pretty technical brady fast. For example i can see the Like the need to. Let's say process incoming documents in this in different formats okay so it is obvious that you can have a significant noise in input if this format is not ideal you know like for example or cr are working extremely well but it's enough that the document needs to be yard. It's the facts for example that's that's-that's make a typical example. You lose a lot of quality important so even if it seems in our structure document it seems something should be handled pretty easily than the turning that into a digital version. You include so much. The noise than the downstream part of the process is pretty critical and other examples are documents that are for example say structured right so let's say documents include tables and all of these are areas that for a business person the app but it's a document the document and so for me the fact that that's three tables or the fact that taza just paragraph it's exactly the same because he includes unstructured information and it's correct but the difference in doing a great job to turn the table into data can be used makes all the difference in the final solutions i i i spoke sometimes with customers that give me like you know handwritten documents even if now there are dr are software that turn a handwritten documents into digital aversion steel data a lot of noise that this kind of middle of the road the middleware software actually creating the price so i think that downplaying the need of having a a very rich way and very once again. It's not the one a one-size-fits-all like having the right set of tools that enable you to do the the cleaning it's perceived as being extremely important in that in that issues were data numbers. But there's not the same perception or at least the same understanding in In nlp so again. i make. I made very simple examples. I'm sure that people that are more technical. They conceive them. Maybe not not very intelligent examples but the reality is that these are examples that everybody can understand and and the way where you have issues in terms of not having a clean. Acquisition of data can be represented by those examples. Cool and when. I think about The consideration bring up your on acquisition cleaning. I sometimes think about how much transformation has to happen to the business in order to bring ai to life a lot of the time if we're stepping into a universe where the data is really gobbledygook. And it's not not exactly easy to set up in integrate. A i in clean things well enough so we can have some outputs that are useful. Sometimes we just need a system that can drink in that ugly data and clean it reliably and then use that to train a model other times we need to talk to a client potentially around kind of changing how they date infrastructure in the first one is changing how they do intake in the first place. Is there a bit of a wrestling match. You know when we stepped into clients and arianna workflow between. Hey what can we just vacuum clean and tweak up in the back end in our of our own system versus what we really need you to systematically kinda fix for the sake of your data health long-term mr client so that we can actually make this happen. Is there often a bit of both luca. I mean it's a very very good point and So i think again. It's a question of if you want you. Can i think there's this perception that is you know magical ride that they saw gold possible issues because those warling artificial intelligence building the definition reality all these aspects. That are extremely important. So and i'm bringing it when even furthest step is not only about The way you actually structure sometimes need to have work to be done before you actually can implement official intelligence but also the way you have designed your processes you cannot do a autonation which includes usually a piece of natural anger understanding if you don't have processes that can be automated so the same way it applies data so i think we need to need to be honest. I mean the point. Is that the successor away. I is really around turning things being practical. I should be perceived as an investment in any kind of other software. It's not something that is you know kind of In its own completely different category and when you face reality and it we practical pragmatic chooses the solution that might be eighty percent today and then do the pre work to make sure that you can turn to under percents in the course of the next month. I think this is a important for people that are enterprises. The dr initiate investing in initiating journey inside the world over deficient. But he's also something that vendors need to be much more kind of You know practical in the communication the custom. Sometimes things cannot be done. And the risk of putting together in something that is unsuccessful than it impacts the overall perception of the enterprise of this new technology that can bring value today if you follow very practical and pragmatic approach. Speaking to that point. luca. I know that the ai vendors that we've spoken with over the years who have mature companies really have had to learn a lot about what it takes to work with clients and have a bit of a white glove approach to teaching folks may be what would it actually takes to bring ai into an organization. Sometimes it's easy. Sometimes it's plug and play most of the time it's not when it comes to serve encouraging clients to invest in their own data infrastructure the cleanliness and access whether it be real time access or just access in general to their data. Sometimes that feels like this. Extra hurdle is extra cost to leveraging an ai system but in fact there's another side of that where it's also an extra bit of maturity that we can kind of invest in so we can use that data more productively for thousand applications in the future. How you speak with clients about being able to and willing to invest in maturity maybe the kinds of maturity wouldn't need for traditional it system. But something that may be a i would need. What does it look like to kind of explain that in a way that executives understand it can make a rational decision about Find the usually a kind of productivity to consider that the arrow i off An investment in deficient intelligence sort of compounded effect. It means that the the kind of investment. You're doing maybe to do the first implementation that could be the two example to me right to create an infrastructure for data that can be acquired in a much more effective way or could be that to designing a process that is actually designed with mission in mind both those investments and it could be also creating a sort of you know kind of general knowledge across the enterprise. Zones is a knowledge. Graph three describes the language of the there. What needs to understand. Is that the return on that. Investment is not the only in the first project that they actually put in place. Their return investment is as a multiplication fact. When you actually turn these seen to cover during different areas in your organization and that's where you know having approach that he's around creating if you on actively intelligence as sort of infrastructure for the company where today you implement your shot tomorrow implement your contract comparison. The next day you implement something to automate europe wants to create the infrastructure than you actually can compound the effect of the return on investment. So when you actually face the discussion in terms it's something that i think Executive reading understand very well. And then that can remake the difference. Compared to do kind of you know kind of set them forget kind of strategy. Well i think this is a really important way to think about things for the listeners. Who are tuned in. We have an article called critical capabilities. You go to google you. Type in emerged critical capabilities. Get an understanding of some of these sort of prerequisites day i deployment data infrastructure being one of many. These are things that luca as you're kind of articulating. Now we're talking about sometimes. They're seen as hurdles right. A company will think to themselves that they're not familiar with ai. Will jeez this is a lot of work. I mean i just wanna make this chat bot happen but you know what you and i are talking about the fact that he s. Maybe these are additional time and money. But also this. This is how we unlock future capabilities. I'd love your thoughts on this as someone you guys are. Pretty mature vendor company had been around for decades very few. Ai companies have been around as long as you guys have in terms of seeing that maturity investment. Be something that clicks with executives. I wonder what that really implies. We could talk about conceptually. Hey you can upgrade your your data now and then in the future you can also use chat bots can also do this. There's a bit of a wrestling match here. Because it feels like number one we could just talk about like we could talk about all the different things we could do with the data and then in all those different circumstances. What's the foundation of that infrastructure. That what would need to look like to enable all the cool things we wanna do in the future. That's one thing a second thing would be. We have a single project and we just upgrade the infrastructure for that single project. It feels like the second one might be less abrasive to the buyer because oh it's less of an investment it's less thinking in strategy after bringing less subject matter experts but it's also unlocking less of that grand future potential weird. He kind of draw the line between really opening up the conversation about what we want to create mr buyer for everything we can unlock together. And you can unlock in the future versus. Hey what do we need to unlock to get you the darn documents search program that you hired us to put in place in the first place. How do you factor in both of those conversations because it really feels like they both have merits. I mean obviously. They both have married. I think one one aspect to think about. It's also how things are changing with time. Okay so obviously. There was much more resistance or at least was perceived much more scifi dot com Four five years ago right so The objective was to find these kind of fast win. That may be. It was creating the end of the day. What gartner caused the technical a gap right. So you find. You do the minimum required to do that solution. makes sense economically. You are out of that solution but then if you have to either one you need to redo things not from scratch but palmos. I think this was sort of tactical way. F- years ago. That was probably the only way to actually go into into the market. I think that there's enough awareness right now to understand that. Implementing is radius structure change for the organization. Where this discussion around. We are not here to boil the ocean but if you just focus on the limited things enable you to put in place one one solution. You're actually missing an actually. You're not creating the system that enabled you to leverage more. I think there's more awareness. And i think that i'm talking with our experienced for sure. Other enders have seen in our experience. Is that now. You can actually count on showing things that happened in other organizations around using approach that you know we have we. I'm sure i saw the vendor have several companies that have for whatever reason a combination of More of innovative approach the combination of the right people right moment have chosen to avoid the easy win or just focusing on these wing Stop this kind of infrastructure. And then they start getting the actual return so now few years went by and you can actually talk Nine ten twelve project in production. That are all around and be for example blase seeming images so so you can actually see that you have this proof if you wanted something that is changed in the last few years yet is. I'm trying to come up with a nutshell insight. Luca that we can end the interview with year. Maybe one way to think about this. Is that on. The one end of the spectrum we do the absolute bare minimum for any kind of integration. In order to just sort of plug in the solution is the downside here is that we don't build maturity in terms of teams skills culture importantly data infrastructure. But we kind of get it up. Maybe we can even get some results with it. The other end of the spectrum is with a goal in mind. We just think of one hundred things we could do with a and we think about the foundation we could build on that would really be the most fluid set of teams in skills and culture date infrastructure. That's go unlock all the future. Ai stuff that we wanna work on that second. One is pretty unrealistic. Nobody's going to invest purely in hypothetical in foundations. The the first one is a bit immature. I think most of our listeners certainly are aware that a with no focus on on the maturity is a pretty bad way to go it. Sounds like all folks that are procuring. These technologies luca thinking about them in the future need to be asking the question. How much maturity do we need to realistically think about or would it be responsible to realistically think about uh before we think about deploying this tax maybe we should have that conversation have that strategic talk about how deep we wanna stick the tendrils just to get quote unquote this one application up and running. It is that something you wish. More procurement folks would pause and think about. Or maybe there another way you would like to to nutshell this idea. I think i think it's It's a framework. What is what is. Also i think helping this discussion right now is that the discussion is not around. Only like one point solution fairly muted and something that is undefined. Because i can do hundreds of things but right now. I don't know exactly. What are these under the things i think that right now there is enough maturity in visibility on also amer. Jay has great content around practical implementation. Right so let's say. The approach new financial enterprise. Now is not that you will there with one point. Sushi can go there with already in the moment when you ask. For this kind of alignment you can say eight. This is one but you can already see in your eyes on without any major kind of are not talking about things not define Duke three four two three four possible implementation that you should actually account for if you do this initial step. So that's i think. What is changing the more visibility on. It's not really pure low hanging fruit but a lot visibility on practical implementation that can point you in the moment when you're thinking about the first solution saying hey. I already see in a reasonable timeframe. If i do this. Investment to completely leverage it also on investment to on use case to use case raiders. Because for and i think this is something different from what it was just a few years ago when you know this scenario is not not likely. Got it again and again. I think you're right. I think there's there's been a sea change in terms of thinking. Part of that is around what we refer to as executive a fluency that is to say does the leadership even understand that. Ai maturity is i think the fact of the matter is for many many years. That answer was a square and firm. No pretty much across the board and and so nobody wanted any maturity investments because it was all extra gobbledygook that the regular. It vendors didn't ask for but but now we've got hopefully more mature folks in part because of smart vendors like yourself we're educating people and hopefully in part because of folks tuning into our show reading materials on deployment. Ai are y. maybe luca. We're lucky we did our part here today. So i I appreciate you being able to join us and enlighten us. A little bit as to the practical realities of nlp and glad to have you on the program again. Look thank you. Thank you very much and you know whoever's listening and they want to reach out to me for any genetic question we don't have to speak only about about this but just unethically about this kind of adoption i think. Can you know. I think we can have a good exchange. Our way slopped. Dr people who actually are facing practically so thank you very much. That's all for this episode of the financial services. Podcast if you like what you're hearing here be sure to leave us. A five star review on itunes because this is our newer and are smaller. Podcast every review really matters a lot plus we take all of our review quotes and look over them every single monday. He would emerge so when it comes to determining our editorial calendar the first point that we check in on is what folks have to say about our podcast and any lincoln comments get our actual written article so your comments really do matter in crafting a great show. We appreciate your listening. And we'd appreciate your thoughts again. Drop us five star review on apple podcast. Easy to find the in financial services podcast. Otherwise we'll catch you next month for next episode. Thanks much for being here.
[AI Futures] Forging International Consensus About the Future of Intelligence - with Jerome Glenn of The Millennium Project (S1E9)
"This is Daniel Fidel in you're listening to our Saturday futures series. This twelve part series is about artificial intelligence. In, this episode nine in this series, we've talked about near-term governance considerations for AI and also long-term. What's IT GONNA take it both a national and international level to sort of forge a future with it's very powerful technology in the decades ahead that hopefully be an aggregate good one, not one that instigates more economic disparity or war or other negative consequences that we'd like to avoid, and there are few people who've done more futures thinking and thinking about policy the future of governments future of enterprises I ended mapping those futures out then Jerome Glenn who is the founder of the Millennium Project to for a quarter of a century has been running the. Millennium Project, and essentially doing just that working with governments working with large organizations from the Red Cross to the government of Korea to sort of map out what the future looks like under different circumstances into poll together varied stakeholders often internationally to think about how we can get along what kind of future we want to craft and what's going to be a forged win win scenario for that future. Maybe we can do to prepare for it We speak with Jerome this week about what that would look for artificial general intelligence towards the latter part of this twelve part series, and we're on episode nine. So this is the back half for sure. We're GONNA be talking more and more about the long term consequences of strong ai when this technology becomes very powerful. The big question in today's episode is will, what does it mean to prepare for that? Who needs to be at the table in order to make sure that that's a peaceful transition? What kind of questions do we need to ask in order to ensure that we're sort of buffering against risks in so much as we can actually do that and jerome with his experience and exactly the space sort of breaks that down in depth I'd love to get your thoughts on this episode in this series in general, you can go to e.. N. E. R. J., DOT COM slash pod three to two question survey. It's not even a survey. It's just a couple of short fields you can share your thoughts. I'd love to know if you WanNa see this as a separate podcast if you like it on Saturdays, your ideas really matter. So please do share them there. We've gotten dozens of responses already in it's really helping me to kind of craft what the next day I'd future series might look like but I'd love to get your ideas as well. So without further ADO, we're GONNA. Dive into this episode this is Jerome Glenn With the Millennium Project here on this special AI? Futures Series. So Jerome got a lot to talk about here in terms of artificial intelligence governance, artificial general intelligence. The reason I think this conversation will be fun is because you've thought through some future scenarios with with very large organizations for many years. Very High Level and you've learned a lot in the process of what is the process for pulling together different stakeholders imagining, what will the future be? What should we do I? Mean very complicated. You go about it. Of course, one of the first things you do is you gotta find out the state of the art of whatever it is you know is there is, let's say five elements to it or ten elements, and you know was the state of the art on this element on this element, this element, this element. Now myself I won't know enough to do that. So we have a global network of networks sixty five nodes. Return Network himself within countries, and so I can say, here's where we are so far and they tell me what else ought to be considered. So there's so as global sort of a state of the art assessment finger. Yeah and then within that with take a look and say what questions were not asked the authorities have been asked. and. What questions were as but answered, superficial. That gives us questions to as in a Delphi study, which is a questionnaire goes around the world. And the results of that then becomes guts content to create draft scenarios. We send address narrows back out and everybody hasn't at Pat and presides over, and then we can say, okay, what do you do about this scenario? What did you do about it? You'll see a good action as well as scenarios this sort of a general approach So you talked about the Delphi study I actually recall you bringing this up the first time you and I chatted I don't remember who has five years ago or something wild like that. Speak briefly about wooded Delphi study is so I like finger on the pulse what are we missing? Pulling, those ideas together and then there's this kind of dispersion to generate even more. What is the Delphi study. Delphi questionnaire. Whose second round. Is. Determined by the results of the first round. And third round is determined by the results the second route. the reason for it was that there were generals and admirals and experts that don't always the same room with each other at the Rand Corporation. The Rand Corporation had to figure out how to beat World War Three. Well, we didn't know a lot about that beginning, but you had all these brilliant people that don't always cooperate. And sometimes in the military, sometimes, people were afraid to criticize admiral if there are only a cap yeah you avoid all this crap by saying here's around one tended out and they response good. But without a names, no one knows that you're a private or a president is state. and. So the ideas become persuasive rather than personalities or your. Then, all those responses but also happens to sometimes someone doesn't respond to somebody else's idea. If they're in person Zanele on that yeah all but then the second round, those ideas on the first are in there. So you have to respond to all that second first-round stop that you wouldn't normally have responded to. The same thing goes on those results. Everything has a chance to be responded to without name rank serial number so that ideas become persuasive by itself. So it was a way to collect intelligence at a have it learned through innovations so it wasn't just saying what's the state of the art of thinking is like you're gonNA improve state of the art is thanking as itself yeah. A I I go by. And that's brilliant. It's Great I. Still Think it's one of the best there is however what happens if you're in a hurry? Light with Kopech nine stuff you know you. Aren't working on that one. Now, you're going to do what we call a real time Delphi. That came up because of time. Sanjay questionnaire. But he will have a sign insulin come back later so you can vote early off. So the idea is, let's say you responded by by just off and we asked you explain there. So they say on a scale of one to ten, it's like five then when it comes to explain yourself. So the people see why you said what you said without your name out you're right same way, and so then come back two days later they later and I said you know I see somebody misinterpreted. So, I can go back my original text and edited to make it clear although as thought. However in Singapore, they did an avid poker contact range this way there's a web link. Then another person comes later sees that. And says, yeah but the Green. Ones Better Watch this. And so you're you're you're getting the CPAC stuff, but then you can say to somebody this Delphi it goes live today at new and we finished in one week. This is called real time Delphi got it and so the original approaches said was developed by the Rand Corporation Right Yep. As a matter of fact, partner in crime had Gordon co-founded when project was part of that original team Okay. There we go. So there some of the the the origin story here. So when it comes to congealing all of those thoughts I, mean some of your past projects I'm trying to think of an and you you do a better job is around you lifted what's most analogous to this project you And I are talking about, which is around the requirements for global governance in what the means of global governance for artificial general intelligence obviously never tackled specifically similar futurist type projects, technology type projects power in government type projects. I'm trying to think about once you've congealed all of those answers from different sorts of parties will talk about who those parties are. Then you gotta turn that into maybe more or less likely future scenarios to some degree, and that also feels like hard work but but go on about what happens after the Delphi. We're not doing necessarily more and likely scenario. The idea is to do plausible scenarios that are useful for thinking. Right, so or example. Currently. Answer to your question related stuff. We just finished a three year study on the future of technology and work. Obviously, a is a big part of that including Janelle Washer, yeah including super they could goes out twenty fifty. Nine not get super, two, thousand, fifty, but we. On the exactly why you alternate snares here's what he hit. Get Etc Sofa. So the way that we do this is we take all that content. And some of his positive negative. Obviously the negative stuff gives me the stuff for negatives Mary positive stuff. And then mixed now. Futures will argue against doing graphic. Way of doing Herman, Kahn also, Rand Corporation also a former friend worship longer live. Did the positive negative in middle because that's the way we think season. Since then a lot of futures now the trouble is that people plan for the Middle One. It becomes a self fulfilling process. However. That is evolved into something of its superficial. Which is people say we'll pay two or more uncertainties. Like you get super I don't get super high you get high unemployment don't get high on. So that makes you a two by two grid. The desert become forced air then what they superficial. Now. Describe that state of the future in quantity fifty or whatever. Aren't that's fine. That's good. Then you say. What scenario? What strategy should? I should I have the works in all of those? All right. That's the normal sort of superficial scenario plant, which is better than doing nothing on argue against because it's easy for people to do consultants work with it into understand. The real reason for joins narrows originally was. Her write a story reels Murray not describing a state. Like in a movie. A. Zach you have to have. With plausible stuff not that's the truth but Klaus we'll start why? Because as you write a scenario, you'll get to a point. You say, I have no idea what happens next this is crazy. I never thought of Oma John We've got to stop and you stop and then you do your research, call your friends you know all all that sort of stuff to fill in. The value, the original value of a scenario. was to force you into a position where you and your colleagues have insights into what you didn't even know to ask. Quick Vignette. Corporation was ideas was what happens if there's years ago? What happens? No thermonuclear war doesn't occur in years. But in Orbison to crisis comes up and you might have. While Caused. Trouble with thirty year cat. Is that you you? You don't know who's in the Kremlin? Furthermore International Affairs have changed enough. You may be looking at Beijing, etc. Right. So the whole idea of mutual assured destruction was you had to prove to your opponent you're crazy enough to press the button. How you convince an unknown power structure in an unknown political geopolitical on the world that you're crazy enough press the button. They didn't know. So you stop writing. And you do research and you think all right and they don't do that anymore. So you're stop and say I don't know all right now what they did was they came up with fallout shelters. Your original reason for fallout shelters had nothing to do with people being thousands of years underneath the ground. Would it had to do with his convincing your opponents? That you're crazy. Enough press the button. How by having real fallout shelters and then you say everybody go. So the in this scenario, the your opponents sees news stories video. Mass crowds in New York City going up massive crowds in. Chicago mass crowds Los Angeles the whole country is going to these fallout jobs. What does the opponent? Thank you guys are crazy. You're actually going to go to war. That was the purpose but you couldn't say that purpose during the Cold War because it would make the whole idea in ballot. Right yeah you never heard about this. But there's an example of where you got to figure out what you don't know as important. So in businesses say Weinstein one because they didn't look there look there they didn't go to the unknown questions. Got It. So the purpose of you're saying you collect all these ideas we sort of map forward. You're saying the hard work to some degree is thinking through these plots and then stopping at like to be honest I have no idea with this combined with. This zero clue as to what's going to go down I need to talk to folks that know those areas I need to speak to people from her from those go regions whatever the case may be. So so this help us flesh out, you know how wide is this suite of plausible scenarios because of course, you could imagine it being infinite, but you don't want it to be infinite. You have to boil down the representatives set that will be useful for thought. And informed for policy for for future action, I'm bound that reality with implausible scenarios that are a limited number. Well. There's two approaches. One approach is the one I just mentioned i. Was your plausible negative worcester possible positive and what's your mix the other courses. As I mentioned, you take your various unknowns that you really want to know about and you make your matrix can have you know a whole mess different scenarios that way and then you pick out of that mix what do you think is the most interesting? What really brings up the unknown questions and so forth. That we have asked before a so and then you have yourself a little steering committee who nitpicks you to death? And you come up with a consensus saying, okay these are the ones we wanted to approach. Now in the process of writing them, this is another good method performance in the process of writing. Is. Ours are develop outside of the original expectation. Let. Avon. A method should not be a prison cell. Method should help you. So as you right 'cause I heard while the story started get outside of the box lot at because you're trying to find the dynamics of cause-effect. Current you didn't think about before so it should evolve beyond your original expectation then you're learning. Got It, you mentioned the steering committee on some level should be open to that learning and be and be responsible to saying, okay. The these are the plausible scenarios we ought to discuss that we believe are most likely to be worth considering I. WanNa Pivot towards your eventually gonna be touching more and more on the governance of artificial general intelligence I happen. To believe that this is such a complicated scenario that it will imply some kind of a pooling of thought rather than hypothesizing in some academic tower or by some some individual brilliant scientists when it comes to coming up with ideas about how the future will be. Sometimes those have to do with real power struggles with the territory lines with the policies of. Across borders and boundaries. This process are there instances of this process where we've talked about very contentious issues with? Competitor's more or less, and are there any unique insights from that? Oh absolutely this is one of the essential assumption on the money and project his to global futures research. Locally not to have a PhD from Beijing. Moscow. At Harvard, with both got PhD's from Stanford know that's not global research. So. He leaves that like we have no in in Beijing and a node in Tehran as well as the than an silicon balance Cetera. So by having a mix of these people you onto the address questions but a question then your question, I assume going into the eye. On as we're preparing to work on a government study. One of CO partners will be the Chinese Academy of Social Science Institutes. Of saw technology saw technology the legal stuff all the well, we're working on this technology of a on. So, this is. Including, China in the conversation from round one. Because if we're going to create these international agreements and treaties and governed systems gotta how? China. And others gang so we start them to begin with. And so yeah, clearly need the stakeholders their right to presume that I'll very contentious issue. Let's just talked to one party will figure out what they think. The other party will think them will bring it to the world right? Even if you did drop something great, you know there'd be zero trust serve baked into that process. So clearly, it sounds a giving people in equal footing kind of to get off the ground is one thing it seems to me that to get folks to agree or to to at least have some degree of I guess maybe consensus is not exactly the right word but to. Serve come to the same page on some topics. There would have to be a lot of that oscillating these different plausible scenarios and say, which are the ones that are digestible for both of us ways. One of the things we're looking forward the things that they're looking for you know right now I think in the artificial intelligence race, the US China's kind of the big consideration. Of course I think we should be thinking about the developing world should be thinking about the future of Europe we should be thinking about other players organized crime because they've got the cash to buy the best software engineers in the world. And they got the institutional savvy. You create a whole bunch of middlemen corporations so that you think you're were. Is and you're not. So we're include that as well. How do you pull those folks into the conversation? It'd be like well, you know we really need a representative from organized crime. Well Nice thing about the money of project people sometimes just show up. And you never quite know why but that's okay. I figure if they can find the back door, they're smart enough to walk in a living room. Anyway. But one part of that is the UN has the drugs and Crime you're and they look at all organized. Now. There's a guy involves in that who is now part of the UN stuff on a I application Irakly directly you? mentioned. Sure that. I don't think. I. Don't think he's that he probably won't be all that offended Iraqis. Proud. About saying Oh. Goodness. All right. Well, there's a chance I'll edit this out, but there's a chance I want to go go on going well, the idea is that people who have more. Frontline encounters we WANNA have in. And but this brings up of essence also. You don't want to give your opponent. Crimes opponents speak. Ideas. And and and our work for better for worse. I did stop. So we don't WanNa make them smarter. But we would like to figure out social judo and s by as far as I'll go. Okay. Interesting. So there's there's there's considerations there to around. Can We? Can we extract from? I, mean. This is this sounds like international politics intelligence in the first place can we can we pull in everything that we wanna pull in but not be permeable ourselves to have folks learn things And there's a parallel here. This is complex. Helped me remember why I started on this. Is Self interest of organized crime to take seriously why has every go to General? We don't know how long it will take to get to. Super. Goes to super without initial conditions being in good shape. We're toes. Now is a parallel. During the Cold War. Carl Sagan went running around saying. If. You have always explosions get enough dust in the air thus clouds in the air that you knock off your vegetation. So I don't care whether Moscow or Washington has a first strike defense or a second strike defense reviser both toast. So you've got to stop it because there's no winner. No matter what you do. No matter what you do you lose. All right. So the same thing we're saying, no matter what organized crime does. They're. GonNa lose with the rest of us. If we don't get the generally I write that moving this Super A. A soubriquet I can mess around organized crime just like they could mess around with you. Yeah. Yeah. Yeah. So I'm thinking forward or I guess we'll close the gap more on Agi and on Super Intelligence you know I'm imagining going through this process with you know high up folks in China High Up folks in the United States and it start at one level I. Imagine there might be a Delphi cycle of sorts and possible scenario through, and then maybe it will go up to a higher level and hopefully there will be more powerful folks involved. It would seem as though it would be challenging to come to a same page conclusion about sort of where Agi. Should land whether it be in whose hands or representing what kinds of values or it doesn't do this but it can do this. You know on these are impossible things but you know we might as well take a swing because if you know imminent destruction the alternative, why not? What is the approach to wiggle around advocates? You know you can imagine all these all these human rights, things that Europe in the US are going to tout may not exactly be you know privacy and whatnot may not exactly be China's Cup of tea no matter how you slice it. How you frame cut is a process like this help try to find middle ground if there is one. Latouche steps. You can have. Workshops national workshops or the workshops in different parts of the world. They can say, okay what do we do about that stuff? So then they have a bunch of suggestions we pull all those new did this by the way the future work and technology studies saying way we got hundreds and hundreds of suggestions these workshops around the world. A lot or lapse of course for many people think the same thing. But then we narrowed down to like about ninety or so or hundred specific actions, and then divided those into different categories whereas governor do business do. New and so forth. All right. Then we send those out is dealt by five separate Delphi's and say now was the good the bad about this, the likelihood to pause, and so then we give back to everybody. Here's the menu of actions. Here's the commentary on those actions. Here's what's been perceived to be the the feasibilities actions. And the likelihood. So if you're in country X., you might pull out that whole menu. Options that are relevant to you you another country might pull out from. So so they don't have to have world agreement. Will we have to have is a better on rotation that we have right now? Now it's superficial and it's not serious long range. So this will move the conversation just like it worked at. That has moved conversation quite a long by now. Got It so. The idea would be to concretize Internet just like well, we disagree about things well, I want power you our but to say, all right well, here's all the ways it could progress. Here's the international governance structures. Here's the local kinds of governance structure. Here's the modes of checks and balances, and here's will be good or bad about these. These different approaches years with this party thinks about it, but this part thinks about it. And maybe that would concretize. Okay. Here are things where there seems to be almost ubiquitous agreement. That's great. Here seems to be the biggest faultlines and then hypothetically you could run another cycle about handling those faultlines. But at least now we know what's contentious what's not where we agree and maybe brought categorize the points where friction exists so that we can work on them. It sounds like if nothing else that would be the output. Take a look at the the climate change. The GIOTTO protocol is a lot different than the Paris agreement. Go on his for the folks who are who are at home in are familiar with the core differences there go on. Okay Kyoto Protocol the original ideas about addressing climates. It did not have much specificity at all. And Set out the door. And then as. Of the meetings occurred going into the climate change conferences against more precise, more precise and more people play in the original protocol China did not say you can inspect me. For what I'm doing on. Carbon. In the later they get. Yes. You can expect me about what so he balls so we gotta start. And that's what. Usually. Good. Starting. Off Ideally, this helps a process that ends up into international agreements into the actual negotiations eventually, a treaty ratification, and then a treaty can be modified updated Yep. So like the process itself that we're we're talking about here. So in terms of you know we've got a number of questions before we wrap about this. Topic specifically now that we've talked about a process and hopefully the further folks tuned in, it's useful to to think through how many ideas have to be pulled together to actually have a conversation about something. So complicated, you know if you think about who would have to be involved for something as important, artificial general intelligence obviously, you know representatives from different countries. Sure but. Deep questions they're doing folks from defense private sector and academia deed folks from different Geo regions within these countries a representative for cultural nuances. Do we need folks of a certain level of rank in the military as opposed to just some guy who happens to be the one region fiction to be interested in this stuff on not that I think it's IT'S A. Separated from reality, but for some folks it is. Deciding sort of need to be in the room is challenging and it might not be that in a first round, we can pull in the ideal players whose opinions we really think would matter in starting to concretize the conversation who might we start with with something like Agi you've already done some thinking on this. Remember that Berry beginning you want to know the date of the arts. Yup what are the norms principles values what does the state of the art on thinking about? Rules and audits or a What is the status of results from different international conferences because a lot of international conference meetings have occurred. Within that body as we go through that, you can pick out who knows the most in different areas. That's one part. Then, within also ask our nose sixty five, not around the world who ought to be invited. So. Most of the people who are invited or not centrally if we didn't right, you know like I might buy you and so. We might do sent but then our Iran note picks crew from Iran will be involved I won't know. Isreaeli Rayleigh no we'll pick who in Israel, and I will not know who should be and so so a lot of picking is done by Dr knows you're supposed to be a group of individuals and institutions cut across institutional categories government business academic. So, that's our brain picking mechanism of global local conversation. Got Us into the goal would be bought up. Well bottom up and top down. His both Okay. You have the nodes doing the grassroots ball, but we're also doing a global assessment like approach as you pointed out, we'll go. Pose. Okay. Got It. Got UNDERSTA. So I imagine there may have been previous were you've cycled a number of times on a particular topic? You mentioned three years in the future of work. I'm not exactly sure how that went I. Know You've you've done other sort of broad projects, governments, etc. I imagine sometimes I passing distillation of these ideas maybe don't have. Someone high enough in the People's Liberation Army as you'd like or someone is high up in buzz, you might originally like or someone is high abed up Microsoft or Google maybe you think would be relevant for. Agi Conversation. But if you build enough momentum, maybe the second time around, you can start to loop them in. You know when it comes to the critical stakeholders who who need enough by into start to get this stuff td mentioned moving from Kyoto to the Paris agreement the nudges along that line in terms of the political cloud in terms of the poll. Businesswise politically, etc does it often happen in kind of concentric circles? So to speak are or how do we work our way to influence? The people that we attract. Are Thought leaders in the sense because they want to continue being fought meters. So they WANNA find out we know something they don't know. Looking at the other student in school what are they now? So, there's a Lotta that going on, right. So result we attract a lot of thought leaders talked other people we don't know. They can't keep their mouth shut as how they keep the reputation by saying unique right. So a lot of the stuff gets rippled along we integrate the players as you point out as much as we can. But a lot of these people don't necessarily like to fill out a questionnaire. Footnote. What we've done in the past, we have our no chair interview such people. They won't fill out the questions themselves, but they'll answer. Yeah, yeah. Drank or so we do that. So we're. -nology but the thing is it permeates out I mean there's no. We don't we run for one study at the next. We don't always evaluate how it permeates out, but you see our phrases in different things inside retired in the head. And you can tell because we pick a year or a certain consequences for. US and you can tell, yes, see who's grabbing your stuff. Leaders light to know that they're the ones who know it all they want to say I want to do X.. Think Thank X. told me to do this. No, they don't WanNa do that. Yeah. It's it's interesting. There's a funny analogy in the market research world in center of our universe you know beginning the company was working with big retail banks in a pharmaceutical companies, etc and we were warned by early advisers who who were pretty high up forrester you WANNA. Steer clear in your early days of Lake. PWS's in the KPMG's in the big consultancies because their goal would be to take whatever your research methodology is, and then they've already got all the enterprise relationships in and they'll just say, hey, we came up with this look at what we found look at how many companies we assess and so yeah, similarly there's. Consequences in your world too. So it is with ideas I. Guess Right easy to swap I can see plenty of upsides there. But of course, there are some downsides back on Agi here you know clearly you and I are on the same page why this interviews happening that much more thought would need to be given to how this would shake interested in some of your instincts neither you nor I know the future, but we have some instincts instincts on. What it would take to cross the chasm where international bodies our nation's maybe let's say would believe that human solidarity around this issue is in central. To be of the belief that at some point, we will have to come to that conclusion if we don't want competing cyborgs and competing strong ai I think that that's that's really the state of nature in terms of plugging in stuff into back of our skulls and in terms of building machines more powerful than ourselves. The state of nature I wouldn't leave very much room for for Happy Little Apes. But what kind of tensions what kind of you know precedence would have to exist four folks to get on the same page and say, Hey, we have to come up with a way to handle this. What do you think it's GonNa, take. That's why we're best prepare the sun. That's why. The but to give us a side point to that might help the audience be a little more optimistic because there's reason to be quite pessimistic as is that in the early days of the Internet I was involved in Getting what's called packet switching instead device that makes the Internet cheap put these little things all around the world and it makes packages and switches to a satellite. The satellite time is almost nothing even though you've noticed computer all day long that's why it's cheap. Now, I was involved in the early eighties getting that into different third world countries and often dictatorships you business. And as a result, I was aware of the conversation. Gt talent and all the other early early players and we are so happy and optimistic. We're going to get a knowledge of the world together. We're going to make it work. I. Mean it was just great. It was like. I got a magic wand and running around the world hurting on all these deals, right I did not think once. Far As remember anything about the use of it for child pornography. For organized crime in wandering and all that Orleans for star, a cybercrimes information warfare as but all the rest of it I was. I would say as a species we were. Lately naive about it as we ran into it. Now. That's not the case. I, mean, there are I'm not trying to keep track of this stuff and I can't. There's so many conferences meetings on on. In artificial intelligence. I can't hardly keep progress over anything. So we are looking at the negative. So thank God for Hawking's and Gates and the rest of the. Traveling this Boston Hey, this can go the wrong way. We did not have any remember in the early days of the Internet your baby him some people say you're weren't allowed saying, Hey, we got to rethink how we do this Internet yes, we want to do it but we can't just let it go. Yeah. Yeah. Yeah. Her what you said that I remember which means that the in my view that there's reasons to be optimistic that we get more collaboration on this one then we did on the other. And there's more downside because going to the super as you point out. Yeah. I mean you know part of me thinks that many many years ago that there would have to be some kind of an I hate to use the analogy I. Don't have a better when I probably should think through a better one before I start saying podcast but some sort of Pearl Harbor ask scenario now that it has to be an attack by another country, but it has to be an event that. As horrible as it is the common enemy idea that I the the idea of intelligence in an in an unbound survey way or in a way that maybe we feel like his dangerous would be so evident as to be real not as to be imagined I think you know it wasn't even the coronavirus here. You know what? What was it? It was. No. One's really sure about what the CCPA's saying. No one's really sure like what's true. What's not whatever I think a lot of Americans might have some skepticism there although you know certainly plenty of sympathy for the folks that were ill and for the doctors who really had a bad golf but you know you start feeling it crawl across. Europe right he started seeing like the pictures coming in with the CCP's not gonNA show you all the carnage from from a poorly handled sort of. Outbreak but you know across your, we were seeing and I think for Americans it was like by Golly, we've really got to think about this. You could still say we acted slow but the ideas like it has to be visceral. Might have to visceral here. Might there have to be some some brain computer interface some Ai. SCENARIO OF SPOOKS PEOPLE. So thoroughly across the glow, they realize we gotta get on the damn same page we can. Develop. In under three scenarios, we did on the future work technology out to twenty fifty. We did have in their some of those, for example, the idea of weapon systems of developing and and almost developing its independence capability. I mean weapons systems, alliterate act independent of human control, and we have that in this scenario. So by the way, the to it. Because government sometimes a little slow sometimes our self defensive and offensive, and see you end up with your independent hackers in the world underground folks. They create a new independent group. That gets ahead of the thing and then fighting cooperates with governments but maintains our independence from governments on they wouldn't be arrested so far so they have. Intrigues on this stuff yes. We have a couple of those. Things are lost control, and in this scenario we ended up concluding in that that particular there was never fully under control. He was a constant thing that yes or somebody hold you know part of it whole areas wiped out the still happens. Yeah. There's some stuff in there. Referrals. You. Have to check out the future of work. It sounds like that's the closest precedent to too strong. I at least in terms of your. Matter of fact, you wonder we had the very end art isn't all of a sudden things are happening around the world. We couldn't figure out why and how is happening and we think of. The Subaru has begun. Yeah. Yeah. Well, I will make sure that we linked to whatever is publicly available in that work in the show notes somewhere. So people can get an idea of what the output to these kinds of processes are and were were at on time here rebound load them I mean twenty bucks is not a big deal. Easing off. So Cool Jerome I'm glad we're able to have you on in this series. You are one of the rare folks brought on because you know you don't know. But unlike most people in those in that position, you actually have a process to figure it out I. Think more of US should be considering it through that Lens very much appreciate your perspective on the series of thanks so much joining. US. Thank you. So that's all for this futures series episode in the business podcast the next three Saturdays are three final episodes in the series on Ai Governance. So we have three guests who are in my opinion, some of the most important thinkers about strong. And the considerations of governance and management and even safety of artificial intelligence that approaches or even surpasses human ability on multiple fronts. Those three guests are Steve on one hundred who is our next interview E. next Saturday. Ben Gerstle arguably one of the best known artificial general intelligence thinkers in the world founder of a number of AI companies and organizations. And lastly, Hugo de Garris one of the earliest thinkers in this space who wrote a paper I think two years after I was born nineteen eighty-nine about the future of artificial general intelligence in his thoughts have been really important me crafting my own opinion about where these technologies are headed. So we've had on some great folks we had Stuart Russell from Berkeley. We had folks in the future of Humanity Institute from the C. from the I Tripoli so many great perspectives or really going to start stretching that into what the heck does this mean for where we're going and where do we ultimately even WanNa go as business leaders. Government leaders is people thinking about the future of humanity intelligence itself so we've got. Some excitement in these next three episodes and I'm excited to bring them to you. So be sure to stay for next. Saturday for more of that and be sure to stay tuned for. Tuesday. When we dive right back into a I, use cases here on the I in business podcasts stay tuned and I look forward to catching you in the week at.
Building AI Products for the Enterprise - With Saurabh Suri of CerraCap
"This is Daniel Fidel. And you're listening to the AI and business podcast and this episode. We're GONNA be talking about building. Ai Products for the enterprise. Which for any of you who've been tuned in for a while now is not easy to be able to build an AI. Product that delivers a meaningful business result is one thing to be able to deploy that with existing. It systems is quite challenging to have to retrain staff to understand existing workflows. And that's going to be the focus of this week's episode some of the most exciting facets of our own research work here at emerge when it comes to our work with large enterprise clients is around understanding the build to buy ratios for different kinds of AI. Applications it's interesting to see if you look within let's say banking let's say life sciences. What kinds of applications may be drug? Development MAY BE PAYMENT FRAUD. Detection are being acquired primarily from outside vendor providers in which you're actually being built in house by the companies themselves sometimes. There are a application areas. Where really the company understands their own. Unique nuance needs at such a deeper level. That any vendor could that they have to build it in house so for this episode. Where the your vendor selling into big enterprises or whether you work within an enterprise and you're thinking about build by decisions you WanNa stay tuned all the way through our guest this week soared story is the CIO and managing partner at Cara Cap Ventures. He goes deep on what he's seen work and not work across his portfolio of investments. When it comes to their interface with the enterprise if you'd like to learn more about what we do here at emerge including how we balance and understand the build by ratios across different business categories of AI applications in major sectors. You can learn more at E. M. E. R. J. dot com slash. Hey I o l. That's a opportunity landscape. That's our core research service here to merge that's emerged dot com slash Ai. I O L and you can learn more there without further. You're gonNA roll right in. This absorbs Uri managing partner at Karak ventures here on the AI and business podcast. So I wanted to start us off by asking about what you consider to be. The biggest success factors when it comes to building. Ai PRODUCTS FOR THE ENTERPRISE. Adoption is obviously very hard. What happens when it's done right right and left? And that's probably one of the most oft ost questions are around in my role Especially on the venture capital side. We have a lot of conversations with people on both sides. Which is the enterprise leaders the? Cio's if you would Trying to adopt a into the enterprise. Dig It off. The startup. Seals are building these products for the enterprise. Right and looking at that when we look at the success factors. I can pretty much make it down into the first and most important one being domain domain domain now what does that mean the domain and understanding the nuances that. Come around with it right. Companies and startups. That can demonstrate a strong domain understanding and clearly articulate the problem. They're solving than to have the biggest success when building these products for the enterprise. There's a funny example. I often use. And that's the word slim right by Stephen. Somebody's bidding products around. Nlp Now the word Flam has a very different connotation in retail as it doesn't healthcare right in retail. You have a problem in healthcare. That's a symptom and does a slight over simplification. Of course by as you're building these products for the enterprise that domain on the understanding of the nuances that the domain is really one of the critical success factors. The second is looking at outcomes and not the CAC of the process right. There's a huge difference between a company saying that they would leverage to help with customer acquisition as as many state versus saying that. Hey Mr Cio. I will increase your overall customer acquisition by X. Percent and reduce your cost by white percent. And that's the difference of focusing on the outcomes right and the companies that do their research and build the products whether it's internally or buying something externally focused on that outcome really win right and the last and final thing I'll go on is the integration of the deployment which is products that are built with business users in mind right up seen always have the biggest adoption across the industry right unfortunately seem to many technologies amazing products. Completely fail because they were just too complicated to use so those two things if you come down to it it's don't mean outcome and east of us are the way we look at it. In terms of the biggest success factors. Got It okay. So let's poke into these themes these are great places to start. I think domain is a really interesting one. Because you know you can see so many startups that you know. They'll raise all the way up to twenty million bucks with you know a homepage that the solutions tab has fifteen things in it you know. Maybe that's a little bit of an exaggeration but will have eight different things in six different industries. You know what I mean. And then in order for those to raise fifty mil they normally boil that down to maybe two industries and then if they don't go out of business oftentimes it's only gonNA be one so it's weird that they have to start. So broad is just a consequence of this tech being nascent is that consequence of AI. Being hot and so even people without a great focus can sometimes raise money. Why we we see so much of that so there are a couple of reasons around that but the one that I look at in working with a lot of these startup. Ceos in these amazing entrepreneurs is I think sometimes they make the mistake of focusing on the most visible and not the most valuable use cases right. And it's a little bit of okay deserves by six most visible use cases across the industry. Let me just throw a lot of them on a wall and see what sticks and the often fall into that trap because as they train their engines as they put in the right kind of deep learning etc. Because they've gone so broad then never able to get to that point of outcome or the point of focus so that's one side of the coin. The other side is yes. You're absolutely right. Sometimes you have to look and sound way cooler by saying I can solve the world's problems by amazing algorithms and platforms than learning learning. Whatever it be. And it's a combination of those two and a lot of CEO's fall under that trap. But I loved what you said that when they move on into this series B. O. C. C. They actually get penalized for that. You get penalized for trying to be everything for everybody. As opposed to being that one thing that solves one problem completely yet to find find a domain like. You said you know you said three times. Is it like domain domain demand and I think those of us who want to see these products mature. I think really long to see more of that for you. Does that require a combination of beginning with the right? Founders and team members that domain or does it often start with just people who were good generalists who get the tech and then who throw themselves at a couple of spaces really find what sticks in other words. Do you see that steering of domain domain domain begin with founder ship in initial skill sets. Or do you see it with people. Developing laser focus after getting a team together. Or is it pretty even fifty fifty? It's pretty I'd say there's more of entrepreneurs who get the tech and then approach the domain. They're just a personal preference. I'd like to see the other way around. I mean we love working with entrepreneurs who been in a domain long enough who understand the nuances and then who wants to break that down. Well we love seeing that because they come in with a very different focus that come in with a very different approach as well however I have been fortunate enough to back some very successful entrepreneurs who came from outside the domain. That actually was a bit of an advantage. Because you know. They came with a clean slate and tried to break that apart but they focused on that one problem right. That's being the key. I don't really care what side you come from. I do care a little bit. I liked the newcomer the inside out. But that's just a personal preference of obscene at work. The other really well as well so it doesn't matter which way you approach it but as long as you can focus on really south that one problem and find out right problem. Don't find the most visible problem but find the most valuable problem right. I had this great entrepreneurs who said the best way to build a billion dollar companies. Find a hundred billion dollar problem and just a one percent marketer really great way to think about an approach. That is a wonderful thought experiment. That is just a wonderful experience. So it's tough to do that in the hurly-burly of trying to get a I to stick in the enterprise. But if you can I guess that's probably the low hanging fruit way to reach a billion dollar company if one could even say such a thing last little question on this sort of domain specific area here. Do you believe that the key to making this work is purely interacting with customers? A lot of you made a really important distinction. Not the most visible problem the most valuable problem. That's really hard to distinguish because of a problem isn't visible enough. Then you can't really use it as your value prop to sell because people don't even realize it's a problem. How do you find that middle ground between visible and valuable in this kind of rough and tumble startup world? Really the answer. The Dad is conversation constant conversation immersion if I might add so constantly talking to customers and immersing yourself into the domain and the problems will spark one of the most valuable use cases as opposed to the most visible right. And I've seen a lot of entrepreneurs do it in different ways but the emotion within the problem set of that domain set tends to produce the best results when I say in Marshon on a this amazing portfolio company which was using its using deep learning and a I in visual recognition and the food quality industry and The Entrepreneur Co actually went and sat and Firearms for over seventy eight months and went to quality. Testing of places. Did that research in marched us out in that almost sixty seven months before coming out and saying this is the problem and this is the problem I want assault one behold one year since she Said that this company is doesn't really have to do outbound. Sales anymore did getting inbound requests left Brighton Center because they immerse themselves on the fund. That one problem. Which is the most valuable problem to solve? Yeah so I like the shortcut are the short answer is conversation and you also used immersion. I guess there's no there's no way around the necessity of that kind of deep dive. And hopefully you come up with a treasure. They're you also mentioned talking about outcomes again really really really challenging. Because particularly with a I in some use cases it's really hard to measure for example. Enterprise search like to find where it fits into a specific workflow and say like we'll save. X. Amount of time on this like if there aren't metrics that already measure that that it becomes tough to maybe sell to metrics but you really emphasize the importance of outcomes. Talk a little bit about i. Guess how have that mindset from the start had through that lens from the get-go building a product right now again whether you're entrepreneurs trying to sell the product into an enterprise or you're an enterprise. Ceo Trying to build a new product if you're not hitting will be called the CEO's cross ridges revenue cost qualities state ability if you're not hitting that off revenue costs quantity and risk. If you're not hitting that what impact that you really have it right in terms of the outcome. Dutch need drives to that. If you can focus on one of or better still all four of discord parameters right. Revenue Cost Quality Risk and engineer your product to provide and focus on the outcome against it your chances of success and adoption and overreaching that you'd start as a company as a product that Cetera goes up exponentially I completely get the problem of obscure metrics are metrics that don't even exist but if you engineer viper in a way that you're hitting again you WANNA hit one of these four metrics with. Let's not overcomplicate right. If you're trying to hit one of these four metrics how'd you engineer the product to drive towards that even when you come about an enterprise search absolutely very obscure? How do you measure pot activity of search right? I've seen a few companies who've done a great job and they started with a specific problem in search. Which is and the price search for field sales folks? Gandhi provide the right data the right time to the customer while they're sitting in front of them to increase customer acquisition of close rate. Something of the sort that again. You're solving that problem. And those companies tend to have a much higher viability for success than more genetic one's behind. Yeah you know the the the shortcut that we see and I hate to call it that but in the enterprise search world has one random example we see that when we really have a hard time articulating. Roi In these really obscure use cases. Like you don't like in mortgage. Everybody's use to search sucking. They're just used to it. Nobody's like Oh my goodness the time it takes like everybody's used to it. It's not it's not really different and instead of selling to efficiency because they're not measuring the time now anyway and it's GonNa take a lot of time to integrate these systems and get them set up. It's really complicated. Instead what they do. Is THEY SHORTCUT IT? By selling to risk they kind of paint this this fearful fairy tale. They say they say imagine. If this could help you wonder if this little tiny bit of stuff that would help you with regulation you know. They don't say it like this. But here's what they mean. I don't have our numbers for you. I can't tell you how much money you're making. I can't tell you how much money saving you'll never measure it after the fact and you don't measure it now but you know what I'm GonNa tell you that you'll be able to find the right kind of things that will keep the regulators from slap in your wrist. And that's why you should give us a meeting and the we see that as like the you know. I hate to say it's like the cheesy answer because it works like we see it working but I see a lot of that. I don't know if you see people using these weird little shortcuts like that option near the fear base out right and it always works to get through the door but not always closed the deal one of the things that we use with a lot of our portfolio companies. Is We try and help them get that anchor? Client the accurate client. Is that first initial client that you know you sometimes do heavily. Discount the work for them but actually work with them to measure out numbers so mortgage industry is a good example. If I just say that right and you say yes people are used to doing it. But if I stopped focusing and get that anger blind and I stopped focusing on throughput right. How many mortgages can I help you process with the twelve people? You already have on board as an example right and if the answer is right now you're doing five thousand a month. I don't even know if that's the right ball. Who knows you're doing five thousand the if I can get that up to seven thousand or eight thousand and I can actually. While I'm helping you do that. I can engineer my product towards it as well and in return on providing you this highly discounted amount to use my product that Cetera et Cetera. That's where sometimes anchor plant becomes critical and sometimes even more critical than the revenue base. Clients that you get a little rock and early stage companies can do that. You can't do that as a later. Stage Company or a mid stage company. Because then you just have a whole different set of problems you have to deal with. So that's why it's critical to focus engineer. Not just your product but the company around that mindset the billet sustainment think. That's what everybody wants right. Everybody wants their pilots to turn into something so successful with such collaboration that they can actually track numbers. But how does this stuff really work out right? A lot of these pilots just flop flop so hard that you just break your teeth on the cement like flopper really really hard. Because we can't access the data we might make a little bit of money. But it's just so hard to get the integration done. And even if we can maybe they're not friendly enough with us to go ahead and do all this tracking and help us build a use case that we can brag about to the rest of the world but I guess in an ideal world. What you're saying is you would get that you get the good anchor client to sink in. Who really likes you? And they're happy to be able to kind of measure the success and then you can leverage actual Roi numbers instead of just a fear story or just an excitement story exactly and I. I hate to ban but now now you see the reason why I prefer entrepreneurs who come inside out than outside in because entrepreneurs will come from that domain already have existing relationships and the understanding to land the right kind of anchor ply. I don't want it to get out away from the entrepreneurs will go outside because some of them have done fabulous jobs right of landing that anchor client and building the right use case but rebounded for the inside out. Guys understand the domain. Have the relationship and know the right kind of anchor to start off with but more importantly you brought up a very good point even if I get nine car. I'm GONNA hit so many roadblocks so many issues after with maybe I can communicate or whatever else and that's where I've seen on some really really good seals do that when they get the anchor client. They literally packed their bags and sit either outside her inside the office building. Poc isn't done. Because that I could almost mean make-or-break stage company. Yeah Oh man I have so many more things we could fly into. But I'm wary of where we are. I'm going to end on a tiny question. I'll try to keep it brief. How would you translate some of these ideas that you've picked up in the startup world to the enterprise in other words you know? You talked about domain. You talked about focus on outcomes. Is there any little tidbits? You'd like maybe business leaders to listen to if they want to be able to breathe those concepts into a bigger business absolutely and it's really funny because a lot of these concepts translate over and I'll be honest with you a lot of these concepts. I learned the other around writing. Been within a the the enterprise world and coming out. And they're too when I'm trying to sell might internal stakeholders all of the Donald business leaders in our large I of move a have to focus on the outcomes right. The chief marketing officer. The chief financial officer is not going to understand that. I've got you know there's amazing deep learning capability and five guys being whatever half a million dollars a year in salary to but would they are going to understand. This is what I'm going to do. I'M GONNA solve for customer acquisition. I'M GONNA solve procurement etcetera. So focusing on that outcome for the business leaders is critical and not go down the TRAP FOR IAE SAKE. The fundamentals hold true and that's the called fundamental because they hold true whether you're a startup building for the enterprise world or an enterprise CIO billing for the company and having those measurable metrics against the success and managing stakeholders. That's the key translates across the board managing stakeholders. Will we try to do with this show sore of as we try to educate the stakeholder so that they can be more easily managed because unless you have expectations about how this stuff works and what all the challenges are you really can't be managed? You're going to think all the wrong things and so hopefully for those of you who are tuned in. We're all a little bit smarter by the end of this episode sore. I know where we are for time. Super respect your time so. Thank you for joining us here in Industry today Dan. Thank you very much. So that's all for this episode of the AI in Business. Podcast as many of you know a lot of my work. Here at emerge involves flying around the world and speaking and also meeting with and interviewing in addition to servicing our clients not that long ago about a month and a half back. I was in Paris for the CD'S LAUNCH OF THEIR AI policy. Observatory on our episode are bonus. Episode in just two days is with the deputy. Cto Of the White House for Technology Policy so lean. Parker is going to be our guest in two days for Special Bonus episode here on the business. Podcast if you're interested in AI. Governance were balancing regulation between innovation and understanding what that looks like at a national level. This is going to be a really interesting peek into the. Us's AI priorities if you are a vendor or you're a company that sells into the government for anything AI related. This'll also an interesting episode to stay tuned in Fourth. So we've got a special bonus episode for you which I recorded live at the OECD in which is going to be live in two days to be sure to stay tuned in for this Thursday. We've got an episode and next Tuesday. We're going to BOOT BACK INTO OUR INDUSTRY. Themes are we talking about the future of logistics and supply chain which given all this crazy hubbub of the corona virus is going to be. I think more relevant than ever and so without further ado. We'll wrap this up. I'll catch you next time on the. A and business podcast.
The 3 Phases of Making a Business Case for AI - with Scott Nowson, AI Lead at PwC Middle East
"This is Daniel Fidel and you're listening to a and business podcast. This is our Thursday episodes. You know what that means will be focusing on making the business case for a in the enterprise, which is making the business case, really mean well. It means a lot of things it means anything that will essentially get the C. Suite to say yes to a project. We need to be able to match the right. A application to the right business problem need to build a potentially understand the state of our data and our company to know if we can actually leverage a certain application and we need to make the. Case we need to be able to convey what is the short and the long term value of this particular project or initiative sometime? That's short term. Financial are alive. That's never the whole picture. We've also got a paint the picture for strategy so this a lot of moving parts to make a I make its way into the enterprise if you sell AI products or services, that's topic obviously to your heart, but if you're looking to buy or deploy AI. Ai In your company. You still have to understand the same skills and we want to shed more light on that process. We believe that the people who are the catalyst who genuinely understand how to make the business argument for a and had to pick the right. Projects are going to be the people who make a big dent in their enterprise in in their industry in general in terms of overhauling it with artificial intelligence. We interviewed this week Scott. Is the artificial intelligence lead for the Middle East for PW C. Scott as a PhD in P and Scott is an excellent interviewee I've got a great relationship with Scott. He was on one of our previous episodes. If you go to our soundcloud or I tuned to search for Scott Nelson, he was in a previous episode about AI adoption. He speaks this week about the three phases of making the business case. So, what does it look like actually? Break up this business case presentation to three different parts what? What do you need to put together to be able to give yourself the best chance of getting the C. Suite to say yes of getting a sign off getting that pilots started getting that deployment, actually rolling and Scott does a great job of breaking down I think that these three phases obviously be thought of in many different ways as probably a lot of ways to conceptualize it, but I think this is a useful lens to looks from so without further ado. This is Scott Nelson here on the. Business podcast. So Scott the next question here is around building digital maturity when it comes to adopting I. It's not just about plumbing in having work. It's also about getting our digital infrastructure right building our teams in a new way potentially organizing our our it structure and some different fashion that often feels like a hurdle feels like a barrier feels like a reason not to do I. How she enterprise leaders think about that. It's not clearly just a negative, but clearly it is a bit of a cost is a bit of a hurdle. You like to frame it. I frame it pretty much like that that it can be a hurdle. That understanding of Ai, and where sits on your journey, and that it is part of a journey is critical. I've actually heard. It discussed on on the podcast before by the likes of Ian Wilson. World the world invests all the time in digital transformation. That's been one of the key things that consultancies in. The Powell world right now, but that's about aligning systems. Generation data. And then that comes ai transformation in our this is this off? You've gotten all your data ducks in a row then we can look at. Because having a long term, AI strategy is key to make a success, but key to that is having your data strategy in place. It's really hard to avoid that. It's just going to be more successful. Bought I- Tampa? That saying this should not be limiting factor for getting started with. Is this is you need to be thinking long term and short term? It's perfectly appropriate to start with small in house proofs of concept. To use off the shelf, cognitive sevices from from a Microsoft or an IBM say so. They are definitely barriers on hurdles and. In are in this region in the middle. East we find a lot of clients who may be onto mature in the data strategy yet. PASSIONATE ABOUT GETTING A I. Like to guide them from both directions his. How you get started quickly his how you can get some immediate value bought here the steps you need to take to get their longtime. And vendor companies I'm just thinking about maybe the even the software side, not even a consultancy sort of like you folks you. They often have to be the ones to introduce the C. Suite to the fact that yes, we can help you prevent payment fraud. Yes, we can help you to whatever count the number of people at India retail store whatever the case may be but. You know this this. This are also going to have to be required to start working with a in. That often feels like this this extra. This extra stuff some of them I. I think tried to make an argument and I think there should be an argument made in. You probably have a better one here that some of that maturity is an upside when we start working with the fraud data, we're GONNA figure out a new way to train to groom to store this information, so it can be of greater value in the long term for our application, and also you guys have a better understanding and more harmonized data set to do other applications within kind of sell, the necessary maturity part as a benefit as opposed to just a hurdle. Is there an approach? There is their way to think about that. The, that is actually an approach that works, it's honest it's true there are things that you can do in the shaping that will essentially form a discovery phase for something else like I'll go in and get your data out full you and it might be painful, but although a lot more about what your data looks like, and can advise you on how you need to fix it to make this easier next time. I will lend things within your data. The perhaps okay, here's some will use cases. If you did X. Thin. Actually you can enable why and. I've built you, but it also foams part of the business case say we've gone in. We've scrape data together. And we've done X.. Imagine what we could do. If you went back and invested in these other capabilities upscale you maturity, we've saved thirty percent on some task, and that was with the cobbled together data. The we have if you go through these phases than you'll have an even stronger business case so. The business case of one foams the business case of another. If you will, so it does work like that. It's not just a huddled. There are benefits to starting with anything. Yet and this, this is actually this rings. A bell and I've got a question from one of our emerged plus readers. It's related to this which all all ask in a second, but we you're saying really resonates with a recent interview that we conducted with a fellow by the name of Babic founded sentient technologies out in the bay area, and he's now with a big consultancy, but he mentioned that through these initial projects are often going to be the catalysts for modernizing data infrastructure, finding new ways to work together with cross functional teams, educating executives on Ai that often that maturity isn't going to happen unless a project has been decided on I. I kinda brought up with him. It almost feels could be a downside if we have project a project be project, see kind of sparking off in different corners of the business, and we have kind of data maturity, anecdotally sparking off in different corners of the business. Maybe that's you know disjointed and potentially the wrong way of thinking about maybe a smart enterprise would be thinking about digital ai maturity at a broad level and have a constant template to refer to any time they do i. do you see scattershot being the only way or even maybe the best way, or do you think that some central planning from an enterprise would be? Ideal I wonder what your take us. I think the for that I think that hybrid approach is much better. SCATTERSHOT is great. If that's how you got that, we started with one team. We did work with them. And then it trickled through the maturity came about it perfectly valid but I do think it is much better to be thinking while you're starting these little things somewhere atop. Quote Unquote needs to know. This is going on needs to know. This is coming on then, so start out as on the sooner you do it. Because as I said that long term strategy is key. The sooner you do it, the less you have to. Fit. All those cobbled together projects back into your strategy, because then you potentially end up with fifty projects whose code down to line they don't have metrics, and by the time you get in with you'll, you'll ai platform. Those kind of redundant. You might have to again. You've been getting value from them so it is that balance of meeting somewhere in the middle at the bottom down that maturity spread buses top down trying to get a policy. Get a framework in place. You know that we very much strategy motivated and It is just much better for long term success. So it's it's better if we can get our internal data, scientists together our business leadership together and say look. We're GONNA start kicking off pilots. We're GONNA have sandbox projects. We're GONNA have early deployments. Have things that we're going to move based on shared priorities, what should we have is our norms for how? Modernize or harmonize data. which should we have as our norms for how we retain the learnings in these projects? which should we have as our north you know. Is that the kind of conversation that should be had before we start these little spark off projects in different corners are what should happen in that conversation? I think that's right. It doesn't have to necessarily have him before as I said, but it needs to happen soon. Because what you find is questions of Rep Likability, if your little small spot, what? Could. You do it again and yes, you could, but you're essentially going to start from scratch so those conversations about how will you support re-use? How will you ensure some standards? How will you support? The data scientists who are they're all installing different python libraries on that different machines in different versions. How will you have that consistency like sort of any software engineering library, but then with ai you also need to start looking at governance responsibility ethics. Data Quality, and that can happen in the pockets, but it's so much more of a strong argument restaurant business case if it's constantly coming down from the top and the data, scientists played big part in that the quality control as the strategy teams. You know the larger the enterprise, the most stakeholders you will have again in the mall time. It will take to to pull this together, but. Those are the conversations have to happen. What do I need data scientists to support my work to make things easier. What do you need is the rest of the business to make this effective and work for you? Yeah! Why am I the Golden? Ideal I. Think would be that businesses would have that conversation like you said maybe not ahead of time. Certainly early on I'm imagining myself is a smaller. Ai Services AI product vendor of some kind. You know maybe I'm an adviser are on cell technology in some way. I go into a business I'm sort of wondering. Should I push this pilot through and get this thing to work with whatever the initial budgets are, or should I try to do my best to get some kind of that higher level alignment around what this maturity Models GonNa. Look like longer term because maybe GonNa, let me have a longer project in bigger customer lifetime value with this client, but at the same time. Maybe it's GonNa. Take an extra eight months to talk to all the Darn Sea level people. People and get on the same page. It really feels like a bit of a catch twenty two. If I'm selling into these big companies WanNa, do right by them. I WanNa have a good clv with the client, a longer term value for me, but an deliver longer term value for them, but then again it's like I. Want to move the Damn needle and kind of get started. He Serum Middle Ground. Is there a way to think about that? As an outsider, selling into or working with big impress? You're right because it's a very. It's an interesting point that you can have different perspectives on so a former CEO, once told me would back in Australia that there's nothing worse you can do than to lie to a client and the so much hype and underperformance in. You. Know even have this concept of a winter. You know we have this ocean. That could just stop at any moment. And I think the last time. I you had me on. The focused I, said Hey. I is harden. That's okay. Yup is still true, and from my perspective from ABC Perspective View. You have to help cons understand this. With Peter BBC that honesty, the integrity is is cool to who we all people trust us with the critical parts of the business, but I'm also aware that in our role we. Have the ability to help the clients. Step back to slow down and say look. These. Are the things you need to get in place? I'm very aware that we can have a strategy. All who can come in the implementation team will just take a step back on white for year while the strategy team. Does that thing in the long term success? That's great. The context the Yoto him out. Though you know not, everyone is in that position. If you have this niche capability, you really do need to sell. You have a technology. I think that is a pot where advising on some of these aspects and you can't. You can't entirely shoot yourself in the foot by saying you're not ready for what we have. You not ready to what we have because I think you'd find. Most people aren't ready for what you're selling. So you then get that scattershot approach that we told you about. You can get a team so long as the team you'll coming in with understand this situation in in, you can help guide them. And by taking that approach that it doesn't have to just stop top down. Then? You can get that in and then maybe there is room for it, and you can get long term success with one team but I. DO agree that being opened that this would be so much better if you know absolutely what we have can work for you bought. We can see opportunities in the rest of the company. It's almost being there chilly. Though coach Your Business, you can have a great success with them, but you want them to be a success to you know we'll build this view This might apply cases, but we'll build this view. We'll help you take it to the rest. The business will help you help the rest of the business. Understand how they need to transform in order to you know how do I. Get some of that I mean that that's something we see very. Very much here, if someone's implementing something, someone will see it in. That wants some of that the ADS. Yeah, child does your business champion. How do you help them? Go on and become a champion within their organization. So you're it is a bouncing act, we a- In a good position that we have both those arms can take time with that yet, but I think other people have a role to play in that too I I don't agree with just the short term sell with a capability that may fail and may not lost. The middle ground of kind of being the internal adviser I think sort of different in some sense, but at the same point, it's a big opportunity. I think a lot of I know. A lot of our subscribers run services or product companies in their selling into the enterprise to be able to be the catalyst that actually gets that smart conversation to happen Scott that kind of centralized conversation that that allows a company to actually mature in in the big picture. That's actually a pretty big cool responsibility. If you're able to help a company that much I think, there's a big win. Win To be had their final note. I'll ought to be briefed on this one, but I want to squeeze it. It in Stephen S is one of our many emerged plus members here and the question is. Is there a market for digital maturity? Itself ease worded it a little bit differently. But you and I have talked about kind of upgrading your enterprise upgrading their capabilities to be able to adopt Ai Stephen saying that maybe not all consultancies have a I-. Chops are not all going to build algorithms, and not all gonNa write python. Is there a market for selling those AI prerequisites in terms of training teams to work together in terms of helping structured digital infrastructure in a new way, etc? Do you think there's a market purely for that or do? Those companies also have to quote unquote due the. One hundred percent that there is again we're in a privileged position Peter EC. We say strategic to execution. I work with the implementation team. My Team Legend has imitation, but we also have strategy. We have the business specialist and you could easily decouple those. We see a lot of projects where. their capabilities. One consultancy firm has done the as is assessment. One has done the then. Up and down the strategy design, another has come in to implement or technology vendor will do the implementation at the end there is absolutely a market for that for the education about a I for having those conversations in upscaling people in how to have a conversation to act as data advisor the as is benchmarking. There is a lot of work to be done in the strategy. I do think if companies are willing to take the time to invest in that I think it's part of the long term strategy to understand. Data maturity a lot in the data transformation that it's the first stage. Where's your data? And what does it look like right now? So there is a role to be played only doing that without necessarily being able to implement. If you can help, understand what's needed to implement. That's great. If you able to do a vendor assessments say like you with the client you put together. The are pay and you help evaluate the respondents, but now I don't think implementation. Capabilities or not? Having them rather are a barrier to work in that space. Cool Awesome, so Stephen. Hopefully, that's a satisfactory answer Scott I appreciate you throwing in that little bonus question at the end era did I wanted to make sure I got this went in before the interview was out, but anyhow I know that's all we have time. Thank you again so much being able to join us here on the podcast. Thank you, Dan. That's all for this episode of the A and business podcast. We did three use. Case episodes Monday Tuesday Wednesday this week. What did you think about that taught me a note on Lincoln Search Dan Fa Gela that G. G. E. L. L. A. on linked in pop me a Lincoln note or send me a request and let me know your thoughts. You like more volume more frequently something listen to every day, or are you just as fine with having two a week or even have a different preference? Let me know bought me a note on Lincoln I'm interested to know. This was a bit of an experiment for me and I. I really would love your genuine feedback for those of you who are actively selling AI products or services in other words. You pay your bills by making the business case. If you're not already in emerged plus subscriber, please do consider This is a resource we've put together for people who are the catalyst people who need to basically get the C. Suite to say yes, not with fancy sales tricks, but by finding the right AI applications. So this is our AI use case library, our Ai Whitepaper library where you can find Roi Information and deep dives into specific use cases as well as our full breakdown of Ai Best Practices so. For Measuring Roi our best articles on adoption and deployment, and these are resources exclusively for plus members, so if you're involved in a services, if you're a consultant, even management consultant working on a strategy checkout emerged, plus it's e. m. e. R. J. Dot, com slash P, one P, plus and then the number one you can learn more about emerged plus there otherwise just be sure to go to emerge dot. Dot Com sign up for the newsletter at the bare minimum. If you're not already subscribed as it stands, so who? That's it for this week? Four episodes hoof I was a lot of recording, but I. Hope you enjoyed it. I look forward to catching you next week. We're GONNA BE DIVING BACK into use cases next Tuesday for our usual Tuesday, use case episode here on the and business podcast.
Using Existing Edge Hardware for New AI Capabilities - with Roeland Nusselder of Plumerai
"This is Daniel Fidel in you're listening to the AI businesses podcast. The focus of this week's episode is going to be on leveraging artificial intelligence at the edge. Hug. We run leaving closer to the edge and what are the use cases get labelled in doing. So this topic that has been of interest for me in our work in heavy industries such as mining and transportation. Manufacturing, but also in retail, one of our largest market research projects last year as many of you are aware emerging, you can think about us like a Boutique Forrester. Gartner we focused on the ROI helping companies, pick high ROI projects and refine their AI strategy. That's most of our work here emerge and that's most of my advisory work with enterprise innovation leaders are one of our. Bigger clients from last year was retailer focusing on what their largest competitors. In this case, the Walmart's the targets of the world were doing with computer vision in store and I would have never guessed at the outset of this project but maybe a third of the written pages from this report had to do with the considerations of leveraging ai at the edge the particular issues with. Hardware and software the particular issues, even battery time battery life, and with these applications actually looked like there's a lot to think about when we're putting a out at the edge in different environments again, in heavy industry, this was patently obvious for me but in computer vision in retail also became patently obvious. We talk about a little bit of all those things in this particular interview. So this was a fun one for me because it touched on our where a lot of last year's focus was with one of our larger clients here in emerge our guest. This week is rolling. Cinder. Who is the CEO and Co founder of humor I focuses on leveraging tiny amount at the edge with existing hardware. So not coming up with new hardware running a I. On existing chips role in talks to us about what new kinds of use cases can be enabled existing hardware, and also what it takes to take a machine learning model that might often be run in the cloud somewhere on GPU's and translate that down and deliver some value whether it's detecting if a person is in the screen or detecting a product is on the shelf and. Doing that kind of processing on a much older off the shelves bit of hardware is it turns out. That's its own technical problem. We talk more about the use cases than we do the tactical considerations, but we do cover the technical at sort of a conceptual level for those of you need to think about, what might it take to get some of these cases actually done Within your business on Roland is one of many presenters at the hardware summit put on by Kazakh research. The summit is taking place from September twenty nine through October seventh. It's entirely virtual. We partnered with Kazakh research last year to promote this event when he was out in California and now obviously due to covert. The entire event is virtual September twenty ninth through October seventeenth you. WanNa learn more about the hardware summit and Kasaka research. WHO's the sponsor of this episode. You can simply go to Google type in a hardware summit. You can learn more about their event and grab yourself a ticket. If you're interested in these of themes without further ado, though we're gonNA roll into this Tuesday I use episode this Roland of Humor here on a and business podcast. So Roland. Glad to have you on the program. I. Know we're going to be talking about a at the edge. I. Think in order to have that conversation based on where your firm is focused we should talk about micro controllers and a tiny am l. This is for for you folks really really big opportunity for ai at the edge he maybe not up what we're talking about today. Yes sure thing very me of course. So tiny mel is machine learning or a I own really cheap low power hardware and then usual micro controllers. Some Mike controllers are very cheap low-power chips and they're literally everywhere are hundreds of billions of my controllers in a roads sets also why they can be cheap. But it's very challenging to deploy machine earning or to run machine learning or microcontrollers and its resulting but maybe it's good if I did a bit of a wide so important to run machine learning microcontrollers. So one way to do this, I mean if he thought, you could think that you can just sense the data back to the cloud depressing Darren very heavy and. Expensive jeep use bid. This is often not a very good idea. First of all, they're bench with limitations. So if you have a camera that's connected to Wifi network and they sent up the whole camera to the clouds, if you have multiple gummer's connected through same wife network that just doesn't work your your wife is down immediately than there are things like latency take. Time to send it to the cloud to person Darah Senate back there reliability issues, diner radiant, and it's done. You still want to make sure that your product works their privacy issues. You don't want to send feed data or audio data through the clouds and dinners energy consumption. That's actually a big issue because sending data to the cloud or even if use WIFI consumes a lot of energy and that's not good. especially not if you have a battery powered device Saddam L. Solstice by running machinery and work by running two workloads on the device itself the very cheap low power chip. But. The thing is it's very difficult to run machine learning on a chip and that's what our companies folks them. Got It. When you say chip in this case, you're you're talking about microcontrollers. Yeah. Exactly. Okay. Got It. Got Maybe. So you've walked through a couple of instances an. Familiar. With you know the edge as as sort of an idea and the intersection of of Iot nfl you focus on this space pretty ardently though maybe we can talk about some of the cases where hyping data to the cloud makes sense and some of the cases where a dozen year security he brought up bandwidth. There's a lot of these practical concerns. Can we tie this to you know potential business cases Hey dan here's an example where it completely makes sense we we gotta send this stuff up it might be at the edge, but we got to send it to the club and here's an example where we really should not be doing that Do you have any? We can talk about The main thing here is that you want to have devised a battery powered. That's an important issue because makes much cheaper much easier to install vices. So for example, if you have a small camera in a grocery shop to detect if a shelf is empty or not. You want to make the device battery-powered at getter with smoke, and you can do that if the machine learning workloads is running on a mic controller and it just sense a small signal to the cloud if the shelves empty or if the if the shelf is enough to empty and you always want to tasks on the edge, if that's bull doing reason, why would not want to do it on the edge is if you need so if the Model is so complex that requires a lot of energy to run and requires very expensive. Large ships skull jeep use said, that's why you want to do hoax over example for very complex and opie models of complex NLP tusks you you might want to do it in cloud, but if it is Bulbul, if you game run it locally, you generally want to do that because of bent with. Issues, because of latency issues, because of reliability issues because of privacy issues. So a good example of adults that you want to run locally is, for example, an H. system where on a heating air conditioning system where you have a small camera which idex if there are human in room, and then if there is a human into room, the heating air conditioning system automatically turn on or turn off. You don't want to send it whole feed your feet to the cloud it. It's not great for your free event with of your Wifi network. Another great example where we ecstasy for between is in retail. So you're starting to see devices that are battery powered that have a small camera and that have a small microcontroller controller, and it runs in a little deep learning model to detect if the shelves empty or not car, and if the shelf has empty than the signal to store manager that someone needs to fill up the shelf again. Yeah. Yup or for example, as small camera detects how many people are waiting in acute and at the stores manage store managers do more effectively locate their their stuff thus like debt or you can do gay seduction off shopper and detect. What kind of products show purser January interested in said at the shop nick and do better product placements those kind of thoughts. been if you're running ds on larger chips that are more energy consuming, you have to connect them today Tristan Nets, somehow that makes it much more expensive to install much more painful store owner. So if he can make this battery-powered, you can just clip it on a shelf or you can just Louis this evening for example, would you really like if you can make these device better half these kind of devices everywhere. Got It. Okay. So that that makes sense I see the battery power as kind of a key threshold. It's one of the factors here that you're talking about as when when might it be better to do it do the Processing on the edge as opposed to the cloud in the retail example, just to be clear and I might be on the right page. I might not. It seems like if we definitely WanNa know these aisles whether their stock or not and these aisles where the gaze of the customer is it may make sense to just straight up install those cameras permanently and then they would have a power source but I think what you're getting at is that well, that's a very painful adoption process ray I that would take a long time and it's more expensive also maybe if aisles are moved or if we want to take different a set of different angles. Angle a different spot and see if we have a better read on inventory a better read on customers I know cases or whatever. Then it's very pliable it's removable and we don't have to start running. You know extension cords all over the place. We can just be doing m. l. without the need for that. Exactly, cool for Amazon, go type of store, which has cameras everywhere in which is completely built up for this. It's not a thing bid for much smaller shop which doesn't have the resources to install these devices than it much better. If can just clip on the shelf and installed very quickly and changed positions, etc, and is not just about berry power devices for many products margins are extremely important so if you suddenly have to install. NVIDIA JETSON GPO in there for example, which can easily cost hundreds of dollars. That's not great if he can do during the very cheap chip dead makes products. Yeah. That Sunday mixed pretty much more tractor for customers. So it's also verbs exempt for sound detection or for simple audio dos. So you can say during the on though GonNa, thaw six, and again you the dude locally also for privacy issues and for for energy consumption issues. Got It. Okay. So that make send us a couple of examples was an h VAC system detecting if people are in the room as to whether or not we want to use power, keep the lights on again these kind of clip honorable tasks that relatively simple we're sort of we're not doing the most robust processing in the universe you know we're not. We're not taking a you know an image and doing some monumental processing task on it. It's kind of hey is there a person here or hey, where's this person's is focused and then just doing that processing right then and there in in terms of getting that information from the device itself and pouring it somewhere where we can make sense of it I'm imagining. Let's say I run a big grocery store. I would imagine that that information would certainly be streaming out of these various and sundry devices into somewhere central where I could get a general picture of all of this even though it's not hooked up to the electrical system would have you in. It's not being in the cloud I can kind of take the processing. It's been done and just pipe in those results into some kind of some kind of a dashboard. This is what I would imagine that you let me know. Yes. So the method data should be sent out the device. So not the video, for example, your to hold out your feet, just a Meta data. So it just says shelves empty yes or no much. Notice data sending out resumes a lot of Energy Yep and it will drain your battery notes. I'm yeah. That's what he wants to prevent. So you want to proceed locally. And just sent out the result just send out the shelves empty. Yes. No to more central and seek. Got It. Okay. I'm going to see if maybe we can touch on even if it's one or two other small use cases before we talk about how this is technically done, which is obviously you folks are working on you know I'm interested maybe painting A. Little. Bit of a mental picture for the listeners to one or two other examples who got the grocery store camera. We've got potentially a security camera you know. Is there a car in the parking lot? Is there somebody walking somewhere whatever that's that's one. We've got this H Vac. Thanks very simple kind of yes. No type processing going on where we're not like. Scanning and image, and then re factoring Elon Musk's face on this person's body unlike Trent we're not doing fancy things. We're just making a simple decision. What are other real instances where those simple? Yes. Knows with some computer vision or or or audio can be really really valuable. What are the ones you're excited about? So one thing that infer excited about is, for example, hence, gifts for. Detection. Guest recognition. So where simple camera and it decks movements off your hands so for example, that can detect if you swipe through right, there were swiped to the left. So this basically enables any display to become as intuitive touchscreen screen. You can bench to zoom with your fingers swab rights left's makes kroll movements, those kind of things. Back this still a bit too complex from Mike Controller we're working very hard making our models very small and very efficient. Do this on their cheap fair cheek chips. Yeah. Because I was going to say, is there a person in the screen or not? Seems Pretty seems pretty viable to me. Is there a car entering the parking lot or not seems pretty viable to me even when you said gays detection I kind of thought to myself, Oh jeepers that's that's a little bit more complex. That's maybe eight or ten I don't know how many orders of magnitude more complicated than is cereal in the slot or not it seems like to your. Point some of these task it a little bit more complicated. Maybe that's a nice transition into how we're getting this done. Obviously, these chips were not built for this task, but there's so many of them. They're being created every day in a million devices from washing machines to you know little cameras and microphones or whatever the case be, and now we're we're sort of bending artificial intelligence if. You will sort of pack into these little devices. What does that look like what? What's the? What's the technical process to make that happen? Yes, it is very challenging because we want to make sure it does this by controllers can do to most challenging most exciting posible tusks and to this use them buying arise neuro networks. Okay. So I'll try to make not technical but anita give some. Explanation. Sure. Generally, if you have deep learning model, you have dense with millions or hundreds of millions of parameters. Traditionally people used thirty two bits to encode each of those parameters, third bits, bits to make things faster to make things more efficient people with thirteen bits to sixteen bits, and now pretty much everyone is using eight bits, eight bits, each of those parameters, each way to need active deflation. BITs, when we do this and we thought how can make this even more efficient? How can we really get something which everything we show as Mike Controllers to too extreme? We thought well, why don't we use just one single bits to encode each way to Nietzsche Commission just used. For each parameter instead of eight bits. So dismayed your mobile, much smaller because instead of using a bits, you only need one bit and it also makes your model match faster. All operations become much faster but the spit Mara a bit too technical. Bits using by North very difficult, very challenging. been doing a lot of work to to make his work. So to be able to use by Rosner Network, she needed to get several things in place. The first thing that beinart networks require new training algorithms and we've been doing an are still doing a lot of research on how these models Saudis by neural networks can be trained because she still wants to make sure that these models are accurate. So the accuracy said. They don't miss for example, human human is actually welcome from scarborough and they still fest Ns Mall. So we've been doing research on net the second part that you need this need to develop the software to trainees by neurons neural networks. So you need to form does algorithms straining worth software, and we've built this over tools and part of this work sexually also open-source lark acute l.. A. R. Q.. and. The third component that you need is something we go compute engine. This is the piece of soap trained animal and executes its very efficiently on a micro controller. So it is basically like if you just have an APP, your smartphone. For example, the snapchat APP, you can use it if you don't have android or. Needs, an operating system and is compute engine is basically the operating system for deep learning controller and to build a fair efficiently. It's quite difficult is actually very challenging. So our team has done great work and his work very hard to make efficient and fast. So those are the three components deck need to build minutes what we've built on which were still improving. And actually we're also covering another layer of stack, which is an sir chip designed for reconfiguring chips sculpt FPJ's. Through technical. Yeah I. Don't think I should go for it here. Yeah. For a particular audience but the conceptual understanding important for our our folks regardless of the use cases the relative cost the relative applications for business value that's that's certainly relevant and one last thing that Kinda floats to mind role as you talk about this, maybe we can end on this point is sort of what's going to happen in this. Ecosystem of hardware and software at the edge. You know there's there's so much more that's going to be happening. There are some people as you're well aware who are trying to figure out what might be new kinds of hardware. We're GONNA WANNA have in whether it'd be self driving cars whether it be drones or whether it be any kind of you know hand held device a cell phones that will be. Better able to handle the kinds of. Machine learning oriented tasks we want a handle on the edge we don't WANNA have to pipe to the cloud there's other ecosystems that are really about kind of adapting to the existing landscape of hardware and saying okay well, how can we effectively take the cutting edge of what L. is able to do and bring into that into that world? Do you see over the course of the next decade just a bloom of expansion on both of those sides of the camp? What's your thought about the future here? He mean both on The hardware side. So precisely, yeah. I guess I guess people let me frame it a different way people who are trying to reinvent the wheel. Hey, look. If we're going to be at the edge, it's got to be these kinds of chips. This kind of processing trying to overhaul everything versus folks like yourselves at least at the present time where, Hey, there's a huge ecosystem of this existing hardware. Let's make cutting edge work there. Do you see as much explosions happening on both sides fence or maybe do you think about it differently? Yeah I think both sides are pushing heart and there are lots of different approaches both on the hardware side as on the software side, and I think is also necessary because, yeah, we really need much for example to do more exciting things on VICI low power battery powered devices need much cheaper chips that run much more efficient software I. think it's very important to both sides keep pushing very hard and try and ineffective thing. Yeah well, and it'll be interesting to see how you folks develop. Your Roland is obviously new new algorithms enormous for a fray I lord knows five years from now what will be the most popular computer vision approaches right I imagine you guys will be adopting adapting to all new technical ways of getting this stuff done as things move forward and we can't even predict all that stuff but it sounds like for you there have to be innovating on the hardware itself what the idealist and also people adapting the current hardware to get more stuff done that those are viable approaches. Young agree and you need to keep changing both sides. If right now is software that the companies in the research teams dead are designing better or more efficient. So far are purely looking at existing chips. You will not end up with the most efficient solution and the same thing with the hardware side if people are making chips for existing learning algorithms again you end up at the local Mexican. So for example, if you get transformer transformer are very efficient for for jeeps. For. Example G which is very popular. That's been designed to be very efficient for use and eight bits deep learning models that are very popular. Those have also been designed for jeep use those motives that are currently out there in the most efficient in best models that are out there the people. So if you design new deep learning algorithms, new models for new chips, you can end up with a efficient and very powerful solution. So both sides hardware to keep innovating. The future is going to be an exciting one and I think there's a lot of viability for being able to have some nimble battery powered solutions in the early days to be able to fill out these use cases and and deliver some value. So I'm certainly rooting for you guys and seeing how things go on and I. Know That's all we have for time rolling. Thank you so much for being able to join us here on the show today. Think. Very much forever me reading joint it. So. That's all for this episode of the A in business podcast. If you're interested in knowing more about the use cases that we cover here emerge, and if you'd like to be able to have a visual explorer of use cases across retail, which we talked about today across financial services, including insurance and banking across defense heavy industry, it'd be sure to check out emerge plus emerge pluses are premium subscription folks that really want to put an action whether you're a small consultant needs to guide your clients with the best smart out there and really understanding what their next step she'd be or whether you're an enterprise leader who really wants access to not only. Use cases that you can using your own business, but also best practices about measuring our y about adopting and deploying adult building a team successfully, if you want to save yourself the hassle of reinventing the wheel and learn from some of the best of guests that we've had here including heads of AI at public companies, they'd be sure to check out emerge plus you can learn more about the subscription e. m. e. R. J. dot com slash p. one that's plus once Em Yard J. dot com slash P. One, INC learn more about emerge plus that's for this episode. I'll catch you for Thursday's making the business case episode Huron the A and business podcast.
Where Conversational Interfaces Belong in Banking - with Shankar Narayanan of Active.Ai
"This is daniel fidel. And you're listening to the a in financial services. Podcast should be grateful that the hype wave around of thoughts started to die down at least in the financial services world. I remember two years ago. Essentially all of our enterprise advisory engagements with banks in particular involved in some way shape or form leading air out of the balloon of the expectations around chat bots. People thought that these things had monumental capabilities. Astronomically beyond with technology is today and subsequently you saw companies like wells fargo and like ally bank make a lot of noise about chat bots and then all of a sudden. Fold those projects entirely presumably. They're working on them in the background. But none of the big buzz ended up coming to life and we in fact kind of annoyed a couple of the chat bot vendors by letting the air out of that balloon but the fact of the matter is there are places where conversation lay i can add value. We have to be realistic about it. We've had some great guests on this show in the past. We've had consist sisto. Who's probably the company that's raised the most money for chat bots in financial services specifically and this week we have a company from the other side of the world from singapore in fact active dot. Ai has raised over fourteen million dollars. They have over seventy employees and they are working on conversational interfaces in banking we speak with their co founder. And coo shankar narayan about where conversational interfaces belong in banking. What are the places where they actually can get fruitful use. And what are the places where it's realistic to expect them to deliver value. We really shoot straight on this topic. And i think it's a really useful episode for that reason if you're interested in more natural language processing use cases including conversational interfaces than download our free. Pdf brief called unlocking the business value of nlp. You can find that at emc rj dot com slash nlp one. That's anna's in natural ellison language visa processing and then the number one emerge dot com slash. Nlp one and you can download that free pdf brief if you wanna take some of the lessons from this interview and go a step farther without further ado. We're going to hop into this. Episode with sean carr of active dot. Ai you're on the financial services. Podcast so shankar. I wanted to start off talking about what elements of workflows within banking where we can really apply conversational interfaces today. I think there's a lot of claims about ai taking over customer service or some other functions but of course it's more nuanced than app when you take a look at where your technologies being applied and what you see in the landscape. How would you summarize wear conversational interfaces fit in in banking right. So there's a lot of hype around conversational So i would like to break that particular meant we are on the very early stages of conditionally. I am in the technologies just evolving as long so in terms of in banking. I think the key use case for conditionally is of several but let me talk about the customer engagement side am banks are looking at cutting costs on call centers and reputation calls which comes into the call center they move into some form of a flow for chat bots and chad votes has to be intelligent enough to understand that alonso's and respond appropriately the challenge. Which we've been seeing and which most of the companies thunder companies are evolving from celebre. Give you an example. This has been restarted. This company was that everything's moved conversation and unstructured data. And we just happening where you have people chatting or come on what they can ask anything. Because there's no structured work or the zone many shropshire that they can ask anything. So you'll you'll heavy lifting is done by your systems in entirely to understand. The piece has to be good enough to understand the intent and appropriate the answer Their tools at one is banks have to be pretty strict in terms of how they respond just to make sure that the brand is kept so the way. If if it's an ai which is open to training or training without any human interface. It can this phone and get trained and If based on property may give a wrong response so if we need to have a better control on that and stock has to be built in that so what we are seeing or the bureau of let me give you an example right when we started in twenty seven twenty eight when we launched our first services with a bank the workload pretty structured the opportunity impact build a lot of variations on radiance fall the the intense again the stroke of the nlp to understand how pavilions for that to respond a car in the food has to happen is let me give an example if i make a query that hey there's my checkbook i applied for it guest today so you may have multiple variants which built in and the system understands what you intend hits and response to it. We launched. We had of art. Sixty thousand interactions per day mid some of the banks on viet launched in india. Where the there are twenty million customers and the operational team was overwhelmed. Because you can't keep having team billions so we have to build a deep learning mortar so that it auto trains and the billions auto bill so this my team both so there was a lot of learning which we act do as we each rated in canonisation layer journey the customer engagement side. That's the sign the law of other use cases which is emerging will the last few years especially in a machine comprehension whether market documents which banks have and. Let's assume that you are a relationship manager and you just want to know that on. How is the apple Gonna be doing tomorrow. And what does the cio report. Amancio information office of has created and the relationship manager doesn't have time to read through it so you have a reading the document which is being fed understanding the intense and comprehending it and you can quit any queries and it will not give up particular on servile pick relevant answers and showcase whether human gan understand it and pick up the knossos. It's such plus plus right. So i see that as a segment which we are working on with some max banks so using a for internal processes you have the rpm which is basically. That's a separate were to complete version. But in terms of con- additionally is fell focus on you. See a lot of use gives us or hr all the mundane tasks which people have to communicate with. A human is being moved onto box or workflow base os and that starts the shift which is happening. And we're seeing that. I have data which shows the in fact last month a one of the banks did six million interactions in a month. Or the because it's amazing but the final. Wally masur doing now just to clarify chocolate. This is six million internal interactions. You're talking about this. hr faculty here. No no no. No your customer writ large of a lincoln howard phasing customer actions retail banking iraq jumps rea-. Now that makes sense humans out there but just imagine a call center will not be able to have that kind of scalable volume now. There are lots of unique interactions which are happening which banks looking through. So i'll give you an example while the banks had to adam. Api just tell where the credit card is going to be delivered on which day just going to be delivered because they didn't have the use case but customer Asking that i applied for credit card. Where is it. I haven't received it so bank said okay. I don't want this to go to the call center. I want to based on customers asking these questions. Why don't i give a particular times time kind of thing where i can tell where the where the credit card is share not share so what is happening with conversation is if banks can leverage and i think banks are slowly understanding the scale of it there is a communication we just happening between a customer and the back and how smart the bank can be to understand this communication and and has the Better engage the customer and that's the value which we see on the corner. And that's what's going to happen in the next. Few years is still in the early stages yet where predominantly botched on function properly they only can do set functions already set queries which i always to the banks you. Why are you putting. Faq's in the bar. I would have a better such features on google. You know i get better response. But guess banks would have to has to create that data pool for estimating agent. The banks are realizing that. There's a lot more effort to be put in for a direct engagement. Because just imagine. I go to a taylor. Who's a knowledgeable teller. And i tell her i want a particular checks to be shoot. She knows exactly what to do right. She doesn't she will go check and come back and tell you the relevant answer so about to function in that manner intelligently it needs to have results connected in the back end to be able to provide the answers to the front so from accurate perspective. We focus a lot on soza requests. The mutsu actionable services. Quickies what we look at. Give us some examples of those assuncao. This is really where the meat and potatoes happens for. Me is because you're saying it's early days. We can't handle everything conversation right. We're where we're just not going to be able to just tackle everything. Customer service services. Too many permutations. You've decided a lane to pick. This is really cool. What are those. Can you give maybe three examples of what a service request is and how it works so today. We do hundred fifty. Plus you skates this in all our banks. So we we alive in about twenty banks most of our banks because the last skill banks are india in india for us and there are some concrete unions in the us who are using our products but india scale much much higher so we are doing use cases. From checking account balance enquiry credit card inquiry services credit card issuance services origination services. I would be able to buy a sell stocks. I can do. Fund transfer functionalities heyman functionalities. So we are covering the entire gamut. Oh yeah loans issuance of loans. Managing your loans. Kurdish scores as well as ablity to campaign management internally via coming about hundred fifty. Use my get. My guess is some of those are astronomically more developed than others and some of those are astronomically more bounded succinct and reliable than others. And that's almost certainly true. You talked about loans. Could you give us an example of a question or checking account. I saw Listen i come to a bar. And i'm basically saying that i'm make a payment. A fundraiser trouser Thing is my balance right now. When i am looking for that particular functionality the bank is similar to say that shankar has a pre approved loan of say. Ten thousand dollars With this particular interest rate heat bank is pushing me that in the conurbation. Hey sugar I know you've been getting salaries as your balance. We are also your pre approved for this long. Would you like to take this long right now. So that kind of interactivity at the time of customer engagement is what's happening today. Got it to being able to sort of. I mean in that case where we're actually talking about kind of for lack of a better term. A bit of a marketing use case i suppose of course loan loan origination. Select that in getting a call. You're basically communicating with the customer at the time of. Its an engagement tool. It's amazing engagement to let me give an example racial today. If you have an app apple structured you open up and you go button. click click. Click click stuck is a vase structure. I i may not be saying that in a chat i may i may because y me is that It needs to all the consumers have to understand that you can the limitation as well as where it can take so. Let's let me give you an example. I can tell that. Hey i'm going to the us. I need a foreign exchange for this. Can you some travel card. And i'm going being these long sentence which upset but to invention that now. What is the future of the is going especially on coalition air. I it needs to understand this intend. It needs to be able to say. Hey sean where you're gonna use. This is the industry's rate. This particular visa Is able to handle this for your travel. You should be used. This particular travel cards and here are some of the bad. You can get a money. Invest in union in the us on on diesel addresses so that would be the most convenient way the bank should we interacting with you know so that will happen in the future. It is not there yet a part of the functions. We are already doing it. Activists already doing that. Most of the dense via able to address. It is not the energy function you intense. You can capture yet but the bank needs to have the relevant the api. Yes and the data tools and convergence of a multiple systems into a data lake where we can pull data from. The reason is bank. Most banks are embarking or already have embarked on digital and they all really seriously looking at a singular a later lakes which can combine multiple products and so that in the future as a single window and customers engaging on that single window. This is a really important point that you drive home. In from interviewing consists though at all the other players at your mid jillian the other players in the space at least in the in the western world know. I'm aware of just how important this is. The fact that we can identify intent. You can give me a hundred thousand messages. And ninety six percent of them. I can put it into the right or almost the right category to the point where maybe i could prompt a response a them to the right person but being able to really provide a rich experiences about having it connected to all the contextual information and also you would need then to train a system to not just do what it's been doing in terms of replies but now you'd have to train system on leveraging that contextual data so you need to make it accessible then you need to learn iteration train on that new layer of richness and it feels like that's maybe the next evolution the next phase four the next step that banks are gonna have to go through his at a a safe way to put it chunk or would you put it. In a different web so the banks are building that base indolence of the conversation piece of iding. The technology. Willing to become contextual. So in fact. Our activity is leaders. Version of products deals with contractual interactions. As well so let me give you an example right so if you may if a customer as okay. What's my balance on visa card. I would like to save right now. And then. let's say the gives you the balance and say is this disease being old by you and you can pay on this day and then i just ask question next question i say. What would my mastercard so system understands that. He's asking for the balance on the mastercard and it's basically correlates with the previous cornerbacks and then starts the process so we build that in our in a stack where we have a better understanding of how Context together now context vivid important and we do to levels oh context. You can't do multiple multiple levels. You can but i think the gets get confused right at this point in because i can't do multiple layers and as a human we we tend we can meandered within different contexts combatants on a it is not. I won't say technically does not possible it is it is but it is. It has to be very nadal focused yet. It cannot be white folks like say. I would like to send some money to dan Fifty dollars and i would like to do the same functions in using my visa card. So the i mean. The machine will get confused and these educators right against it's we are also not building tools and cape For something which is not gonna happen today. That requirement of the bank is also very limited in terms of is understandable. Let's take an example of alexa It'd be launched alexa with checkbooks and other things and you know for a fact. The data the most of the usage on alexa are predominantly in song or basic functions transaction base functions on not happening on alexa. Right and not yet. I mean that's the did i saw thousand nineteen right. The predominantly moss flare if particular song or music music is is the biggest banking vetiver university. Of course yeah in but we lost it in. Yeah i would be have about five thousand people using it but it is it is it is. It is something which is going to catch up in another twelve to twenty four months. Beden service requests will happen. Maybe not transaction basic westwood solicited stock scaling up that i warned us. I warned that he's royd this piece for me. This data that would happen right. I think people have to get much more comfortable but it is happening right. The compositional method unstructured method of accessing inflammation is happening. It's not going to go away. it's just going to just steamroll. Become bigger and bigger. Bigger technology will catch which village-level technology will will will scale accordingly but the demand is going to be there and fundamentally the demand is being created by the huge young population coming into the financial services to win and they want everything instinct. I mean you are in the tiktok era. You would need to be done now. So when you have that kind of mindset of youngsters. The basically the heavy lifting has to be done by the financial services stupor why that particular function. I was just going to poke into that. Sean car where we are at time that i wanted to throw one kind of very related question. Exactly what you're saying. I think is a great point that we could. We could wrap up on. Because it pertains to part of why i brought you onto the program on you are referred to me by by ocoee. Who runs a different company out there in asia and you folks operate all around the world you operate in the western world operate in big indian market. You're talking about behavior of how we access banking. And i happen to know. India china california. You're gonna have very different norms about online shopping. Very different norms around. You know chat between friends very different norms about how people buy things or or interact with companies when you look at the east and west. Are there any big dynamics in terms of the use of chat or voice or or other things that you really see kind of the differentiating trends differentiating realities of the boots on the ground market. Because i think you have a very unique perspective. Your boy is the phenomenal difference. In i in the adoption of chat especially chad based on wise based technologies in asia. As you make substantial what you're seeing is at least a financial services and we have credit unions who are product. I think one of maggie's on box which runs in one of the unions And we just launched another one you lear. Seeing the throughput of the engagement is happening. People like it in the us and they are getting used to the box and the services what we were surprised support. The asia market is at the scale. At which the an option. Which like. I said it's Six million correction with one bank and people want drive and they'll very comfortable interacting with. Yeah very comfortable. They know the limitations and india's largest sorts of market. I believe four hundred is the largest. A lot of things by are so many folks out there. The market gigantic young. Dan china than india. Right into the thing. Good thing for accurate was in. There was a great launchpad for us Very karate we biller product. Got it to scale our ability to see sale. And as i was mentioning earlier twenty eighteen onwards start setting the base in the us and focus a lot on the credit unions and a long tail of the banging and that's very focused on in the us market. And we fundamentally seeing that the us market especially in the banking side is is a bit slow and catching up with a lot of the services because when we show our hardened sixty use cases customers in the customers in the us as while you can do this and you have gone live with us. This is amazing. So i think this much slower in terms of the take-up but yes they are moving towards that because that is a demand in the market does a need. It's interesting to me sean. Car let me know if you agree with this as we as we wrap up but this is your articulating something. That's really bringing something in mind. I spent a lot of time in bangalore going into the headquarters of all the unicorn. Ai companies out there the in mobis and make my trips and a lot of those other folks leveraging ai and exciting ways and i came out with an understanding that india in some regards. There's a lot of challenges in india. That's its own. It's its own topic but indian guards is going to kind of leap over much of the western world in terms of technology. Based because they're skipping a lot of these legacy systems it also feels like asia in general is leaping over the cultural gunk of how i interact with bank. Oh i go to a branch. Oh i make a phone call right. It's all these new folks. Just get introduced to banking and technology in india and it's all fresh. It seems like the culture can jump start as well. Does it feel the same to you. So there's two things right Interestingly the technology is coming from the west right over laurel yup see. The adoption is what's happening in asia. We are building utilizing a lot of the technology. Lot of the knowledge from this but the adoption is Is phenomenal because of i. Think the demographic rate though india's huge jump yes they do. Yes they are. The asia generally has it's the demographic uptake and predominantly to try. New things. Right really focused on five stolen and when we launched we launched with several banks and we have seen the adoptions we will also surprise in terms of the scale of the adoption and so we had to be on our toes building products evolving product according to you. And that's why. I mentioned we have like coloring fifty uses because the banks are demanding it. Max's i want these. I want this ten other products to be been when we launched our first bank the internal. Hr department is a very interesting discus. So we launched the bank retail banking service to the customers. Now all the eight to ten thousand in staff and the indulge staff says they usually call the call center to get information from the branches they said. Why don't wanna call the branches eleven. You'll give me into bark quit. You fully thousand three hundred thousand tough. So it's basically inflammation flow within the man. Craig people always looking. For instant information either is the soviet side of the On the sales guy who within the bank or the operations guy. All seeking information and In some form of service requests within that particular but a global tickle so there are a lot of the internal opportunities which are not customer facing a available to be to be realized utilizing this technology arts. Yes we are seeing that it will link that sort halford that oh. I'm just focusing on retail banking customer facing. But there's a lot of stuff which can be put onto the technology. It'll be interesting to see shocker. In the next five years whether it be your company or all the other players that are somewhat in this space east and west adoption trends differ and also internal versus external Which become the norm which become really fleshed out and robust. But as you've articulated here there's certainly a lot of opportunities. So i appreciate the extra detail on those points as we closed out. I just didn't leave without asking you that stuff. And i know we're up on time but chunk are thank you so much for being able to join us from the other side of the world here on. Ai and business podcast anchor muslim. So that is all for this episode. If you enjoyed this episode in the and financial services podcast and make sure to check out. Our other podcast. It's called the in business podcast. You can find it on apple podcast. You can find it on. Spotify soundcloud basically anywhere. Podcast can be found google podcasts. For sure again. It's called a in business. We have approximately ten times. The audience on the and business podcast. It's quite a bigger bundle of listeners. And also we cover a wider range of use cases financial services in the mix but we also talked about what's happening in life sciences. What's happening defense and military. So if you want your eyes open to more straight shooting insights on ai use cases and deployment advice and be sure to check out the i. Business podcast otherwise. I certainly appreciate listening to us here on your financial services show and we'll catch you next month for our next episode.
The Future of AI and Defense Analyst Workflows - with Michael Segala of SFL Scientific
"This is Daniel Fidel and you're listening to the business podcast. We cover a lot of industries here in our use case episodes every single Tuesday on the in business podcast from banking to life sciences, and beyond we occasionally like to touch on defense that's indeed focus today. Our guest is Dr Michael. Scholars. The CEO of SF L. Scientific a fast growing AI consultancy here in the Boston area they've gone from zero to. Something like forty or fifty folks on their team in the last five years and they've worked with some rather large customers in addition to the military, and they've been awarded the INVIDIA services partner of the year. The last two years running Michael speaks to us today about the workflow of a defense analyst someone who's poring over data at aiming to find anomalies the might help inform defense's objectives so the military is looking to. Figure out where terrorists are going or maybe looking for clues as to the paver of some ruler. In some faraway country, there's a lot of ways to be able to proxy that data streaming from various sources in the world in defense analysts are burdened with what is often rather monotonous work to put together insights to bring to bear to military leadership Michael talks to us about what it looks like to embed a. In that existing workflow in where it can actually add value. This is useful for essentially any industry working with oodles and oodles of data aiming to make sense of it in terms of reports and interpretation, but I think defense just gives us a pretty cool color I always like covering defense use cases. So without further ADO, we're GONNA fly into this episode. This is Michael Seagal SF L. scientific here on the business podcast. So Mike You folks have been doing a lot of work in federal. We last caught up maybe a year and a half ago a Lotta grows for you guys a lot of accolades with invidia. The government space has been a big place of growth talk to us about the workflow in social media and Kinda, the job of a defense analyst and how that's being done today will pivot into where a it. Yes. Sure. Thanks Dan, and of course, we're seeing tons of money being thrown into the federal space from an AI. Perspective. For lots good reason right in most of the the opportunities here is traditionally if you look across any of these large programmes from the army to the Navy to the Air Force to the NBA. You have hundreds to thousands of skilled individuals in manually looking through images or social media to basically find in assessories right? That's what they do. They sit there are trained they're very, very good at it, but the very programmatic right in the take what they're given as. Gospel and then they move it to the next level where they're supposed to then have somebody take action on that that action might be monitor more closely or eventually hey, there should be a military kind of required there. So the use case that we've been working on for about the past year or so is with some of the some of the departments within the army that sits obviously outside of the US. What they're trying to accomplish is to basically be. Global listeners of all information that's being distilled to them. So you can imagine on a given day you have millions of tweets in Lincoln Post and newspaper articles and things like that are coming out positive negative all sorts of different things across all sorts of different languages, right to a given region in as traditional analysts that sits for instance, in the army your job is to basically assess all these look for risks look for patterns and then basically pass it on. To the next person who would take action right. So that is a very traditional way that the problem has been solved, but it's it's hard right? Because people are subjective meaning what I think is a risk. The other person doesn't and it's not scalable right where now exploding in terms of the content that's out there. So now we have a problem with subjectivity in explosion of growth, right and that's where traditionally we've been in terms of analyzing this kind of information. Got It. So it sort of it hearkens to a mental image that I have I spoke with Mike Brown who heads up the what used to be called deacs. I think. It's the dia you now. Who Talks about the visor man back in the early days, of project, Maven or something where these folks just a little green visors looking at screens labeling stuff manually it just tremendously repetitive can be very drawn obviously, and it sounds like in the social media space. No, it's much. The same we're we're we're looking for stuff we're using judgment super repetitive in order to scale it. You just need more human beings sitting in in seats. In this particular case, I can imagine risk just for the audience, Mike we can't get into incredible detail here, but I'm trying to clarify the image for the listener. We might be looking for things that. Seem like Russell's in the breeze for terrorist activity. We might be listening for things that seem like Russell's in the breezes hints to what the government is up to. We might seem to it could be sort of I imagine maybe there's categories of risks were looking for here in red orange green or something like that These folks have those strata in front of them as they're looking at the social media says big. It is in a can't be right and it's not just risk in terms of is something militias happening. It could be risk in terms of do we see spikes in Kobe in certain populations or do we see anti propaganda wear? The government is saying, Hey, we have no spikes in Kobe but are people are saying we're all getting sick right so it's really looking across the spectrum of language to say, Hey, something just doesn't seem right right in that could be a lot of different things. Yeah that's incredible. It seems almost overwhelming Mike I mean because over well, when when you talk about you know training system abounded reality is That's what we like. My good man. That's what we like my brother but you're talking about will risk it could it could mean These things could mean references of these things we're. We're talking about an infinite spectrum are the the folks who are trained I. Imagine they're trained for maybe a core set of of maybe main risks, but it sounds like they also have to be able to be flagging and be aware of all these tertiary could bees tertiary anomalies at the same time in it's me it's the ladder in most cases. Ideally we WANNA say, Hey, just have your eyes open for this, but things happen to quickly right things unfold as. A great example, right one day you heard a corona and you thought of beer the next day you hear corona and you're thinking about this is an actual medical problem. This is actually a problem. Now kind of locally with some of the Pharma companies. But like terms change technology changes the way that we think about vocabulary changes. So you can't just have a rigid definition like you do in traditional images with me where a building building building. Need to start distilling different languages. English Spanish but Spangler in. This phenomenally anomaly complex space that these analysts have to deal with, which is the whole goal of helping them from An. Tool right which obviously vote we can talk about now. Yeah. Yeah. We'll spend into so and just just for clarity's sake. So the way we frame up use cases and it will pivot right into where ai fits into the workflows talk about what is the business value at hand. So we're talking about what the what these people are doing I imagine the goal is we're potentially creating Schwartz where potentially updating some. Colonels generals with reports on topics of their interest or maybe even just notifying somebody when something really spooky seems seems like it's GonNa Happen. So I I feel like these analysts, their output is what might your at summit up like while they're doing this they can just entering stuff into some big net database that pumps out a report or they often doing the right themselves. Yet. So let me give you one more small little intermediate piece of knowledge or so what we're trying to create for them is basically a google search functionality just imagine you throw a Google right now and you can type into a window and you can ask a question what are my risks today? That's an absurd way. You could say something like that the goal is to enable them to get back information and then write reports about what they're finding in that information. Okay. Right and then serve that to their leaders whoever serving right to actually let them make that informed decision as well. Cool. Okay. These are the people that are looking for the INFO as well as creating the reports that are gonNA get settled up to the. Top. Okay. Great. So yeah, we talk about where I could fit into that workflow already you know my mind is dancing with all the places NLP could fit into the mix and and other things like that. But for you folks, you've figured out maybe what their problems are where a I could could witless way in what's that immigration looking like? Where were those junctures? Where is able to make its way in? So the first areas, the obvious one, right I think on a given thirty day window because we're basically looking across thirty days of legacy data right in data is literally hundreds of millions of records to automatically ingest these records into right. We're using these modern day tools, called Bert models, or elmo models, right all these fancy little names for these deep learning models that make them sound simple. But basically, saying can automatically ingest all this information in an start understanding those patterns such that those patterns can just be shown. To a user who is looking to get some better level of understanding, right so the first obvious places to saying if I may use her in I want to search wear is my risk I want a computer to be able to understand what risk means the context around where risk materializes itself, and then basically give me like a Google search of the top ten areas where you need to investigate further. Maybe it's this tweet maybe it's this document maybe it's this user writing give you that almost like search base functionality. Yeah does that make it does it does. So I'm trying to imagine an example again we're I'm using extrapolated examples of course because am not bob in this project and be you know it's pretty sensitive stuff but I'm imagining okay. Let's just say these patterns that you're referring to. Maybe we have some that relate to the spread of a disease could be cova could be something else maybe there's entities were sort of tracking here. Their sentiment were sort of tracking here would the display simply be hey, here are terms phrases topics that seemed to be exceedingly repetitive over the last let's say trailing thirty days trailing twenty four hours etcetera is it something akin to that? You Know Hey, here's ended. Okay. Yes. Because from an analyst perspective, you have to realize you can make complex as you want, but the end of the day you have. Not, an unsophisticated but in untrained individual in a I consuming those results yes. Yes. Yes and most importantly you have somebody who's not going to sit around for twenty hours waiting for your model to return a result. So you need to build something that inferences at an SLA that they care about in produces a simple visualization around. Here's the top keywords here. Are Some topics right that are meaningful to them that they don't need to think about the math behind it such that they can almost take that probability score as their net new Gospel he and then on to the next level right so it has to be simple individualization but complex in bill, right? Yeah and that's that's the challenge with I in general, right? Writ Large. That's that's the issue that we're gonNA run into just thinking out loud here taking that New Gospel I mean that's that's a lot of weight on your shoulders Mr Mike I. that's that's a big deal. Right? Because these folks obviously they're gonNA, use other tools of course but sounds like this is going to be another tool that will be a layer that maybe they'll filter where tensions go. Hey, it's eight. AM sitting in front of my computer again do I just start reading stuff or do I maybe poke into the things that are beet red and if change in the last twenty, four hours well, why don't I start there? It sounds like it's more than efficient use of scanning time tool maybe more. So than a definitive defining of what the risks are tool, it has to be right in there is in adoption curve. Just, familiarity with the tool in the output in is just not in this use case, this is literally every day. Everyone. It is. I've done a process the same way for twenty years in you're gonNa tell me this computer's GonNa give you the tackling. You don't do that overnight love you have to gain their confidence. So you do that by maybe spending the first several weeks running side by side and parallel in showing them. Hey, I'm giving you every time the best results or almost the best results in getting them more and more and more confidence such that they relied more broad system. It's always a decision support tool. Yeah. So this is this brings us into topics that we really liked to to drum home here at emerge. One factor that that I'm seeing a lot of in this kind of covid era is that tools that are that we believe I think arc as gonNA. Just, truck in the next two years I think you I've path is just gonNA I don't Lord knows how much they're gonNA be worth it to your but I think efficiencies everybody wants efficiencies ai for some weird reason often gets couched as efficiencies only which are so limiting in a terrible way to frame it but deep really hard integrations of either involve a lot of data sources and overhauling workflows I think just haven't even lower chance of getting adopted when our budgets are down were sketchy about the economy. Everybody's already fearful I think what we have to be able to do, and it sounds like this what you're talking about. Address another podcast is find a place in the workflow to fit it in their not doing data science. They're not really getting too crazy. You probably need in the beginning. Mike to bring on some of these guys help with engineering, the features, coaxing out what you want to pay attention to you probably need to partner with a cluster of them for a while but day to day on the dash were they just they're able to look at it it's not really changing what they're doing. Do you put this into dash for their big at a new screen they have to have open? How does this fit into flow of their attention? So. It absolutely depends on how they're going to adopt it. So I think most people do it the wrong way and I'm going to talk about the way that we do it. You can say it's right or wrong. It's a few I think it's the right way. So most people when they do real integration of data science, they I m L., whatever you WANNA call it. They start from the fun side, the Algorithm side, the training I'm going to get the greatest biggest NLP model in the world going to train it with a billion parameters in it's going to be ninety, nine, point, nine percent accurate. But in reality in a production environment, there's no way. Anybody will ever consume that it's too slow. You don't have the hardware you don't have the data to support that in an analyst isn't GonNa wait for it to predict or you have a Visualization. So to your point, you always have to start with one of those fundamental business requirements from an influencing point of view from a visualization point of view from the business side how are you going to actually get our ally of at solve that problem I with the assumption that the model works you don't need to prove it just assume the model works. So starting there and saying, do this needs to be in the same visualization dashboard or did something different because that can have profound differences of what you do or don't bill maybe you can have a different technology stack or Or you don't in your limited, right. So starting from that end user requirements has to infer what kind of data science modeling that you can do and then work your way to that side last right and then actually do the monitoring. So everybody is uniquely different depending on what they need to accomplish. Yeah. But it sounds like gonNA maybe it's vetted in existing dashboard ex baby has to be its own interface in some way shape ambler. But yeah, just depends on how it's skin fitting. Obviously, you guys you deal with custom stuff per so okay that's useful context. I think here we'll. Talk about that Nitty Gritty of getting past as adoption barriers I know you folks off also work in care. I gotTa tell you. My Am I say this with all due respect just any sector I was not gonNA sell a high into if I was doing the technical work, luckily, we just do market research here. Right but hosing the tackle, it would be healthcare because of just how many hurdles there are like the CEO loves it and it's going to benefit the patient but the doctor has to learn it and the nurses workflow changes a lot this stakeholder mix there is just I mean it's scary. INDEFENS- of courses also complex diesel. If we talk about what it takes to sort of, you know get folks to to start to use this. You know we went in with great intentions. The folks had signed off on this new that we were good at this. They believed in the vision they thought this would be really really helpful. We want to get people to Kinda get some traction with it. What is that convincing process for lack of better terms that internal traction process talk about a bit of that Mike The beauty about healthcare is if you're a researcher or a doctor, your inherently a scientist who is open to collaboration in ideas, right? Okay. That makes it easy that that's a good starting point. When we get involved in healthcare in this could be in research hospitals or big Pharma companies or something like that. We have to embed ourselves into their SME. So, for instance, when we're working with the hospital system and they WANNA do say medical imaging on. Radiological imaging for images. It's something like that. You don't just start building. What we've done in the past to help them bridge. This barrier is literally go to the rounds in the morning sit with them while they're discussing patient cases in literally talking and understanding what are they doing when they're looking at these images? How much time does it really take for them in how much of it? Is that to really then them talk to their patients and helping them build that story of saying like if this was solvable in if we can predict conditions in Sepsis in relapses in, give you a better patient in. Dr. Experience would that be of interest? It's Oh. Yeah. That would be great and then you start teaching them. Okay. So we've broken down that you. You think the process works from a process perspective and I understand what you're doing is literally sitting in looking at your rounds with you, which is Harvey Experience Beer that. Terrible. Well, it's not terrible, but it's a reality check Then you start needing to educate them right like what does it mean to build an algorithm? What is a probabilistic score mean? How is this going to help you in your day to day in? Really Treating them as you know, they're brilliant in their discipline. You have to go in there and show that you're brilliant and yours in coming together the common understanding and that really breaks down the barriers. If you're just GONNA go in there and say, Hey, listen I can predict cancer better than you can laugh you out you'll never have a chance but you really have to develop A peer to peer relationship. In. It sounds as though Yep so this is this is sort of even in suggesting the project itself wrapped as our last little bundled question here as we close out this first. Sitting down for figuring out what their day to day problems or maybe suggesting having it be their idea a little bit of inception action. Yeah to. Try. It has to be you know. I think about. Philanthropic efforts right if you go, it's the same thing you go into Africa I can get you guys water. It's like aw do be careful you you have to you have to sort of find a way to and so in your case it's the same ball game. You go there figure out what they care about where they're bumping up against come. From that place where they're not gonNA listen till. They know they're understood right. So you make sure that they're understood. Then you can say, Hey, we can have this in this way so that it would make this easier without the useful in. It's it's sort of like a almost certainly s if you could do it kind of thing, and that's how you. Have to suggest it. Then you have the issue you brought up with defense, and we can talk about whatever industry dynamics fascinating maybe defense specifically around what it takes for them to start to use it. Once you've built up and built. Now now it's available. You said running this fifty fifty cast deserve to be a framework of thinking for you as. The vendor to say, okay, we always know we build something. We're GONNA run a fifty fifty tests. We're GONNA talk to the economic buyer. We're going to tell them. That's what we're GONNA do when it's done built because we know this is not just going to be a rolled out thing. There's always gonNA be a little bit of a wrestling match. Here's. Part of what you're planning forward is that kind of Oh my God. Yes especially in. So the biggest issue with medicine even though they're brilliant is they don't like to think they're ever wrong but statistically, doctors are wrong in his own at the time. Right? It's just a fact yet. So as part of this journey has to be helping them see in designing a system that allows them to understand what we're predicting what they're actually predicting in showcase that these things could leave cohesively in. You're not replacing them. You're augmenting in making their jobs more efficient, right? So you can't just say. You're ninety three percent accurate historically because that's what already law radiologists are are machines ninety five percent accurate were better than you. You can't do that. It doesn't work that way. So you have to build this workflow same thing right? A Ui that's friendly that shows them in explains rate it's a heat map it's local it's something that gets them comfortable with Y, you're making your decisions and then that's how you get through adoption. Yeah. Yeah. So again, a lot of human factors here you guys are in. Your the Services Industry Mike and yeah got Yorkie man you gotta learn your soft skills my good sir clearly, you've learned a lot of so interesting I. Think maybe this is a good take home message as we close out for the folks that are working on these kinds of solutions is to think about how do we inset and really collaborate on the origin of the idea and the solution and sort of the way it's going to be frame, and then how do we do the same thing with very soft nice framing around getting them to try it getting them to adopt it making sure it's not an automation risk in an insult to their intelligence. People WanNA pretend it's about the Algorithm Mike but I guess it's. Not go I would say I, mean at this point we worked on almost every use case in every industry right? Realistically there solvable given the enough data in technology in hardware we can solve every problem in the world that's easy. The people is the hard part. The hard part, and so a little bit of bloviating with the claim there. But I will say you're driving home a point that everybody listening in does need to tune into. If you listen to the show for long enough, you're well aware that that's the case Mike I'm really glad we got to dive into that aspect of use cases as well. Today I know that's all we had for this first interview but thanks so much for joining us. Awesome thank you Dan. So, that's all for this episode of the A, and business podcast. If you like what you're hearing here, if you enjoy these use case episodes, if you enjoy are making the business case episodes on Thursday where we talk about a deployment and return on investment than drop review on I tunes, it's now called podcast, but it's very easy to find us just a I and business on Apple podcasts. Your feedback is not only tremendously valuable. To me and my team helps us inform who we wanna let on the show. What kind of topics we want to cover city can make things better and better for you. In fact, this twice a week format that we're doing is actually based on reviews feedback and Lincoln notes from those of you who are loyal listener. So I want to say big thanks for that. Already that's GonNa keep us informed moving forward to what you'd like. To know also, it helps get the word out about the podcast itself every now, and again, I'll share one of the nice five star reviews me on apple podcast. Let other folks know what what people really see value in the program and that really helps us a bunch as well. So helps me help you if you enjoy the program, drop us a five-star view, it's a business podcast on apple podcast if you haven't already Checked out our other show the AI in financial services podcast check that out on apple podcasts were on spotify soundcloud or your favorite platform as well, and be sure to get all of our latest coverage on financial services banking wealth management insurance, etc. We have the whole show dedicated to that as well. So that's all for this episode of in a catchy here for our Thursday making the business case episode on the A and business podcast.
The Power and Promise of AI
"Welcome to CCC talks empowering it and business professionals in their digital transformation journey. Find all the latest tips tricks and strategies at our blog and resource center at cloud credential dot Org and our host. Ccc Managing Director Marco Laughlin. Hello everybody and welcome to another edition of CCC talks with Markle up on the cloud credential council. Now today we're joined by Daniel for Gela is the founder and head of research at emerged. I'm Don thank you very much for joining us on. Today's podcast. Glad to be here market so much on your company. Emerge is focused on research in the field of artificial intelligence So maybe to tell us a little bit about what you do on what emerges all about Suri. Yeah I'll give you the very fast version here so our work focuses on really the Aurelie of AI in major sectors so we track the startup ecosystem. We try to know use cases and we categorize what is enabling within those industries. And sort of where? We're seeing a return on investment so Visual Map irs are alive for leaders of companies that a Serb don't want to allocate their funds without knowing the landscape. I I got a good proposition. Especially the visualization of an Aurelie. Also the new thing but I think companies should do more of that abbey rather than just pure numbers of it may be looking at it from. I'm sure looking at it from different aspects for sure. Yeah I mean For some people ease of deployment for example is a paramount import. Because they don't have that much internal data science talent for example for other people. They really have an emphasis on building a particular capability. Like let's say computer vision. They WANNA be able to identify things with images. It's an e commerce business or retail business and they might particularly want to screen for that so criteria will vary but I think companies do it. Well you assess where they WANNA put funds companies. That don't just you know. Waste a little money on a pilot here on a pilot there and then they'll they'll learn the hard way that they should have some some strategy so that's kind of where we had good. I've learned to light as a hard way to do things at an easy way to do things find the people that can help you do the easy way and the answer. But I think what you're saying is your focus on the value from AOL. Ai Tune Organization and trying to understand that before doing ai for the sake of doing it. Is that what you're about? Yeah so some companies have been honest with us About checking the box in other words. Oh we did a pilot project because we you know. Our competitors did press releases about pilot projects and we wanted to do one as well. I'm there's obviously a lot of danger there. Germs of of wasting money so on most leaders we work with a lot of heads innovation heads of strategy. They don't quite know what they don't know they don't know the totality of what they could invest in and sort of. Where is the low hanging fruit and a lot of time for Big Orix? They want wins. They want relatively quick wins on with. It's not always easy. But they have to see the landscape to do it so yeah better than better than hurling money into random directions or like you said. I'm doing it for the sake of doing a savior's now we're going to start with him. Simple question very simple in terms. I'm GONNA ask you. What is artificial intelligence will? What is a yeah so you had said before we started recording here? Mark that You'll get ten different answers when you ask ten times and I speak fulltime with a researchers and heads of AI at big companies. I probably get different answers from them as well so to some degree even the folks that know. The space are bickering. Lot about Specifically where AI fits in compared to machine learning etc. Broadly speaking a I is a computer doing something that otherwise people would do. This is the big umbrella artificial intelligence that goes all the way to really complicated natural language generation stuff for You know that kind of models a human conversation which were kind of borderline cutting edge big companies like Google all the way down to relatively boring stuff like detecting fraud for a credit card company or a potential even more limited applications. They're so big umbrella. It's a big umbrella they can. I think automating still for the computer that humans use to do. We'll talk a little bit more. Besse AMBASSAD industries a fan. You know as you said even if it's just credit card fraud or insurance fraud and I could have taken a lot of people hours to try and figure out. Where may I system might do that? Quicker faster better can ask 'em again in broad terms. Why should we be excited about? Ai Buffoon from two views one from the organization perspective on one from a consumer perspective. I'd be excited. I think that So from an organization perspective I think Excitement can come from. You know the potential to win in the market. I think there's excitement. There's fear there's a lot of motives for organizations it's just about being able to stay relevant so for particular for very large companies on have already budgets Who are kind of enterprise level The winners and losers of the next ten years and a lot of space is heavy industry financial services etc have to involve some capabilities unlocked in some ability to have moved fluently with these new capabilities. They emerged so being excited about that. Means being cited about staying ahead of the market as a consumer and ultimately again. We're more on the business side for our work but as a consumer. I think we almost might see it as the next level of the next layer of convenience like the Internet was know I I used to. I don't have to go to blockbuster. Renna video now have net flex is kind of an Internet serve convenience level and kind of layer on top of any given service whether it be food or entertainment or connecting with friends. Or what have you? I think I will be potentially another layer of that certainly in consumer tech. That's moving quickly you know. Syria was out well before those kinds of applications were available to be and I think that consumers actually going to be the fast moving area. But it's mostly it's convenience on for for your average kind of person on the street. Absolutely on he said interesting thing though blockbuster being the the US DVD video rental store which along exists. So I think I think we're saying the likes of these digital technologies a helping organizations change our business model the way they do business and how they do business because it allows people to consume things differently as well so I think it's a great opportunity for organizations to do something new but there's also the danger that is they don't do something new that some competitors might do that so it's not a challenge facing. Yeah it is. I think it's most relevant for the larger organization so in a space like banking where we do a lot of work insurance wealth needed. It's kind of like global top one hundred companies so if you are a mid size random bank in the Mid West. You sort of. Don't have to direly stay ahead of ai because you don't have the budgets you don't have the RND and the technology so nascent it's GonNa be tough to get it off the ground with a lot of that that are indeed but for the folks who are in the top one hundred when JP Morgan and Wells Fargo and the other players Start to get that edge in customer experience. Start to get that Agean in You know how easily they can onboard people or how well they can calibrate their loans and lending for example you're competing with the biggest right now essentially across sectors on matters survivability wise people realize that added and that's a lot of impetus to get started is actually less the excitement more the nervousness nervous. I think that goes back to your opening. You know what you're saying about understanding the Ohio. I'm the value from a now. Everybody is going to get value today as you said. Those mid Westerner Midstate by they could put a lot of money into see no return for years because they're not under threat from these other industries in other areas. Yeah I mean it's also just. That is very hard so to do I quote unquote This is very choppy language to say do but to essentially enable a in the enterprise that is to say we're going to train our own algorithms to achieve a unique goal on if we want to do that we need data scientists. We need a lot of data. That's organized we need the ability to fail because a lot of the time even with great data and smart people there may just be an application. That doesn't work as it turns out. This data doesn't correlate to fraud. Sorry that was nine months in a lot of money like okay you know not the end of the world but that's the nature of Ai. So the stomach for that. Kind of risk of non are alive on the required budgets. Data in expertise just aren't at these companies so it's it's not just that they don't compete directly with J. P. Morgan. That's part of it. But it's also like they couldn't enable the technology will have to evolve in move down market to become more push button. At which point it will really be doing. Ai They'll essentially be leveraging data mostly trained on other people's information about IB software to point it out right. We'll move into the next era. We're not there on right now. So the highest that big players. Yes I ended up going to be a very interesting space when we're almost consuming else's AI. Or the learning that they've created a now clear is in if you're the leader in the industry let's say banking you used a You've got to return but you now have this super knowledge or source that you can then salad. Maybe spin off as a service to thanking industry. That to me is very interesting. Yeah it's it's interesting to see where and if that will happen so let me paint a picture for you of kind of how it could happen but places where I actually don't think it'll happen so you bring up a great point and I think not enough people frankly are asking about how. Ai will evolve. We have to think about that a lot. Because companies we work with are planning you know five ten years out they at least want some vision there so to your point you say. What is it like to use other people's data while you use net flicks right now? I presume or use facebook and I don't have a Netflix account. But people I know have Netflix. And that's essentially being trained off of Amazon as well Spotify were listed music These are systems that are trained on people like me. So if if I use spotify and I listened to a lot of classical but I also have like this random nineties dance like songs. I like for some weird reason because I am now. I was twelve once and I listen to the radio back. Then they'll they'll have folks who like me have maybe had similar tastes in. They'll be able to suggest songs that I would also like so. That's training someone else's data we. You're talking about I think would be a bank who really masters anti-money-laundering and then says I wonder if other banks pay US A TON OF MONEY TO MAKE SURE. Nobody uses their bank as a terrorist route for money on. Maybe we can sell this so that may happen on. We are right now for the most part seeing vendors serve that role so vendors working with bank XYZ whatever and then they're taking the data from all of them and their goal is to now have a from scratch kind of pull the cord and we can automatically rip that up to speed. We're seeing more. The vendor ecosystem take that role but some big companies may do it themselves to I think a lot of companies are worried about giving up their crown. Jewels like that but But the vendors are trying to work at it they're trying to basically what you're seeing today over. The course of the next five years will obvious but what people don't realize today as vendors are trying to drink as much data from the top one hundred is they can so they can service the entire mid market or less pushback not entirely but more or less really own that market so that's happening across sectors for sure. I think that'd be keys. You said more or less push-button it's like worse offer as the service got today. Yes is in complex to consume now behind the scenes to make it work. It's a little easier with API's on the software as a service players integrating a bit more but they still want to keep level separation. But that'll be. That'll be interesting. It sounds to me like there's a bit of a gold rush happening at the moment. Would that be good phrase? Very I mean there that's That is not hyperbolic to say you would be on point to say that there's a bit of in a gold rush. Are we gonNA see cool off? I think that that's really the big question. Because a gold rush kind of indicates a height bubble that is bound to burst on. I think it's being debated as to where that is in. Ai I I do think that the expectations were bloviating in a great number of sectors. But I also think the traction is reasonably strong enough of those where we're not going to see a real ai. Winter again in the same Gargantuan sense that maybe happened in like the eighties or so But yeah you're safe to say venture capitalist the startups everybody's in Ai Company. Now I mean there's a certain amount of the hype that is Beyond reason I say that. I'm sure there's a lot of challenges to organizations in adopting a I. Would you have a few things to do? Consider a few don't for organizations who are thinking about this. Yeah we put together a piece recently called something like a composite of fifty different interviews call the the prerequisites to ai adoption what we refer to as critical capabilities on hand. There's a number of factors here but if we think about big ones one of them is just skills so That's not only data science skill so having data scientists that's also understanding how data scientists and subject matter experts need to work together to enable a out because it's turns out you could buy fifteen Carnegie Mellon Ai PhD's on which is very hard to do by the way. But you know you could buy fifteen of them. Set up their own little office with a cereal bar in water. Slide and all that stuff and that doesn't get anything done. You actually need the subject matter experts in you need. It on both sides to be connective tissue in to work with those folks in and understand the same lingo understand problem sets in a similar way have access to the data they need. So I'm how teams work together and data science skills. That's the skill side. Another aspect of this is culture So as I've mentioned before doing in the enterprise today again five years from now. We'll see this ease off but right now it is doing science. It s s the hypothesis spending time collecting the raw data developing what we think the right features of that data should be training algorithm over the course of many many months cleaning gate over many many months before that and then we don't know if it's going to work and if it does it'll never be one hundred percent ray might be ninety something percent and he's going to be enough to save money so that kind of the kind of our D stomach. You need to have the kind of risk stomach you need to have to deal with? That is hard for most companies. And there's some bigger firms again like I said top one hundred players where They do have to now develop that kind of survival wise. They realized that that innovation focused kind of isn't isn't optional anymore. And so that's another barrier this. There's others too but those are two really big important. Once I think that's the key phrase you said. Innovation is not optional anymore. We HAVE TO INNOVATE. I in in my experience. I think organizations. We've forgotten what is innovate. Truly you know and I think These new technologies won't they can't achieve allows to refocus on my ideas on the technology itself but on wall to potential of technology might be but as you said I think You know some of the risks. Are you have to take the risk? You have to be preferred prepared to not get something right you have to everybody's talking about I just fail fast but you do have to accept a week a lot of money into this and not get the result that we want. But as I say. Edison didn't figure out the light bulb in his first. Go or first hundred goals or I you know but he had to have been funded some hell to keep going or would have stopped so we have to think about that intelligently and failing the right times I learned from it I always say it's not failure if you from it but it's mistake if we don't time yeah can't Can't disagree with you there and I think that The people you're seeing doing the Edison thing as you well articulated. Great analogy are really like the top five top ten businesses in these major sectors often. It's it's written in all truth. I mean even financial services it could be argued in some sectors. Like the top three. That really liked the less. Throw money at it right. Like the the like they're like Gung Ho on a lot of the number forty five on the global top one hundred banks for example they often really WanNa know the existing precedents of what has had are y for the bigger players right for the oaths of because they can stomach a bit of this but they actually don't have the unlimited pocket so to your point we wanna fail and fail fast but we also want to be able to pick those spaces where we have a chance to win on and businesses below the top five. Let's say sort of do do require on information about that because it's tough to lose too many times when you're you're not the the biggest town. I love that we've got the FAM- fail fast but we've also got to pick the place to win. Yeah you want at your odds right. Stockyards in terms of how easy to beat the deploy into basically proved management? Like hey see. I told you this wouldn't be a waste of time because no no headed innovation. Strategy wants to say well. I told you we could fail and we did. But I'd like more money right. It's much easier when you can say this little win. I got for you now. Imagine if you gave me five times as much right. That's what our clients folks are. Be excited to hear from now. Then you use a phrase. I've heard a few times am creative. Disruption now tell us what does creative disruption main in regards to this world of a. Um have then how is a is driving creative disruption? Yeah so I think this is really in some ways. This is not all that unlike any other business creative disruption. I think we could say automobiles the Internet. You know we talk about blockbuster minute ago. We can talk about Kodak in these famous examples of people who were destroyed on the process of the next wave coming out I think we'll see the same in so the problem with Ai. It's the opportunity as well is that we don't know exactly what these new norms will be will be the new norms of how you interact with a call center. We'll be the new norms for how you get financial updates on your investment accounts or your savings accounts are checking. Or what have you will be the new norms for how we shop. Some of these things are being felt out the big tech players but some of them are going to settle somewhere and as it turns out there may be firms that just cannot get the infrastructure right to handle on conversational interfaces for example or to handle voice very well and it may have a tangible crushing impact on their ability to service customers retain customers etc. And so sort of like again Internet blockbuster there will be some of these facets of ai that will be so critical industries at some folks a win and some just Just won't get gobbled up. It'll be so similar analogy. I wonder these days. Are these chapel. That we see replacing calls Going to disrupt in any way or they just more of an annoyance to some people because they seem to do no lower than enlisted people going to in the past. They don't tell right now Chat bots are so we. We have a very robust landscape of banking as are numerous sector. And I'll just speak from the perspective of bank in banking are there are on it chat bots conversational interfaces in terms of press releases from the top banks is lettuce the re times bigger than any other category of Ai Capability in terms of what banks say they are doing but when it looked when it comes to investment there are rules of business functions like compliance like fraud like cybersecurity. They're getting astronomically more money. And it's because there's actual results there the conversational interface stuff looks great to your customers to seem like your hip. It looks great to your investors to see like your hip but at the end of the day on really. It really isn't garnering that much of a of an Roi in Modi's actors. I'm not I'm not a conversational interface pessimist but I. I am a realist in that. It's a bounded technology and if it isn't treated as technology that can handle some low hanging fruit and otherwise should be routing people to a human resource instead of trying to talk win is running out of what the hell say if we're promising that that continuance that real conversation. I think we're over promising in ninety nine percent of instances today so Benjamin is where these data basically screams height. You just look at where the money's going you look at what people are talking about an. It's so obvious where the hype bubble is again. I'm not a total pessimistic. Good vendors there but off yeah chat bots today. I think the way to go with them. And against some of the experience I've seen with chapel not only have place but their their position than scituate to eliminate the person at the back rather than as you said to still route to a purse. Not The back when the opportunity comes across so we need to speak to somebody. Move away from the There is somebody there human actually go talk with so I think what we're danger of is using some of this technology to replace the human at the end of it rather than to facilitate some of the interaction still have people at the M. though the the chain so to speak leave me into this. I I talk a lot about this subject on the humanity of it. The human the person aspect of it. I guess so sometimes I I asked. Are we losing the personal connection with people and customers by automating more the question? I have down. Maybe you could help. Is You know how does this? A affect our personal connection with customers as a business as business start was yes. I think this this could go both ways on on the one hand you know not not because I'm a bad guy but because when it gets there on people will lay off call center people because the world because real actual the world because the state of nature is so angry and mean. And if you don't and your competitors do and you spend more money than they do you will lose and so at some point there will be sloughing when it makes business sense. New companies grow when you companies grow. They will grow without the the employee bulk. Now I'm not a fan of that and I think you know It's necessary. I'm not saying we should resist all of it. There might be regulatory needs for these kinds of things but I don't think that If you are the CFO If the choice is keep losing money until you lay everybody off and we all find a new job or we gotTA slimness department down because by Golly all of our competitors are reducing their costs. Those are those are real adult conversations. And I think we're not really seeing those crash into financial services retail today for is specifically. But I think we will when it goes to customers. I think that this could potentially go both ways. It seems obvious that if a I- stretches US away from our customers it feels very hard to win doing that. It feels very very hard to win. So there's there's some examples like an apple for example where I'm getting in touch with customer services like impossible but the product is really good and like so people put up with it but I think generally most part A. I will be about learning from all of our conversational. Interface interactions all of our phone calls and being able to better understand Meta trends. It'll be about looking at our interactions on our website and better being able to service our customers. It's almost impossible to see a writ. Large wave of customer neglect customer distance. Plus business winning at the same time that they seem essentially diametrically opposed so. Hey I for companies who win will almost always in some way. Connect them to the preferences. The needs of the people who pay them the dollars. So I'm not too worried about US being Disjointed inherently by in that way it should connect us to our needs and wants and desires. Whatever that is. That's intelligent way. I think of aggregating a maiden things like that. Well we'll see I mean. We had manufacturing industrial age as well as the revolution on the Industrial Age Are we seeing people replaced by machines? I think about the next stage now where we're seeing some of the service economy being replaced by software driven by big data facilitated by cloud. But like that when there was this displacement from the mechanically machine air new jobs came up new things came to. Hopefully we hopefully we. We continue on that tangent. Now down in one of your headaches talks. You've done a number of those into really good. I think our listeners should go have a look at some of those as well. You also talk about the emergence the automation economy. Now we did a segue into it there but what are talking about you talk about the automation economy. Yeah the automation. Connie I mean so I. I don't know so much if if it's economy. I think I know the text you're talking about is the one at University of Rhode Island which I do on so I don't know if I use that phrase particularly but I think I did talk about the rise of automation certain job categories. Do you WanNa kiss explorer that Lord. Yeah because to go into it. You're talking about a Driving let's say white collar automation so to kind of senator. I know you'll should. White collar workers be be worried or do they fear that this automation economy will take their jobs as well. Yes so the It's a good question. I think that the broad answer so the tedtalk actually does a reasonably good job of tackling. Nece on it's just Dan for Gela Connects University of Rhode Island or whatever but Pretty easy find on Google Talking about three main facets of what are the job security pillars that we will likely be able to stand on us and so on one of those is is what I refer to as context and so we talk about what should white collar folks be worried. Serve depends on their role. I think right now because we're not seeing the bowling ball. Destroyed pins across finance and retail and heavy industry. Where we're not there yet. So so real visceral. Will my children eat a meal? Worry is probably unwarranted almost ubiquitously in most white collar jobs however if we want to look to the future and secure as much certainty as we can about own Value in in an increasingly automated age. One one factor here is is context so I use a bit of an analogy here until about inputs. So that's whatever lands on my guest or my computer screen joke about my my work on those inputs. What do I do with this thing in? Front of me is a spreadsheet that manipulate in the same way as a form that. I check in the same way is it. A whatever is it repetitive. In wrote and the outputs do. I send it to the same inbox forward to the same person. Put IT in the same file structure. Whatever the case may be do the in the works and the outs more or less. Look the same without me. Having to know anything else in the business in other words I don't have to look over into marketing onto look over into procurement. I'm just in my world of inputs work in outputs. Those are the roles with no context right. They just. It's just work on those roles at the highest risk of automation across the board regardless of industry regardless of of. Gio allegiant and so folks really feel like they're in those kind of physicians would be in the spookiest spot for sure. Then other things. I guess for people to think about about their careers. I guess we're no longer in the era of Goat's college get your degree on the whatever that is go to the bank in this example and have a career for forty years get the gold walked off nothing fancy pants and I think those days of leftists behind I'm then we've technology changing so much that win away icy it in win people's careers going house to almost reskill to three or four times along the way to keep up to keep relevant about the rate of change. Were talking about this You know automation economy is going to force us to do that. Or if we don't do it we may get left behind. I mean nobody could deny that. I think even before we talk about you know. My Grandmother had one career upgrade that she has to do which was learn to type. You have to look at a typewriter typewriter. And of course for my parents were a couple more right. My Dad never really got with the Internet stuff. He ran a little carpet store and God bless a meat. He had to learn some new technologies and new ways of doing things. I think that the folks who now will in that store are having to probably even have an email list of customers who send messages and things like that and at some point. Maybe that'll turn to marketing automation on. So yeah I think that the technology tools. The ways of doing things are Involving faster than ever there is reason to be nervous that not everybody will be able to or want to endure that kind of quick hustle light and this will create that stratification of society writ large folks who who either want to or are adept in on consistently evolving driving forward into new problems to be solved and learning new ways of doing things and folks who For ability or for preferences just absolutely do not WanNa live in that world where it's a new dam tool it's a new dam workflow every six months I think that society. Maybe we'll figure out a way that that'll be tackled. But I think it's it's worthy of consideration in terms of how that's actually going to happen but we found in our Recent said global digital skill survey was one of the critical findings looked up and organizations now requiring people to have the ability to learn on rescale quickly and apply those skills as opposed to coming into a new organization with degrees diplomas. I'm certain types of learning. It will last for the length of time that they're indoor organization so the expectation is always there but we also found that not everybody is capable of reskilling but we do believe that they should be afforded job opportunity one way or another two jobs relation or twitter or social channels. I think that that's that's a big thing and the other any other. I guess ethical implications of say using a uniform organization or is there are ethical points from an society in general. I mean there's there's all sorts of considerations ethically in terms of Longbow societal impacts. There's things like you just brought up. Can everybody keep up and learn everybody able to rescale? I think it's it's somewhat obvious. It not necessarily. Everybody wants to do that as the aptitude for that. It's not the thing I'm it's very demanding. I think for some folks on and I think that how that is tackled whether it be universal basic income whether it be in this broader governmental sort of factors that I think we could say have ethical import in some in some fashion umph within businesses on. I think that a lot of the time a is being applied to kind of snipe out things that could be those kinds of risks so an example here is regulation or compliance in the financial services. Space there are things that I can say on the phone when I'm selling you an investment that by Golly are not okay so that they're just not an there's there's ways that may be money to be transferred that by. Golly if the regulatory folks knew that money came from this party and ended up at the end of the day going through the Shell Company. Into this party we would be slapped hard because you know funding terrorism or crime in some ways not right but also just because there's punishments there so we actually see. Ai Aiming to kind of fire away particularly in finance also in life sciences at these things that are on compliance unethical risk factors so I might open up some but it might also help. Close the door on some of these things like fraud. like Insider trading sort of compliance risk for example. So I think it'll go both ways. So yeah I think there's not the boat. Waist there should be careful on both dont critical question. The final word can make the world a better place in come. It's far too much. It's it's the title of one of my Ted talks. Which if you're you're not ready to consider very scary far off post human intelligence stuff. You probably should avoid watching that talk altogether if you wanNA give yourself nightmares. Maybe you can But Yeah I think you know in the long term like forty years out. I think we're GONNA look really wild shifts sort of human condition and hopefully that that is for the good. I think it seems safe to say that in the near term on the aggregate Ai. Kind of like the Internet will be a net boon for wealth. Broadly reminding to regulate it differently. We might need to change the technologies. We allow our teenagers to us. If it's making the whatever the case may be right but I'm on the aggregate I think hopefully net boon To sort of productivity globally on at least in the near term and so my hope is that at least from the business perspective. The answer to your question is yes. I think long term. We've got bigger considerations but relatively near term an optimist. Good good the optimistic with you there as well. I think there's always good. At least we gotta get regulation in order to get to that point. We have to have things out there then catch up. We have to see what's happening. What the potential is look at the Internet? You know think regulations coming in year on year four that it needs more book. You couldn't regulate it for that years ago. Because you wouldn't have known the onto the scope you will know people would've used for so unfortunately it loves KOCH OPENING GIFTS OUT WINDOW. I think of opportunity for US TO BE MYSTIC ONTO. Try and fail and as you said fail faster. Put Your money where you want to stack your chips. Don't go all in and try and try again. Daniel Fidel thank you very much for joining us on. Today's CCC talks. I think that's been really enlightening and they're thank you for your insight. Thank you very much. Thank you for joining this episode of CCC talks. We hope you enjoyed this episode and walk away with a ton of actionable insights if this is your first time joining us. This is US extending a personal invitation to you to join other. It and business professionals. So please subscribe on itunes youtube or Google play. If you're struggling in any capacity in your digital transformation journey contact we'd be more than happy to guide you and find you the right certification courses to help you manage the challenges. Modern businesses are facing this was CCC talks until next time.
Best-Practices for Discovering Valuable AI Opportunities - with Adam Oliner of Slack
"This is daniel fidel. And you're listening to the business. Podcast art of our mandate here at emerge is to constantly build our library at emerged plus that is to say construct. Best practices frameworks infographics. For what a strategists and a catalyst really need to have on deck that is to say ways of building strategy. Wade's of streamlining adoption ways of building. Ai roadmaps and also ways to find a opportunities within their business. These are the non technical business skills that bring ai to life. And that's ultimately would emerge plus is about and we have a huge panoply of these frameworks that we've built over time and people often ask. Well how do you do it. The good news is we don't have to do all the hard thinking. Some of the smartest folks in the world when it comes to applying ai real world or our guests here in the i in business pot gassner in our network in our rolodex emerge artificial intelligence research and it is from brilliant minds. That many of our best ideas and frameworks have actually come from and our guest. This week is absolutely no exception to that. Mind theme adam. Oh liner at. The time of this interview was the head of ai. At slack slack obviously was acquired. Not all that long ago for many billions of dollars slack a very well known silicon valley unicorn. Adam is now the founder of a stealth firm in the bay area. So he's no longer with slack. Think after a company gets bought sometimes very talented people spin out. And that's quite a natural transition. And adam had been with us last year. And i decided to pull him back in and talk about a topic that he hinted out in his previous interview. But we didn't get to get into in-depth and that is how do we find a opportunities. What is a lens of thinking. What are a set of steps phases to uncover. The i fit to uncover where a value can be unlocked in our business. How do we actually look through a pair of goggles. that will show us. Here's where the business needs. And the data assets could come together and actually deliver value in the business atom obviously has a very robust. Technical background is head of ai with silicon valley unicorn. Certainly need to be well rounded there. But he does a great job of being able to convey these phases and steps in the way that he thinks about the process in a way that essentially anybody can listen to can use can apply. So i appreciate adams way of explaining things being able to break things down. And i hope that it makes it easy for you to apply some of these ideas in your own business. If you're interested in using the wider library the best practices for finding ai opportunities building a strategy building a road map and even conveying the roi when it comes to making the business case to leadership you can learn more about those best practices as well as our full a use case library at emerge plus it's e emmy rj dot com slash p one p plus and then the number one yemi rj dot com slash p. One you can learn more about emerged. Plus they are without further ado. Let's fly into this episode is adam. Oh liner you're on the ai and business podcast so adam blad to have you back with us here on the program. I really enjoyed our last chat about these strategic advantage of data. We've got a couple of good topics to start with today. I wanted to kick us off on the theme of building executive. Ai fluency the show including your last episode to try to get smarter and be able to enable this stuff in their business when you think about you know executive teams leaders who might not be technical building a affluency wiz that involve for you. What do people have to learn to make stuff work. Yeah thanks for having me back then. A good question. I think about in terms of sort of a constellation of potential business problems that you would want to attack with ai on the one hand and then a bunch of data on the other hand and the process of assessing those business problems assessing the data and then assessing the potential bridges between the two. And that's the sort of image that i have in my head when i think about identifying a opportunities We can dig it on each of those so if you think about the sort of business needs like what are the brake problems to target with machine learning with the mostly you're looking for a well formed problems with measurable impact so i sort of negative example would be. I just wanna understand my business better. Well okay i. What are the units of understanding here and how. It actually measure. Whether or not. I've accomplished us. You can certainly slice up that problem and find good sort of a opportunities within that problem. But as a general like that's the project it's not really well formed and doesn't have measurable impact a good example would be something like i want to reduce the average time it takes a user to perform some specific task and in fact that's a great example from a class of problems that are often really good targets for machine learning inside of a product which is still look for user friction or look for dead ends basically any place where a user is asked to make a decision or perform some repetitive tasks. Those are good opportunities for automation. And i think this is maybe sort of unsexy application of ai. But it's a really good one at a lot of people on a chase after the shiny new product or the shiny new feature. Those are usually more expensive and harder to get traction for but if you look at all of the places in your product where repetitive workflows or really just unpleasant or difficult for users. those are great targets. Yeah yeah so all right so many things to tap into here. I've got this mental image. Hopefully listeners do as well of business problems on one side. Eight on the other side. We're gonna fill in the blanks on both sides together with you. Obviously working at slack. You guys have woodall's noodles of users and thinking about user friction is is a nice guy for you guys. Probably a rather commonplace where male would be deployed in used and leveraged when you think about kind of fleshing out that grocery list of conquerable bounded business problems that have a measurable impact. I'm imagining this in the mental left. Hand side here. if. I'm building that grocery list because i want to be able to pick the ones that are going to be. The best fit for my business would have other questions. I can ask myself to flesh that out. Because i fear that a lot of folks are just looking at what their competitors are putting out in press releases and thinking that that's what the landscape is. What are smarter ways of thinking through it. So one way to think about it is in terms of what are the kind of m. l. capabilities. And are there. Mlk abilities that are common across different business problems so to to give an example of this We built recommendations api at slack internally this contrive recommendations throughout the product and so we build that one emily capability and now we can drive recipient recommendations and the composer or channel recommendations when you join a new channel or from slack bod and so on and we don't have to do much new l. on the end it's just sort of wiring up the front end and so if you have a class of business problems that can all be served by the same back end. Mlk buddy that might be bridge worth building. Cool and i can imagine that being once. We have the grocery list on the left side. We can say do these cluster. The word recommendation comes up seven times guys. Is there a way for us to build something that can kinda tackle all of those that sort something. You're getting at the exactly. Yeah yeah and that might help us. Pick the business problems that are important. You also brought up something and maybe there's some color sprinkle onto it around are they're headed workflows fry their employees or for users is at maybe another useful answer. What are the you know. it's it's almost. it's almost a little bit of a a limited lens. Because i think that the idea of a equals automation is a pretty limited way of looking at a i however for low hanging fruit. It's not a bad one. I mean it should be in our bag of tricks. Do you think that can be a useful way to start building that list on the left side. I think it's certainly a useful way to get started partly because when you have those repetitive workflows you're getting pretty structured label data often from users. They were faced with the question. They perform the sort cognitive task of answering the question in some way. That is the label. That's the answer right one and you get a lot of these examples in now you have kind of a nice data set for doing machine learning and this is specifically on the product side. There are of course lots of potential business problems that you could tackle internally for example so if you're not making a new product or changing existing one you can think about on the back end. How do i make my services more performance or reliable. Or how do i keep things from breaking in the same ways that they've broken before and things like that you know to give an example of this. We recently built a spam filter at slack. So people were using the invite to slack feature as a way to spam people with email and so we built a an internal tool that filters out spam and that was a really nice and i'll project it was very well contained and often when you're shipping internal tools. The burden is a little bit less. You don't have to jump through as many hoops to ship. Non customer facing feature and so looking internally is another good place to identify a our opportunities in order to do this spotting of opportunities. Is it useful for you. Know we're talking about executive fluency. Is it useful for leadership or the people who are in the room during the brainstorming here to have some familiarity with the use case range in their industry or adjacent industries were some conceptual understanding of how data algorithms come together to solve problems. It feels like a lot of the time adam. Those conversations happen with. I don't mean ignorance in insulting. Why it just just not really good context on those two things how important those or there are other kinds of knowledge. You think the people in the room have to have to spot problems at worthwhile. Yes so let me maybe finish fleshing out these like two constellation them because i think that gets to the answer that question awesome so on the other from the business needs you have your proprietary data so this is your strategic asset. We talked about this. This the questions you need to ask about the data are things are they clean is it. Clean is it. Reliable is a timely. If you have a single clean current correct. Tabular data said that exactly includes the information. That's relevant for a problem than you're in great shape but you almost never do so the question becomes what will it take to get you closer to that and modern doesn't really require table like i described exactly but the further you are from that it is so if you have data scattered across a dozen different systems inconsistencies like mismatched ideas that make it. Impossible to join across. Those data sources are missing messy values than if nothing else you know where to start in your readiness journey so might wanna google concept called future store and just start getting your data into a state where it's at least tractable to talk about. Do you have the necessary to tackle one of your problems. And then of course. Now you have your your problems. You have your data. How do you bridge the two. And i think even at the executive level you don't need to know algorithms and how they work but it's useful to understand what kinds of things l. can do i think if you imagine it as just a a magic box. That thinks like a person does. Then you don't have any ability to assess the distance between a data and a solution honestly. The list of things that l. does is kind of small in a sense so it does things like prediction. You have some number of inputs in. It's going to predict a number category or something you know. Or clustering is another example. Where you're just grouping things and honestly what will like those two things that i have just said really does kind of cover the vast majority of machine learning because you can think about you can think about forecasting as just being prediction for the value is for some future time you can just think about ranking as being predicting numbers for a bunch of things and then sorting them based on the predicted numbers you think about recommendation in the same way so in a sense like once you understand that. Like i'm trying to think about this. As either grouping things or taking some set of inputs in making a prediction by number of category whatever. Those capabilities are serving as your bridge. And so if you can kind of understand those which is not. I think that then you can start to think about the problems you tackle so going back to thinking about workflows and user friction if you're presenting a user with an empty drop down and they have to select something from that drop down. Well they have a bunch of information about the tasks they're trying to perform the context of this drop down like. What are they trying to accomplish. And so on and they're just trying to pick something from a list. So that is a great example of a prediction problem. If you can give a model the same information that the user has when faced with that blank drop down and you just try to predict which of the things are they going to select and you could make that default instead of just showing them a blank drop down. It's a very small example. But that's the kind of thing that can do and really everything else is just kind of building up on top of that in various ways. Yeah one of the one of the as i really like what you're saying here about not remotely technical at the end of the day adam but having gone through injure ings course painfully and slowly in core understanding that ultimately you have your clustering and you have your you mentioned kind of prediction. Kind of taking whatever wherever m. l. is is being applied in saying. Well it's one of those to apply this kind of lens. I think that's helpful. So for you thinking through kind of hey do these kind of things and then having some examples of each for you. That will allow people to make problems. Click and say. Oh yeah that's prediction our own. That might be clustering or something like that. Sounds like that concept is important. I think if you read press releases or you take corsair courses or something you can be led to believe that has to be really complicated and difficult but ends base. There's sort of a small number of fundamental capabilities. And if you can understand those you're off to a great start and then there are other ways to think about it. Which is that. Some people have hooked together forecasting clustering and prediction and really complicated ways. Where now you have you know self driving cars or something like that but a lot of the time. Those solutions are reusable. So an example you can just go in download an object recognition model. That's been pretrained. You can show it images of sort of every day type of things and it'll tell you what's the image you don't have to go and rebuild that and so if you have a kind of a list of those sorts of capabilities that you know are available to you then you can start to think about ways of plugging those together like to give an example if i have an object recognition model than i could if i have a product catalog someone could take a picture of a thing that they have and it could say this is a chair so let me now look up on my website. You know other chairs. And i could give them a list of those things taken a picture of it but you know that there is. There exists this capability of going from an image to a list of things in that image in taxed and sort of building up. That list of things in your head is sufficient. I would say to start to map out the bridge between the data and the business needs right. So i like that. Hopefully that's the take home lesson for those of you listening in last question on our last sub question on this first topic together add on is around the other side of the table which is which is data. You had mentioned you wanna look at things in saving. It was a clean reliable timely. It's normally not so how far away from getting their one bit of color. Throw on that is do we look data first and then figure out business problems. Do we do business. Promise first and then go say well. What data would we need for that. We do the both concurrently. Do you have any kind of order. you prefer here. If we're going to think about finding opportunities usually start with a business need. I think having a sense of what kind of data is available to you when you think about the business needs is useful if you construct a list of problems you wanna tackle and find out that you have data for none of them. That feels like a bit of a waste of time on the other hand. If you have a problem that sufficiently valuable to you even if you don't have the data or data as a mass you can go about addressing that yeah add logging to collect the data that you need or do something even more heavy handed like you know use mechanical turk or something to get the right labels and if it's a sufficiently important problem for the business then you can sometimes find the data but if you start with the data and then say you know. Where should i go from here. I think you run the risk of targeting low hanging fruit. That isn't necessarily the most business being seen. As sort of churning out an endless list of low-value features is a good way to not change the culture at a company. He add stuff to get by in unless somebody can tie something to something meaningful as well. The the enthusiasm for toys is limited at best so okay so you make the business prominent say okay. What data with this require. And then that's when we go look at the data. Because the i guess the way i see it is. There's so many pockets. Were data's being store in so many ways it's impossible to just you know audit everything or something in one fell swoop but if we say look. Here's the major cluster problems we have. Here's the kinds of data. We're gonna need for them that we can do. Our investigation in auditing only where it makes sense that we don't take twelve months to say here's the state of every drop of data in the business. It seems like that might help. Honar auditing process to some degree. Damore yeah i think starting at the business problem and then thinking about what the possible bridges that could get you there. What data those bridges need to start sense and For those of you listening it's useful to have a data science savvy. Technical person like adamant room to Validate those hypotheses because what a what a pure business person might say. We would need this kind of data to solve his problem. Sometime isn't isn't exactly at so multiple expertise in the room. I will say you know someone. Like me is not sufficient over sure but i mean even someone who is an absolute expert on all of this will not necessarily be able to tell you exactly what data is required for one of these problems so you might pick a good business problem and say okay well. Let's start logging x. Y. and z. That should be enough. You might do that and then go and build a model and find out. The predictions are okay but maybe not sufficiently good for the business problem that you're trying to tackle. There's not a great way to assess that until you look at the data and actually try it big time. Yeah we can't get around the fact that this is iterative. Can't get around the fact that this is what probabilistic i mean. You know who knows how many problems at slack even you guys have been like. This should be solid. And then it's like you know what the data just doesn't shake it out as it is we gotta take a different approach Imagine that happens every now and again. Absolutely yeah and there is not a great solution There are a lot of questions that amil engineers will here and they'll sort of cringe because they know that there's never a satisfying answer to it. Which is like. How many examples do you need to train this model reading. I have no idea. Like i'm never going to be able to give you a satisfactory. That question i'm sorry. And this is the sort of form of that but it's even worse because you may not even have the data yet and so what what are the features that matter. I you know. I can make an educated guess as an expert but i honestly won't be able to tell you until we try it. Yeah i i. I guess i would say this and let me know if you'll give this a thumbs up thumbs down. If you got smart subject matter expert folks in the room business people that know what matters to the business in the bottom line and data scientists will at least have a damn clue as to how data has been used historically. We've got our best. Chance are not not a guarantee we've got our best chance at running down a rabbit hole. That isn't empty. But like you said it might still be empty but but at least love a better shot. If you've got a little bit of a thinking mix to go in. Is that correct. That's certainly true. Okay so second team. Here that we chatted about off mic which. I'm excited to dive into is really around assessing ai. Readiness companies are listening into you right now and we're going to be turning this into all kinds of additional content and really wondering where are we starting from. And what do we need to know about ourselves to know where to begin with are a journey because so many enterprises are in exactly that position. I know you've done a bit of thinking about this beforehand. Where did you wanna get kicked off on this topic. Yes so if you've gone through the assessments that i just talked about so you have a reasonable data story. You have an understanding of the kinds of things you can do with that. Data candidates set of business problems than the last step is to evaluate some particular solution. You're selected a bridge that you might wanna build now. This will seem like a strange thing for me to say out loud. But there's nothing you can do with. Ai that you can't do without it just possibly much much worse and on the flip side the ai. Solution is usually more more expensive from an implementation and maintenance perspective. And that's the question that's before you let's say that the best possible solution is basically up against. Let's call it a year ristic. Sometimes you almost solution will be much better but other times your data just doesn't cut it in the ristic will win again. The ristic almost always wins on implementation costs. And if there's an easy eurispes to try you should almost always try it. I if nothing else it sets a baseline that an mel solution would need to be if it turns out that the ristic isn't good enough than this is when the company needs to ask a few questions to decide whether or not they want to build the solution. There are basically three of them. I would say the first. Is this problem important enough to the business. That i'm willing to invest in a better solution than this ristic. The second is how much new infrastructure do. I need to put the solution into production. The more emily info that you already have the less new stuff you'll need so a mature company will already have most of this infra and will instead be thinking about something like additional cloud spend or something and the third question is do. I have good reason to believe that the solution will outperform this ristic by enough to make it worth. It might be better. But if it's only better by one percent is that worth the investment not require. Any good engineer will have run some kind of an experiment to investigate this improvement. But sometimes you don't know until you try. yeah. I liked the idea of going through this lens. So what you're saying is when we're picking a project use use a. I went into the right tool for the job. Right if it's like. Hey we've got a rule set here. That really cuts the mustard. And you know like you just mentioned with think the expense to actually eat out. Another two percent is going to be pretty. Big from where restarting maybe. This isn't the right. The right move so for for you. Those three questions might allow us to take a further filter. Got the grocery list. Got the data. We've got the bridges that might be potential projects. Okay let's run through this new ring. This is kind of what you're advocating here. Yeah and so. I can give you an example from slacks recently published a blog post about the spam filter that we built and the state of practice when we launched that project was that a bunch of people were manually curing some risks around what constituted invites bam. What didn't you know the the word casino in sorolla this this word and so on and so every time. Some new pattern of spanish behavior would emerge. They would have to manually. Go in and jerry those restaurants and so there there's a baseline like we know how well that's performing. We know roughly. What's getting through in. What's not in particular one of the things that we identified as a problem where how many valid invitations were being filtered out by arrest it turns out some people run legitimate casino businesses and maybe they wanna use lack at so. The target that we needed to beat was the performance of this year. A and the question was like how much is it gonna take to get an nl solution into production you know. Do we have the data that we need to. we have the labels. Do we have the infrastructure to train a model and put it into production and so on and so we went through that evaluation and it turned out to be really successful. We had the data on the labels that we needed everything sort of lined up and so that was an example of a successful project. But it did start with ristic and now is sort of doing their duty for a while. Yeah yeah but like you said then you have the baseline right. If you just went in with the male model you could always be asking yourselves honestly guys if we just had a list of a thousand two rules of casino and comic sans and whatever else you wanna do. Would we be saving money and doing better than we are at this thing but now you know you you can you can figure it out can say how much should we think we can improve it. And then you can move forward from there. And this touches on maturity itself so one of the factors that you brought to bear was hey based on where we're at what's the additional investment. We're gonna need to actually potentially enabled us and company. That's done dozens of ai projects in different corners of the business firm like slack. You guys have lord knows how many algorithms in deployment but you know enterprises just starting off. Maybe to have less of people or somewhere between here. What does it look like to get a sense of where we stand. I can imagine. I may wish osama coo or on up ahead of head of compliance or vp of fraud. Somewhere to bank. And i'm saying well okay. I've got some problems. But what is my hand maturity. You know we're we're r restarting from around here. What are the couple ways you would wanna define an maturity to know what we're what we're standing upon how much more investment we need to take. I imagine a lot of non technical folks would need the conceptual understandings that you have three parts. The data which we've already talked about infrastructure. Which can be sort of basic infrastructure like the ability to train a model and to serve predictions and production at scale. Things like that but can also include m. l. capabilities that you've built for other purposes so a recommendations. Api for example might have some models that you've already trained on your data and that's already in production and a company that has api like that might be further along in the maturity spectrum and then the third part is the ability to put these things into production right like actually ship features or shift capabilities that deliver some sort of a a value to the customer to the business and some of that is cultural. I mentioned that sometimes you don't know whether an initiative is going to be successful until you at least try it on some smaller scale and that is sometimes a cultural leap for a company. The the idea that you would say all right. We're going to try to ship feature acts in in q. Four and then halfway through q. Four you say now we. We did a test model and the performance wasn't really good so we're gonna abandoned that and do something else for some companies that might generate embarrassment or frustration. But i company would say yup. Okay it was. It was the right thing to try. We went in. It was a good experiment to run. Now we know we either can go and collect more data or a try this again in a year or whatever or just say like okay. It wasn't a good target is not good enough. So those sort of different elements the the third of which is may be more like organizational or cultural along with data infrastructure. I think are the key pieces to look at when you evaluate maturity cool so good conceptual understanding there for the listeners. Here last little sub question to the second topic as we wrap up. Adam is around using initial projects helped to build some of that. Ai maturity you know like you said if you have a certain amount already in terms of you got some talent. You've got a culture that can kind of embrace ration- you've got in some of our listeners. Familiar with our model for a maturity as well we can do a little bit more but projects. The r y. The project is not just made this financial return. it's also. Hey we ness stand on this new higher level where we can enable other things we can more nimbly adapted move with ai. Broadly how do you think about picking projects that are both good fit for the data in the business need but also maybe a good fit for leveling us up. We call kind of capability are why if you will you like to think about that. Yeah it certainly the initiatives that people like to talk about are the high value ones where they say this moonshot if it's successful will totally do in in practice the companies that do that have money printer right. They have some part of the businesses printing money and they're fine with a massive upfront investment. Because again it's not necessarily guaranteed that this will be successful. And i think if you talk to people who work at for example self driving car companies. They will tell you that. They've poured billions and billions of dollars into this problem. And you still do not have the ability to go and buy a self driving car and the fact is it just turns out. It's a much harder problem than a lot of people were expecting or hoping or thought it would be where fifteen years since You know. Stanley winnings the darpa grand challenge. And still you know. I don't have my self driving car. So how much longer is it gonna take. I don't think you can get a really confident answer from anybody so again unless you have somebody who's pouring billions of dollars in your initiative you probably don't wanna go for the moon shots so instead the type of project that you want to go after something that is sufficiently valuable that it's worth doing. But maybe has a lower cost for some reason and this cost could be lower because it doesn't require all of the infrastructure or a massive amount of data maybe like our spam filter. You really only need the invitations and some simple labels and you can train the model on your laptop and the traffic goes to the prediction. Serving is relatively low. And so you know. It doesn't require some massive serving those predictions at you know billions of times in second or anything like that and so. That's a great target where it's really valuable to protect our brand and protect our users from spam and doesn't require all of that investment and the good news. Is that anything that we build in service of that project in now be used for other things so if you for example don't need to train a model but you do need to serve. Predictions woke great. Now you have prediction serving infrastructure. And the next time you go for a project the the cost to that as much lower cool if i try to nutshell this going for the wow this would be just a a rock star project with a double our revenue with this amazing model right that. That's maybe what looks cool in a magazine. But what you're getting at is an practice. Where iterating experimenting. Something's are winning more than others. Some things are flopping. But we're okay. Were being a prudent about our experimentation. We broadly build up this floor of ability to be adding value in enough places and take advantage of enough new opportunities that that's really the advantage here more so than thank goodness that home run workout for us. The big projects are high risk and high reward. The smaller ones often have a relatively low cost and are sufficiently valuable that there were doing and they often reduce the cost both of putting into production project number two. But also even the cost of evaluating the feasibility of project number two so if you have the ability to really quickly train a model and passed out like running experiment in production to test. You know with one percent of your traffic whether these predictions are serving purpose if you can do that relatively cheaply than it's cheaper not just to to build project number two but to ask the question of whether it's worth doing at all yeah so there there's there's an roi in just the learning like you said and do enough small projects you get a feel for what's viable and what's not unlike if you take a big swings i where a lot of your guesses. You're gonna learn the hard way for the first time. And maybe that's not good for all companies and if if on that maturity spectrum you're relatively low in the organizational and cultural maturity than shipping. Those quick wins those things that are likely to succeed. And don't have a huge cost that can start to change the culture at an organization because they say oh. This is pretty cool. This works really well. That was a win and so they'll be more likely to say yes to project number two even if maybe the risk is a little bit higher yet. That's that's gonna be. The reality gradually Frog the frying pan. It's the wrong analogy. But for some reason it's the only one coming to mind where the c. suite isn't gonna wanna listen to this whole interview. Adam take it all to heart. They're doing other stuff. They're not bad people. They're just doing other stuff but but if they can see enough chip away value than we may get some investments for some bigger projects will the track record of success. That's that's the right way to think about it. Yeah that's sets probably better than a frog and the frog dan anyway out of this has been an excellent second interview. I really appreciate you jumper back on with us. And thanks again for sharing your insights. Thank you so much for having me again Episode of the and business podcast. I hope you enjoyed this episode. We do our best to work hard to find a good mix of talent here finish show. We like to bring on startups. You might not have heard of. We like to bring on big blue chip companies. You we've had head of ai at raytheon really high level folks at comcast hsbc etcetera. We also like to be able to pull in the folks. They're moving the fastest with a and that is to say silicon valley unicorns so adams perspective is important to us. I hope it's important and useful to you and if you want support the show and you've learned some things he'd been able to apply the ai and business podcast. It would mean the world if you could support us by leaving us a five star review on itunes. What is now called apple podcast. You can search for the a and business. Podcast drop us a five star review and type up what you like about the show. What you've learned how it's been useful for you because it is your feedback that i bring back to my team. When we think about our editorial calendar we think about our interview calendar. It's really your ideas that feed the show helped it to evolve over time at your ideas that have helped us to recently spin out the a consulting podcasts. So for those of you aren't aware we now have a show called the consulting podcast you can find that on itunes find on spotify etc. And that was your idea as well. So you're reviews help us generate great ideas and they also really do support the chefs if you want to support show consider leaving tribes view on itunes. Miss now called apple podcasts. And otherwise stay tuned for the next episode next tuesday here on the a and business podcast.
AI at the Edge - When is it Better Than Centralized Compute? - with Geoffrey Tate of Flex Logics
"This is daniel fidel. And you're listening to a in business. Podcast one of our here from our market research work emerge artificial intelligence. Research is that ai. Executive fluency is really the linchpin to return on investment for projects. If we have leadership that conceptually understands what a i can do what it takes to make a project successful in terms of a deployment or Failures if we have executives understand a representative sense of use cases so they know so realistically what might fit where we would be able to avoid a lot of the early stumbling blocks of where the i kinda fell on. Its face in the enterprise and fortunately executives are becoming more fluent. We obviously want to perpetuate that with our work here at emerge and with this podcast and there is a topic about what it looks like to put. Ai into use that we don't cover a lot but that is worth having on the radar and that is your choice around hardware and compute. We're not gonna talk technical. This is not a show for people who write code for a living. It's a show for people who decide on projects build strategies managed budgets. That's who you folks are. And that's who were delivering for but hardware is still a useful and interesting topic depending on what industry you're in and it's something everybody's going to have to think a bit more about in the decade ahead our guest. This weekend's jeffrey tate. He is the co founder of flex logics flex logic makes a hardware. Jeffries fell out mountain. View where i used to live myself. And we speak this week around when it makes sense to do at the edge and when it makes sense to have centralized Advantages of both. And what are the instances. When ai at the edge might be the smarter. Move to play. We've talked in the past about a in the in retail for computer vision. We've talked in the past about ai in utilities and transportation at the edge but now we talk about conceptually and this is the making the business case episodes. We get a bit more conceptual here. On thursdays we're going to talk about. When does it make sense to have your algorithms doing their work. Far away from your central compute versus. When do you want a pipe. That data somewhere to have the work done. They are having a simple rule of thumb for dealing with. That is awful useful. If you're thinking about use cases that involve complex compute needs and. I think. Jeffrey does a great job of simplifying. Some of those insights so hope you find this episode helpful if you want to support the show and so many of you have been kind to do so over the course of the last six months. It means a lot to me. Please do a five star review on itunes. Let us know what you like. Most all of our best ideas in the last year or year and a half about the show have come from listeners. Like you either contacting me on lincoln or dropping a kind review on itunes and letting us know what kind of episodes you like what you like about the show that's helped us to mould our material and also really helps to support us so if you want to support emerge go ahead over to the ai in business. Podcast on apple. Podcasts dropped five star review and share any particular episode. You light or any themes that you want to see more of. We'd love to hear from you. We really do want to focus on the community this year and you is a listener are a big part of that. So thanks so much for those of you who already have and otherwise without further ado. Let's hop range this episode. This is jeffrey with flex. Logics here in the a in business. Podcast so jeffer-. I wanna start off with certain difference between doing business with the data center versus doing business at the edge. I know that the hardware you folks are working on. Is you know what i think. Most people think hardware. They think about the big rack sitting somewhere in the data center the edges different the edges. New it's its burgeoning. How do you define the edge when you talk to people. Because i think people always think about those racks but it's clearly a blooming ecosystem. Yeah well azusa terms of different people defined differently but where we basically look at businesses any system. Outside of the data center there can be things like cellphones stations in verizon stations that are kind of in between what we're looking at your robots this. He'll cars field ultrasound systems in the field. So these are systems that are separated in well removed from today's yet. Okay and and obviously pretty wide. Berth as to what that could be almost any industry this this could be applied retail. You've got cameras energy you've got i don't know some some big turbine out there. You know generating some power killing in occasional bird. You know it's pretty pretty vast swath of of what what can imply does that. Broad world of edge cluster. In any interesting ways. I think industry would be one that makes sense. Maybe you can talk a bit about that but we also use case what you see edge sort of us for a certain way tammy how do you think about this whole new space. Those of us at home. it's it's sort of. It's new it's novel but how we want to break it up. Well we're just touching the top of the iceberg so we've engaged with a lot of customers see a lot martin's segments and they have different potential sizes so one obvious one is countered. There's cameras all over the place. You mentioned walmart's wells fargo's there's cameras today wired into servers in the back offices of these places and serve servers not in the data center and right now those cameras us recording video in case somebody shoplift something that got her the tape. Now they can add inference start tracking their stores checking buying behavior along the lines along the take to get through the lines things like that. So that's an application where you need object detection and recognition similarly when you're talking about robots robots moving around into distribution for a warehouse. They need to know you know what. Say iraq to put things on. What's the what's the person to make sure you don't get them. Yeah so you're detecting. You're recognizing them taking action appropriately. Same thing happens with cars so those are all object detection recognition models like yolo the three which do an excellent job of doing that. Which are people are flying now and we see medical imaging and there's many types of limiting there's crazy. Mri machines are much less expensive or numerous ultrasound. Machines x-ray cd stamps and stuff in between so there with the people are using models for his more specialized object detection recognition typically. You're standing your knee. It's stuck in there. It's not moving but you're looking to detect some anomalies in the x ray on the ultrasound. South is the baby. Ok is the got busted acl. So it's helping. The radiologist do occur. Job of diagnosing powell. That's what those kinds of models need to be doing. Things like Scientists gamma joining or life sciences. And they are in many cases what they're doing is looking to clean up images using network approaches through extraneous information clarify the. Which if you've ever seen like an ultra sounds like i just recently hasn't surgery. The doctors were trying to find a vein of the my shoulder. I can see the ultrasound. They were looking at a teaching hospital. And i couldn't tell what they said when they when they started a doctor they couldn't tell was on either but eventually they figured it out so these ultrasounds are hard to make out in. Computers can make better judgments which results in better outcomes says. A wide range of applications are seeing for inference models. I think we're just scratching. The surface gets more powerful and cheaper is going to go into more more more systems. Big time yeah and clearly you know The the hardware will also get cheaper with time as well. Cameras have gotten a lot cheaper. You know you're talking a lot about vision certain kinds of equipment opponents and obviously the core hardware like what you folks are working on. You mentioned a lot of vision. Applications the two latter ones. I would actually presume. Maybe we don't necessarily have to be on the edge if i'm at the mayo clinic in they're scanning me. Is that one of those kind of not necessarily at the edge ones like with those latter examples healthcare more traditional sort of in the data center or or do you see plenty of healthcare applications outside of the data center that maybe we can talk about as well while we're talking to people who are already declaring. Hey i in their systems so for whatever reason. They chose to have the the sanction inside their locks. Okay not connect to the data center connecting to the data center as latency. You need a network interface which has always be out and be reliable and it is charge money. It's not free unlike. Google search if you wanna use. Aws or something you're paying so their decision is at the right price is it makes more sense to deploy inference inside their also interesting okay. So you're you're seeing that yet. You know who knows when prices shifted use cases shifted workflow shift if more and more medical devices will not be billed as. Hey we're gonna pipe this over to your data center but hey we're actually doing calculations you just kinda jacket and you can see the results so it sounds like a and is possible that some of the stuff the days after sure we got. We got nothing against that. We see a lot about patients where latency is importance. If you're on a car and you drive in you know you can't be waiting for data scheduling on definitely. Not and if you lose if you can't say well sorry. I issue you know. The data center was available. So you clearly. Applications were real. Time is critical. Anders others were perhaps critical. We see lots of applications where the customers want real. Yeah so so. Let's talk maybe about that. Those factors that encourage shirt on sensors. By the way is i did mention a bunch of vision. But we see widar infrared x ray knight laser. Seen every kind of you know electromagnetic sensor you can thank oh used in various applications. So it's more than just your. Yeah and well. I guess two questions all tack onto a few things you said one of which again healthcare example struck me as potentially not as urgent as autonomous vehicles. And maybe that would be data center. Were but it wouldn't surprise me of the ecosystem evolves so that some of this stuff was just done at the edge. Those products might end up succeeding more. Because you know hospitals setting up the level of maturity they would need to actually pipe this into their own data centers might just be too much of a too much heavy lift compared to doing it doing it on board if the harsh cheap enough to do so. What are the factors. One of you mentioned was latency when you think about what's growing the demand for edge in other words being outside of the data center what's expanding that demand physical distances one. Okay if i'm driving in some obscure part of alaska with some kind of transportation vehicle. Maybe you know. I'm not going to be pipe into the data center says physical distance. Maybe you mentioned need for latency. If we need to make snap decisions we absolutely can't even at a lick of lag and that's going to be another factor that's encourage us to lean in the edge direction. What else do you see. Is those kind of magnetic poles towards making the bigger deal okay. Let's talk about rea- costs. It costs money to run a universal model. And i don't know what the costs are but it's not free so we've had he'd goal. Tell us back when we started into this. Remember we talked to own doorbells as they could recognize people but they actually had to have a little internet connection so there had to be data center connection than the images of the rebound back and they were looking for chips to replace that and put it into their the unit because as long as they were using the data center they had to charge monthly fee to pay monthly. See they were just want by the the on doorbell install exactly yeah so Sanders or wire every image. You're paying for it if you buy chip you pay for it. Wants to the next fifteen years so cost the dates actor you can do. And why use the data such as cheaply fast. Yeah but but is okay. You know in our world today. It's not black and white thing. Computers freezing uses land of local imputes ans- and they're using the data center. When you need to find some information you go to google and you find stuff on their data center but you're excel spreadsheets and all these things are running on your local compute. So you know it's not an either or and our and our regular life using both. It's done based thought responsiveness costs and other totally in. It seems to me like that same kind of blend is going to be natural. You know if we think about where. The computers being used in different industries. It'll be splayed out differently. Right if i run a company with a lot of energy equipment out there in the world. Maybe i'll have x. Percents of my compute that. I could technically categorize at the edge. If i'm a financial services firm just doing underwriting and accounting type stuff maybe we're going to be looking at vastly more in the data center so causes another founder. Yeah there's sermon intermediates. Which is i mentioned earlier. Wells fargo walmart's have cameras is cameras. Don't have any intelligence. Oh is that the wire going back to a server in the back room. And that's worthy capture. The image store them for later use so in that case. They're aggregating the were to into a server rather than having intelligence in the campus so we see customers who wanna buy our gourds to with in servers so than one server in control cameras. And that's the right tradeoff era. Still the ouch. But it's not all the way out to the actually cell sitting in each individual camera again. That exams hardware fit is going to be different. Good or you know you know. Some robots assembly lines in there might be eight robots robots controlled by one infringed. So you can do. There's lots of different trade offs making gifts those best. You know for their outlandish. Yeah again this is not a developed enough space where we have all the best practices right at some point x number of years in the future if i'm best by using completely arbitrary example. I don't chop that much so having a member here fun best buy open a new store and i want you know detect in inventory levels. There's going to be sort of a pretty tried and true orchestrated way of sort of cutting the mustard with my cameras. And having that steph setup hardware wise like at least a couple of cookbook ways of doing it. The a million other brick and mortar folks have used right now. We're feeling that out and like you said might be in. The server might be in the camera. Hey we're gonna we're gonna experiment in world define what's going to be right for the client application so so we have physical distance. We have latency. We have cost. I'm thinking from the perspective of the customer here when i think about. Is this gonna cost more to be done. I loved your doorbell example. That was a great example. Jeff was customer. Doesn't want to pay every month. I wanna sell this thing. I wanna i wanna outsell competitors by being able to even have an accessible price so i don't want to build my customers airmont. I want to have smarts but pay for it one time. That's going to help me grow my business. That was one example. Used where cost made sense that clicked in my head when i'm a business. Functional business leader in retail in energy in whatever sector. And i'm thinking okay. This particular use case is gonna be pricier to run in the data center or pricier to buy quota kokomo edge computing for lack of better terms. Of course we can't give anybody blanket advice but are there ways you like to think through this to kind of find those pockets where edge off makes sense one. Other thing to note in the data center is that data centre has way more power but the products that renovators interupt mice run really really big models and they're very expensive. You know at the edge. He wanted much less expensive solution. So it's not necessarily clear that the products in the davis center will be very good at running edge applications in in the edge. They're running models have billions of weights. I sorta in the data sets. They're running models of billions of ways for running models that out sixty two yearly weights. Which is a lot in. Their chips are optimized for large batch sizes familiar with that term in a data center the others thousands of servers so they can aggregate a whole bunch of somewhere votes in on the parallel and they combined a unison. They crossed the sixty four images at the time at the edge images. Coming in from one camera one at a time and you have to cross the on fly. So it's a different kind of inference that you need to do with the edge than what you do. The data the data center solutions all very powerful are optimized datacenter problems. Which are different from the problems. Got it that's an interesting distinction is well and really makes me think here you know as we go into the future will there be data centers in almost. Certainly the answer's. Yes but neither you. Nor i know the ratios. I'm sure will be different for industry and geo region in everything else but will be data centers that have entire chunks of them built to process much more limited sort of number of features for particular kinds of problems to be more energy efficient. Maybe there will be you know soon. So hey this this chunk of the data centers running through this stuff because we have a lot of it but you know we don't want to spend as much money but as you're saying right now right now they're not optimized for that they're not optimize fredge problems and so that that's part of what makes the cost arguments or strong. You guys are in this space and you're coming out with new chip in the world to sort of operate at the edge clearly for you guys. This is a bet worth making in the edges going to bloom. You know you talked about. We're just at the tip of the iceberg any quick closing notes for folks who are or wondering. Hey what's gonna make edge take offer. They're going to be transit. They're going to be some sort thresholds. We're gonna cross where we're really going to start this snowball of edge being more and more of the compute ecosystem. Any anything you can lead people with today is the solution was. The people are using now for established leaders work but they don't run as fast as they cost too much so the edge market adoption of is still relatively august. People are predicting. The market's gonna grow to ten billion dollars from a half billion today but the reason or the way. Marcus grow. Semiconductors is you've gotta liver equally good performance a tenth of the price. And that's what our new does. It weren't alluring performance this like the market leaders today but a fraction of the cost so that will enable people not just to make their current alter sauces better but to put us into systems where they can't afford to quit quality high-performance inference today because it's too expensive. And that's been expand the market dramatic. Okay so for you. And i again can't make any judgments on your particular product but it sounds like forces in play here are as the use cases expand become more popular. It's just gonna become evident that the cost factors are just gonna hold us back from actually adopting things that we know are going to work in the industry and so we're gonna just have to jump to the edge. There's going to have to be a lebron. In people's way of thinking and managing their harbor eighteen analogy is at the android audi the options for giving a computer maserati and a mercedes benz. Okay got kind of and we. We'd all like Mercedes so there's lots of people who have enough money to buy it toyota or got it and if you can give them a good product at a much lower price point and gives you good performance. Maybe not quite as good but almost is good but it a lot less price all of a sudden a lot. More people input high-performance products. And that's what we're trying he had. It would be far far less cars on the road. If it was only maserati is available. The other analogy. I'll use here. Is that right now. There just aren't that many roads on the road The the united states is pretty pretty well rooted out the ecosystem of. What can this stuff do that will actually deliver value is actually not all that. Well wrote an outright there isn't a clear cut. Hey we all know how to solve fraud at checkout. Hey we all know how to do facial recognition efficiently. Hey we all know those roads just don't exist so these playbook need to develop that people actually have for things like what you guys are working on but luckily use cases are not stopping anytime soon. You guys are in an exciting space and hopefully for those of you listening in some jeff ideas about what's going to get this to tip and where Edge can make an impact going to be useful for you as you think about your own business to jeff. I know it's all we have time on this interview but thanks so much for being able to join us on the show down. That's all for this episode of the ai and business podcasts. Thanks for listening all the way through. I hope you've enjoyed this particular episode. It was kasaka research. Who put us in touch with jeffrey. Awhile ago asako runs a number of ai hardware summits. Kazakh was also worked with in the past about letting people know about the hardware summit. But jeffrey was actually a connection through them. So i wanna give them an extra pat on the back and a big thank you for introducing us to somebody smart who made it on for another episode here on the program and definitely check out kasaka research if you're interested in learning more about hardware otherwise stay tuned right here next tuesday if you want to hear more about use cases because that's what we do every tuesday here on the i and business podcasts or look forward to catching you that.
Text Analytics and NLP in Financial Services - with Ram Sukumar of IndiumSoft
"This is daniel fidel. And you're listening to the a in financial services podcast. There's a term that i don't really liked using because it's awful played out but the term is dark data. Maybe five or six years ago in the podcast. We had people actually saying the term dark data. Turn into a bit of a buzzword for a while but there is some truth to it and in the financial services universe. There's a tremendous amount of dark data from microfiche to physical paper to absolutely unintelligible documents and images stored in various and sundry places within a a banker financial services institution. There's a lot of info the cannot get to. We have to manually pick apart and look into so being able to do text analytics and apply natural language processing fluently to get value from that. Dark data is a big deal for business helps us be up. Operations helps us automate certain workflows and can even able entirely new capabilities. We speak this week with rob sukumar. Rahm is the co founder and ceo of indian. Software indian is headquartered. In cupertino with the preponderance of their workforce in india. I interviewed romm from india for this episode and he speaks with us about particular workflows for text analytics. What does it look like in operation. People get it wrong. where can it fit into play. And where the potential value for textile when it to ideas for use cases and when it comes to the practical realities of what you need to expect when you're deploying these technologies. Ron provides some useful guidance that i hope will be helpful for all of you. Podcast listeners. if you're interested in more use cases and if you're interested in more best practices around ai adoption in our than checkout emerge plus. It's e. m. e. r. j. dot com slash p. Want this is a resource explicitly for enterprise innovation strategy leaders and consultants and advisers. If you need to bring ai to life if you need frameworks and best practices for finding roi if you want your fingertips on our entire library thousands of use cases than checkout emerged plus it may be a useful resource for you if you wanna take your insights one step further again. That's e. m. e. r. j. dot com slash p. One that's as and plus then the number one without further ado fly into this episode. This ram with indian soft here on the in financial services. Podcast so from where i wanted to start us off is on extraction. I know text is your world when it comes to and obviously a big companies whether it's healthcare financial services. Whatever are struggling with texts that they then have to put into their system. Somehow if we just talk about financial services as an example. I know you work there with that process. Look like now you know we get invoices we receipts. We get paid performs. What is the manual process to put that in today dan. Different companies are adapting to this differently at the stage there different levels of majority some just use a large bdo coasting centers to be able to process these documents and use a manual process. Some use the a semi automated Process be maybe a combination of some most often. Don't automating dimension. I think from what missing is an of course. There are no of some emerging that the are trying to democratize this whole extraction process. I think the challenges. The complex city of documents varies from company to company b. obscene With pumping night from new york to singapore seeing a but idea of documents with different complexities. Different types of labels nestor tables of dubai off beneficial data types of data are different. So i think where we use. Our technology is solution. You're right yes. We we think of on gordon text in on capabilities deployed a goal competence in solution around solutions. And we're seeing increasing use as we obviously use some of for example vision uses in august act. Which isn't we didn't apply about more like cnn. Seattle and you know. And then we limit these using tensor of on a lot of by libraries us also use ns dm a algorithm are deep learning to them so he is a combination of these to actually find the boundaries of the devils and look at the federal data within the documents and then extract them and convert this into of meaningful information for the enterprise. This big different resume bank statements busying invoices probably among the some blow ones that got invest in mestre analysts reports. Who have been. I mean people. It's difficult to go through a full debate. The has one extract and somewhat so seeing an opportunity where we used both out. Solutions will extraction and somebody's them for the analysts to be able to go through in a shorter span of nine reporter company. Let's say reach so. I think we're seeing several. Use cases added up and some of these deboning out the dems and these can be applied and benefit could be improved the speed at which the documents can be processed the accuracy at which it can be processed think that the benefits are are many. I think will yeah. That's the promise the promises we can do it faster. We can do more accurately and we can have data. That's in a form that we can search we can learn from. Maybe we can even train algorithms you whatever. The case may be so You know we a cleaner data ecosystem faster process. I think that's the golden dream of wet. We'd all hope. Ocr would turn into you know from from the folks that have been in this space for a long time. There's also a lot of data that's coming out from weiss text so extracting human that information so being the legal industry being the medical industry. So there's a lot of contacts data. I mean it's just not the medical terminology the legal terminology so there's a lot of context jewel loaning that needs to be done. So i think they're auckland industries that where we can text extraction ordina bo financial services committee claim processing where they wanna use it for violating accuracy of many claims. I think i think the use case is clearly they are. yeah extraction. Is you know applicable anywhere. We need to extract texts from with it. Whatever whatever it be an image ugly looking in word doc. In store it somewhere. I like to use individual examples. Because i like the listeners to have a mental image of the before a mental image of the after. And that's so Without a conceptual grasp very very hard for people to make the right strategic decisions about these technology so we we just think about if we hang in finance for a second. of course we've got we've had nuance on the program. I mean there's jillian players in healthcare and the other domains here but if we just hang in finance we think about when this information is being kind of pulled you mentioned. Vpo that's a business. Process outsourcing firm often. They're going to be in india. Maybe the philippines these are people who are gonna read these things they're going to copy paste the right sort of data or manually type the right data into the right form fields and then push enter and then it's all going to be sorted in the actual system the right way. Imagine some ocr systems do that as well for most extraction problems Is there is there sort of a a big custom. It feels because we have to be able to find how we're gonna to store. This right at every bank is going to store things different. Every insurance companies Store things different. What exact fields. What exact format so it feels like. There's a bespoke new from the get-go about hey you need these invoices to go in this way. You need these reports to be put in this talk a little bit about how that has to be customized tweaked. Hit on the name that yes. That's one thing that we're seeing this. There's no one size for. Why would we have done as but it. All these algorithms than the. I jin to be able to act this process or for it but linda. Financial services industry so by side in a body missed out aboard board in our bank statements in so many different types. And it has to be customize. The twin we use on services capability. And where we have an expert does where we've bringing data scientist to be able to customize this and the way this is is consumed by enterprises also seems to he. Did they want the back in the of it wanted to become. Api's some of them. Just want them back as in the former consultant file format. Devil the in use within their databases and the with their teams to consume. So how do fermentable prizes consumed is. Also we see a lot of billion so it's difficult to do one size fits all solution. But when i would say i would fifty percent sixty percent of this as being a lot of commonalities in water the models we're using and the something of training and that we need to do to that particular use case and then the final. The output is how each customer also uses the final extract. information is again customized. Yea because it's everybody's business process is different as like. Oh well for us. We take these documents. We use a summary. Somebody reviews at this level and then they go into storage this way and then another bank might do a totally different thing with different orders in a different process. And maybe that has to do with some legal concern. Or who knows but you gotta find fit for for all of those. My guess is that over time when it comes to certain kinds of docs there might only be like a set number of categories of things you would do with them like one of them would be. You said well. We need a simplified form in this format. That's one like Turning up a piece of paper into another piece of paper that simpler another is we need to take these seven. Bits of information for all these. Docs you know who's the client. What's the amount to be paid. You know whatever and we need to enter it into these fields and push enter in push that into a database. So it's another instance. There's probably only so many broad umbrella categories of what will we do with this document. I my missing any there. I mean i've touched on too. But i wonder if there's any other big ones aren't the those pretty much covered wants man. There's also this whole volume is not a question that comes up their customers with this one time just as millions and millions of that to the somewhere in almost ninety of documents but the volume so far less so i think there's a idea of So so the extraction. Challenges wade's consumed. Also we've seen radian some like it on the cloud where we thank to code options gov vote volume is less than their some one and on prem of and where where you know. It's it's managed as a as an enterprise solution. So we we think the way concord loma use these vide- significantly. But but yes. I think to give bond that. The gm challenges that the customs is facing. This yeah i am familiar with kind of the bpo world. We've actually for our market research work with a number of clients in the bdo space. Where trying to now get into actually automating these processes fred because they realize if we're just a houseful of humans there's gonna be limitations there my personal hope from your in india. My personal hope is that a lot of india wakes up to. Hey what if we could climb a little bit upstream in keep some of this value as opposed to just have the lower price value proposition. Which i don't think it's gonna last for another ten years so i'm i'm rooting for india in a big way to do a lot of what you're doing here now south But i'm very familiar with that world. The other approach of people are trying to take today and we'll go a little bit more into where things the state of the art is more old school. Ocr approaches when you think about what ocr used to be eight or ten years ago and then you think about the state of affairs now the different combinations of algorithms what to you is. The main of improvement uptick there. What can we now do. That was particularly hard with with older school. Technologies for extraction. Jails come a long way. I mean these are some of the that we have seen. People grapple with today's technology options to be able to use some of these algorithms at scale itself is is. I think it said think there's a lot of our technology is is at play there. And i think even look at her doesn't act a just an open source yada engine. I think there's a lot of majority. That's that's gonna go. that's not the end. I mean our businesses to take that. And of course us i on top of that because that just gives you. The was your thing to be able to do that. Scale thanks to a goss and the technology of the options that are available today a long way from the seattle not even five years ago I've seen some of those yard. I don't think they can work at scale than that could. Solutions are are able to and for example. Let's say our solution solutions on google's cloud we have an instance for customer. We have a customer process. Saying let's say a million records a month so you're gonna be so we're using some of the open sociology within that and just to be able to even who that the offer that kind of scalable solution letting creditable when some of them the policies have prompt. So yeah we. We've got more compute to just crank documents through we can serve. Deploy a lot of that compute up in the cloud instead of having it sits wherever the heck it used to sit twelve years ago but also the capabilities of the tackle. Obviously more expanded a you talked about the contextual data. You know the different kinds of formats challenges here that you addressed in than a ton of work in the search and discovery space in familiar with some of the challenges where the format for x. Kind of doc whether it be an invoice whether it be a certain kind of report whatever they're gonna come in with variants in variety that a lot of the time we just don't have control over and to there's going to be a lot of this manual training. How is that adapting to different formats at adapting to different kinds of documents. How is that more powerful now than it was. Then what sort of allowed that staffing. Obviously that was the are available. But i think there's a lot of the become is if able to do a big thing. We want our was. You're able to see this. Handwritten documents are documents to a different quantity images. I think differently. Those taxes is significantly invoked in line. with other digital's. I think some of the models be able to even implement some of these bottles of by town libraries You know that has the michio it over the years back now. Tens of lower. There's a lot of these markov models that we use which also using in speech recognition and so many things that have been known for the while. I think the overall lot dick options available with some of these packages can be used in deployed at scale with higher quality and I think that's really has made a difference to be able to afford this. What we're doing today and that was able to do. It's five years ago okay. We'll get into now sort of what it looks like to deploy a into this process so back in the day we have. Ocr's it was which may be a little more limited in scope to a certain boundary boxes of documents that we could reliably train. Maybe not as much variety though and then we have the bpl where we can kind of. Send everything in you know. There's there's folks that aren't going to cost a ton of money in our to put that stuff into fields. Now we move to a world where we can start to expand what is machine dual and hopefully again faster hopefully better quality this dream right. It's not always easy on the first go. What is it. Look like to to get that process going. Let's just pretend you know. I'm a bank. I come to you. I say hey you know. Here's these three different kinds of documents that we just never been able to handle those cr. What are the steps to sort of start to integrate into that process and get the output that they're for typical probably gonna customers gamblers with a problem statement. Saying you know we have these documents and we've had Challenges using traditional cro. Whatever other debatable out there to be able to extract can take a look at this. The first thing we do is we. Actually we created and train our models to that particular document and an opioid is only one aspect of that which would contain to make to whether it is account information. No old side of the table. And sometimes you have negative tables that we have to deal with so we use audie i models and again wrote. Use a lot of machine learning foul better to be able to do this. And we've been customize the model to that particular use case so we have seen Right from a document that has a drawing and there were tables and then we have seen a document that had a lot of table in the textile and then we have a symbol simpler structure the table dot which is all the more easier wants to handle so i think each one presents a different challenge and i think i focus on extraction and then comes the context tool. That's where some of our summarizations capabilities coming in a lot of Solutions some video faction but we. We thrive in a complex. And we try. When there's more analytics that needs to be line is to be applied on that expected intuition so apply context to that particular information. That could say any misdemeanor. Banco might abbas needs in. His will be very different from well. Let's say a bank operation but it's no someone from many claim processing. These are all each one presents different set of context which we then train the the solution to be about. Everybody's a one hundred percent accuracy yoyo. Obviously that's that's a big and so we started the process. Where bit is somebody of mandolin. Dimension to be able to achieve hundred accuracy but over time as as a model london. Some of the You know mitchell them you know that kind of gets into a diamond becomes hundred. Listen i automated tosses. That's yeah that's the goal anyway. So let me know if i'm correct on the the steps here around because again i liked to make a visual in the minds of the listener the way i understand these technologies. And you'll have maybe a more granular way to articulate this than i will because you do it while i'm a market researcher And i really look at a million other things so the way i understand this is will find these new complex documents. Okay we've got images and then we got tax who got tables text on both sides of. It's very kind of very donkey to train. On in some cases it might be possible to ask the client. Hey is impossible to get these reports in the first place a little bit. better format. Sometimes the answer is yes. Most the time i would i would guess. It's totally out of their hands against so we can try. But then we have the docks as simple as we can get him which is often not great. And we've got a essentially do some some human lifting here and say okay for this document. This field is going to get entered in this place. These two fields. We're gonna exclude. We're never going to do anything with them. This this field at the bottom looking for these terms. And we're gonna put whatever we find into these two slots and then we're going to enter it into the database. I'm making up an example and so there might be fifteen slightly different variants of the weight of these docs come in. And so we need to take x. number of each of those fifteen in manually like. Here's the bounding box. Here's what were searching for. Here's how we enter it. Here's what we extracted and get some of this human training and then we'd get an algorithm to look at a new dock and say which of the fifteen is it and which routine and my going to run. It feels like that's the process at the end of the day. However again you're the guy i'm not so let me know. Know you hit it on. I think that's that's a nice way to put it for for someone to understand. Because of wesley. Customers valley change their document for the We wish thing that happened so so that'd be nice but but it rarely happens but i think Yes what we do is a human. Would i look at the document take visit documents. Okay got account information at one oppose manually do was is what we chain on algorithms and then that uses mlm deboning algorithms to be a bad combination of those on repeatable. Add scale As as as more documents come in and said pretty much what we do but but obviously not come to us to throw up a lot of humans at this problem so we be used that step only learn how to do it first time so that weakened and train out models be able to finish your best position so we're not customers come to us. They usually a customization and setup costs that we have a which could take anywhere from a few days to you weeks. And that's what we do in that. Yeah yeah yeah. I can imagine again that that time is going to be a little variables. Maybe sometimes they're just working with docs. That are fortunately quite simple. And you can just you get a certain number labeled examples and you can just run because you've seen a lot of similar stuff and maybe in other cases it's going to be a little tough so customers period. My guess is you and your subject matter. Experts on the feature extraction kind of side of the house are gonna come together. You're gonna look at dr. You're gonna edit docs. They're gonna give you as much as they as they need to tell you for you to manually score a bunch of them. Probably there's going to be a period where you're going to run a bunch of fresh ones through and you're going to work with them again and say how did we do and calibrate things where you need to so again. There's there's kind of this trailing period of training those normally always going to be some ongoing level of. Let's say quality control. Nothing's ever one hundred percent they're gonna come up with new docks. We have to adjust to the algorithm. I drift a little bit and start doing things we don't like. What is that long tail. Look like for are maintained communication with the clients and make sure catching edge cases. You know keeping things improved controlling for quality. Can we go through that a bit again. So if you look at the solution until we of these demo of textile so you long to do you have one of the document you're done with the customization starting the process. So there's something called of be able to be processing module when there's some degree of processing that began then there's also a juicy module where we actually have to do random checks of some of the process data so i think that qc dimension is something that we're seeing is ongoing. This game of that depends on the number of document again on volume. And but i d could be just our few hours. Among could be several members mendez amounts. So davila depends on the volume an ideal but we notice some of your flop choosing dimension that needs to be there for in charge on a cent water. The tables over time as was the model machar's as learning happens. The machine learning algorithms dacca's he gets better and better and we always have a let me start with gusto must come to us with desert customization and then there's a proof of concept period where we actually doing this for the moment and they're seeing the output on the quantity and obviously they doing this without that makes their life easier so that's where we strive for as much as wanted one hundred percent and again if somebody of qc that we have to factor in overtime. Yeah what does that look like. So i imagine what it might look like. Is somebody at the client. Company is calling the database and when they see errors they click a little button. They fix it and then you have a dashboard of what all those quick fixes are. And then you you have. Maybe a call every month. They're however long to sort of get a sense of if there's any themes they've picked up on that we haven't to kind of uptick it. I just made that up off the top of my head. I don't run your company. But how does it actually work in in reality when what we do is we look at We have a dashboard that talks about number of documents processed. And then there's also some degree of huseyin that the customers do especially in the show face and then we have a dashboard. Wears they report any facts that we look at the okay. Where was that. Was that because of clubs kind of a we need to fix out of them dude alone. Them particularly the context data that we have had so far is something new we haven't seen at is again sped into the algorithm. So that even a minute happens in to joe. So i think we don't have it a click. Get something as simple as that but we do have a dashboard where the process they dow to present documents processed indiana's and then we can kind of build gazans. Then we end of the eminent extreme. Kind of what's on why that happened. And then we can along the model and and we mean chicken doesn't happen again so then deal it really depends on how long depends again on the complex at the end but idea of the documents but it can range from a month. Do even several months of continued juicy work total. Yeah yeah and i imagine you know in terms of marking the things that are ariza keeping track of those errors. The people on the client side have to know at least somewhere to to keep track of them is. They're often at champion. Who sort of heading up like this project. Disintegration this partnership with you and part of their role is sort of coming up with a mandate of look when you find an era. Here's where we're going to store it because clearly you have to have that data populated and somebody in their. If nobody's responsible for it than it's no one's gonna report error is is. It often like their champion there. Or how does that work or you know. There's every enterprise we're talking to them. No the bean and the need for this come from different buttons it could be customer service department could be operational and financial in may be a mess journalists team that has does need so the need bibi's rama. We have seen this usually not with his companies are now having these vote van summation beams abby beans so there's usually jambi looking at better and smarter ways to these things so it usually known as we're seeing that kind of champion was who's driving this working as single point of contact but they're also dime. Who ended this. I an event. It's mood of a support. This will a person najem this usually someone from the innovation team or someone from the internal operations was leaving with these things and challenges. Tempting cool okay. Good to know figuring out the you know you've got all of us have plenty to learn in the next couple of years as to where these things fit in. What's the right combination of our talent in their talent to come together girl you know. Make it work as quickly as possible. Those lessons learned in my opinion as hard if not harder than the right combination of algorithms. It's cool to hear your take on that. Because i think you're going to be learning a lot coming up here. It'd be week being finding different. Use cases different customer. Needs so call while i know that we went a little bit into overtime but ram. I think it was completely worth it. We some fun stuff. I think we've got a great mental picture of the before and after sincerely appreciate you sharing your insights with us today. Thanks so much. Thanks name and w you join the conversation and appreciate all the good stuff. You're doing any minds. Enda trying to create these awareness about use cases in the i n. We'd love with the future guests. You bet remained. So that's all for this episode. Thanks to rama for sharing his insights here. Thanks to you for listening all the way certainly appreciate having you. Here's a listener if you enjoy what you're hearing that be sure to follow us on social to find emerge at at e. m. e. r. j. on twitter emerge artificial intelligence research on linked in or on facebook. We've had a growing social audience and more engagement than ever over the course of twenty twenty s. We've become more active on social channels. Sharing our frameworks and best practices. Charing our latest interviews and also sharing all of our latest articles as soon as they come out. So if you wanna miss a thing and you like use cases and you like best practice information than follow us on social at emerge otherwise stay tuned next month for another episode here on the financial services podcast.