4 Episode results for "Pete Haas"

Amy Stapleton, Dave Kemp, and Pete Haas Discuss First 100 Episodes - Voicebot Podcast Ep 105

The Voicebot Podcast

00:00 sec | 1 year ago

Amy Stapleton, Dave Kemp, and Pete Haas Discuss First 100 Episodes - Voicebot Podcast Ep 105

"This is episode one zero five of the voice by podcast Amy Stapleton Pete Haas and Dave Kemp join me today so we can discuss some of the most important insight shared by voice by podcast guests in our first one hundred episodes. Hello everyone out there and voiced by Land Brechin seller your host for the voice by podcast. We're having a little fun this week with a panel discussion about the first one hundred voice by podcast episodes so no this is not Meta voiced by podcast. There's been so many rich insights offered by voiced by podcast guests that it wanted to figure out a simple way to unlock if you those nuggets in case you missed the episodes I also wanted a vehicle to remind listeners about what has been said you may have heard the insight right the first time but if forgotten it's not top of mind this is a way to bring it back up. I also wanted to get some discussion going on some dedicated and erudite listeners about what was said by pass gas and what they thought mattered most so with one hundred episodes to choose from we narrowed it down by asking listeners like you to let me know their favorite episodes why our panelists also weighed in and by my count we discussed today parts of at least twenty three episodes that means it over three quarters. Were not even touched I will any of your favorites included. You're about to find out is Amy Stapleton Tells Pete Haase from bear escape and Dave capital Oaktree join me on the one hundred episode retrospective. What a long fruitful trip? It's Ben then but before we get started. I have to do a quick shout out to this week sponsor Trinity Audio. Do you love voiced by podcast. Do you get a lot of value in insight shared by my guests each week. Thank our sponsors on social media and visit their websites if you do I'd appreciate it so twenty audio enables publishers to turn readers into listeners by transforming their written content into lifelike speech. We have voiced first interfaces today and that's leading to more demand for audio content Trinity Audio's helping hoping fill that demand with its contact. That's content technology solution. They characterize it is audio fighing your entire website or contact catalog studies by the company show the adding audio increases website engagement by five live acts and it's free so to learn more good at Trinity Audio Dot A._I.. That's Trinity Audio Dot A._I.. T. R. I N.. T. Y. Audio Dot A._i.. Check out the trinity player for your website. I trinity brief for creating a type of flash briefing from your website content in trinity skills that make your website content accessible through Alexa who even know you could do that so checkout Trinity Audio Dot A._I.. And if you get a chance send them a tweet tweet at the Trinity Audio you can tweet at them at the Trinity Audio and thank them for supporting independent media like the voice podcast all right now for some insights mind from the commentary of people like Adam Chair Cathy Pearl Brad Abrams. David's Bisky Liza Foulkes V._J.. Balasubramaniam Ron Krone and many many more. Let's get started okay voiced by nation. We have a special episode for you today. We are reviewing the first one hundred podcast episodes of the voice about podcast. I know many of you joined recently. Let's say within the last year and you've not had a chance to listen to all of the episodes episodes so I have three guests here that have listened to a lot of them. I won't say they've listened to all them. Maybe pizzas and all of them but I I don't think anybody is listened to all of them but they've listened to a lot of them almost all of them and they're I'm GonNa tell you some of their favorites some of the things that they took away and then allow you some of you will be able to skip over because you'll get the main point others. You'll say oh that's a topic I wanna go much deeper on so my guest today our first up Dave Steve Kemp Dave. Can you introduce yourself to the voice about audience. Yeah absolutely thanks for having me on Brett <hes> so like you said my name's Dave Camp on the business development manager at ultra products. We are a distributor of Clinical Oh supplies in assistive listening devices to the hearing healthcare professional <hes> so that's my day to day job and then I got real immersed into this space the whole voice space and started a blog called future ear fugier dot co <hes>. I'm started to write a little bit about what's going on in the voice base how it impacts the in the year device computers that are starting to kind of emerge herbals <hes> wrote some articles for you over at voice about as well so really excited to a chat about all things always spot today great. Yes I think we have maybe a half dozen here. ables articles from you and those are very popular. It is a topic that people really care about is tightly integrated with the way we think of voice and mobility going forward. Take a next up. We have amy stapleton. Amy Is not new to the PODCAST. She was with us back. In episode eighty four. I think <hes> amy introduce yourself place to the listeners or reintroduce yourself hi. I'm Amy Stapleton. I'm the founder of tells and we create conversational story experiences and we also provide a platform that lets independent authors contribute their original stories to our existing voice apps so example of that is our my box of chocolates skill and we try to publish a new box of stories every month and those stories are by different different authors each time and <hes>. I think you said you wanted us to perhaps mention the first podcast you're listening. Tell me the first one you heard. The first one that I remember very definitely is episode number nine with Joe Do Quinta yes yup. Good Ride Jos Jos a well-known developer in the community and he <hes> he really was very helpful and I think for a lot of the early listeners talking about his experiences building games M._C._i.. That was a good one dave. I forgot to ask you that question. Do you remember the first one you listen to. I was just going through my podcast APP to look and see looks like the first one was <hes> Yon Caning with job. Oh Oh yeah that's that's back in the day to. I think that would that would probably are I tend somewhere in that range. You guys get you guys definitely go. You started early so that that's a good thing yeah yawn <hes> build a Jovo which is a like a cross platform. I'm open source publishing tool <hes> which is really cool and they've done really well since then. I've been happy to see the growth there okay third up. We have Pitas Pete. Please reintroduce yourself to the voice by podcast audience. Pete may be familiar to <hes> some of you. He was on as a guest in episode seventy where we talked about the bixby launched last fall pete say Hello Hey guys this is Pete House and <hes> I will say. This is my favorite podcast. I listen to a lot of them so I am very glad to be here. <hes> I bid. I was thinking about this because you're gonNa have to introduce myself doing account. I think I've probably since two thousand five written probably a couple hundred absent only use voice start a Ni- V._R.. Space but more recently maybe two or three years writing for Alexa for Google and most recently for Samsung big outstanding okay wait am i. I they may first episode was episode six with John Kelly Oh John Calvi yes of bespoke in talking about acting and some of their development tools way back then that he was really focused on yes Johnson on a couple of times and I think it might have again on on this this summer. I gotTA reach out to them because what we've done is I think I've got him the last two years and so it's for a few the folks that I've come back to. It's nice to see their evolution over time and what they've been able to do each year. Okay so that's great. I think what we're GONNA DO <hes> folks for the listeners will follow along. We have a number of episodes that have been submitted by the panelists is their favorites. I think everyone submitted three to five. I threw a couple extras in there just for are fun to talk about so we've got a few different formats but we're going to start going through them and I up is is is a podcast episode that it is distinguished by being the longest we ever did. I can't remember if it's an hour forty five or an hour fifty but it's with Ron Krone. It's episode ninety four Ron for those of you are familiar and haven't listened to that was the original founding C._E._O.. Of Nuance and it took so long it was such a long episode for a few reasons one is interesting listing insightful guy but also we talked about nuance the biggest voice application company around three billion dollars in annual revenue. Just in you know selling voice software <hes> talked about his origin of identifying the technology building the team getting spun out of S._R._I.. All the way up through all the different milestones they did and then how more recently he's come back into the industry and has been participating as we're hitting sort of a new evolution and and after being out for ten years he took the company public chairman for a couple years and then started doing other things and so it's really interesting to have him come back and to give his perspective and how voice had changed and that's that's why I was so long and I really really enjoyed sitting down with Ron but I'm really interested P. Why did this one stand out for you? I am a sucker for good origin story so anytime those those come on for somebody like Him who's certainly industry veteran <hes> I was all over that one I actually back in the day was using new onset rhyme V._R.. Platform and scans off in this thing so I I knew about that world <hes> so this is fascinating to see him for from really nine technology background come into the space identified big opportunities and was really laying the groundwork and I know one of the questions they ask is. How do you take this from kind of being a science experiment to a commercial product in at the time? This is unthinking ninety four for that was in question never really been done before yeah they they were involved in a lot of firsts and one of the things I do like about that is there's a lot of people who are new to the industry and that's fine because that's that's what things have changed the platforms different capabilities different but it's always nice for people ever perspective on where this all came from amy. Did you have a comment yeah. I just wanted to say I definitely remember listening to that episode. What is well I know is I had to visit my mom at the time and I was listening to it as walking through the park in it <hes> was a was a really interesting long episode? I like the fact that he talked about. You know the the evolution of technology and then how nuance aunts was not the company that had bought what was the scans off scans off was the one that had actually bought nuance but they decided to take the nuance name which was interesting <hes> but I don't really remember exactly what <hes> what Ron's take on things today is did you can you kind of sum that up because I'm trying to remember like what is he doing with the Lord Mr Jones stuff whether yeah so first of all he's an investor now and he he uses Alexa he was intrigued by when it first came out more intrigued he said than he was with Siri when that came out <hes> and he's never seen a reason to use another smart speaker which is interesting because you think about some of the things that are out there now you know he probably would have been playing with all of them fifteen years ago by intrigued by them <hes> but one of the things he's really interested right now is this concept of artificial humans and and and now his I think the last twenty minutes of the podcast we talked about that as well and that's this idea of it. It's not it's not unrelated to voice assistance. It's this idea that you know we take someone like a celebrity. Maybe we we have an example of this like John Legend to voice for some interactions on Google assistant not too long ago but it could be this type of thing where people like interact with with that assistant in it sounds like John Legend or it sounds like some you know <unk> perfectly newly created solution and it has a personality and you can actually have a full interaction with them and some of the things he's investing in not just the voice assistant but that's part of it right. That's the brains behind it and the voice but then there's also avatars and other types of manifestations that can that can come out so I thought that was really interesting. A lot of people have commented on that part in particular. I thought you know be beat. Beat starts out with the origin story and I love. I love that part of it but a number of people on Twitter said Oh Jeez I love that part about the artificial humans at the ants glad we got to that okay fine Ron Chrome. We gotta keep moving along number. Two on our list. Today is episode ninety so again. We're still in the nineties. A couple recent ones in this was from both amy and Pete and it was about was Steve Arcana Vich so amy. Why don't you comment on why you put Steve Arcana Vich on who is best known for the Alexis Skill Big Sky? Yeah I thought that was just a really great <hes> introduction to to an indie developer and he was <hes> you know I know he's a very interesting. I mean he's very accomplished. He's a philosophy professor. He talked a lot about how he uses he thanks through precise language and using using his background as having to write clearly to help him articulate things in his in his skill experience which was kind of cool <hes> but you kind of see in him an into an indie developer that started from the very beginning did not have a background in coding taught himself how to Code and then found the perfect use case which is weather and at the time and even today the native weather voice APP is not that great the native experience so he got in there he did something and he was able to really leverage that in monetize ties it and I think you know one of the few that found a use case as an independent developer. That's really got something going there. I mean you know whether an an ambient sounds those are the those are the places where I think an independent developer right now can still <hes> or could in the early days gained some traction but you know now. The question is what what's going to happen. If someone for other ended in India developers now it's kind of harder to get in and he he was very forthright. About how many people are using his experience <hes> the the <hes> the ability to monetize it what he's doing to monetize it how much it's costing him to get like some of the radar information so it it was just very interesting to hear his perspective yeah. He was definitely Kelly an open book. I love that about it and he is. He is a very interesting guy. We had a really good conversation because we're both interested in the space. You know a lot of that was just sort of normal back and forth when you're doing some joint discovery together I love the conversation about him. Competing Against First Party <hes> the native weather weather APP and what that meant for him and how you can still differentiate even a first parties got some solutions in demonisation and some of Viz user numbers. I heard a why of a lot of feedback from listeners really thankful that someone was saying okay. This is what a successful skill looks like on the platform now pete. Did you have something to say about that. Yeah no oh that was a very interesting one especially as a developer in coming at it from probably a similar point of view but it was a little discouraged because he says he's not even a developer and he's writing in Ruby which is some developers are listening using now will will will understand. It's typically not how it's done but they have such a you know the hit hit skill and and to be so successful is really fascinating <hes> and he was really open about sharing that so that was just great here yet is funny. He built that ruby framework. I forgot about that <hes> because his the first developer he hired new Ruby and basically did some sort of weird translation with probably Java or python at the time yeah I mean he probably hired a web developer at the day at the time that was ruby yeah and then so then when he wanted to maintain it he had to teach himself Ruby Yeah. That's the not the conventional way to do it but I guess he's sort of an unconventional developer. Okay third up. We have episode ninety seven Amir Hirsch a he is the founder and C._E._o.. Or Co Founder C._E._O.. Of audio burst and Dave camp this was on your list at the top yes I I thought this was one of the most fascinating episodes that you had had <hes>. I mean this whole concept of <hes> cataloging all of the different <unk> audio <hes> sources out there starting being with podcast and radio. I think is it's paramount to the future of really this whole movement in the reason. I say that is really. I think the killer use case we keep saying what's the killer use case for this and what's the use case but I think it's pretty much staring us in the face right now. <hes> it's audio. It's audio consumption and <hes> you know I think that you look at the the rate at which <hes> just audio contents being produced with all the new I think I heard on that episode. He said something like seven million episodes are currently out there across hundreds of thousands of different <hes> podcasts and so this idea of being able to say <hes> to your assistant. Dan You know Alexa or Google Assistant Tommy about Tesla for the day Gimme Gimme an update on what's going on the Tesla and rather than just feeding you like what's what's trending on business business insider or something like that. It's actually feeding you. Snippets of different podcasts that people are discussing and I just think that's such an important component to this idea of ultimately getting to the point where you have this curated feet of audio content. <hes> that is real personalized you where you have people that are <hes> you're not even exposed to but your assistant is starting to learn based on your behavior that you might like it so you're finding all kinds of different content and I just I think that what Amir's doing audio burst is going to be the foundation in the backbone by cataloging hall yeah so I really like Amir story at a couple of different things I like about that. One is it's not just about Alexa skills and Google actions. It's a different way to use the technology in this idea of making audio content discoverable also as someone who creates audio content. We both do that. <hes> you know that's obviously attractive but I think if I think about the value proposition for voice assistant users. I think it's really important that we think about how we can how we can generate content. It's not just reading off a wikipedia page just a bunch of research on this like yes so many of these queries you get some sort of Wikipedia <hes> information or maybe news from like a print new source when there's all this rich rich audio content out there that could be a better could be a better solution. The only thing I like about is like a big idea. It's like to try to the scale of Google did exactly it's Google in this space is kind of a takeaway. I had different amy what what did you have to think about that. I just wanted to say that on that episode to me. That was a really good example. Have how you do such a fantastic job of like breaking down of very difficult technical subject so people can understand it because I remember there. Were you kept like drilling down and asking specific questions about how the technology worked and he would explain it and that's just you do a really good job of like ferreting out information so I just wanted to kind of bring that up because that was that was a great example of it okay well. I'm glad I'm glad that came across we. I'm I'm basically asking questions I want answered to the extent that that's helpful to the users our listeners so much the better okay so I want to talk dog next about what I call the design block and this is several episodes so let's see what I have on my list here. I've got episode thirty with Kathy Pearl Episode Thirty Three with Mark Webster Episode Twenty eight with Tim mcilwraith. Oh I'm sorry Kathy Pearl. She's now with Google was essentially at the time Mark Webster founded. A company called say sprang which was sold to adobe last year may was two years ago now it was it was a while ago <hes> Tim Maccari <hes> is search in the leadership over at discovery communications so discovery network those types of things episode forty as well Karen Kosinski who is an independent voice in solution <hes> user experience design consultant <hes> I would put Amy Stapleton Episode Eighty four in this group as well as Lisa Falcon episode three and Emerson sklar episode sixty six. I may have forgotten <hes> some of these episodes. Sixty six was actually Schley a compilation around design. We've really done a lot of voice you acts or voice design episode sense and because we hit so much content <hes> the one I did with Emerson where we I think we we went through like twelve different episodes will snippets snippets of twelve different episodes and in feedback people had around design but I want to throw this out there <hes> in and get your feedback on it because early on design was so important <hes> and there there were just so little information about design out there and voice user experience design in terms of the general conversation of how voice assistance we're going to be different than I._V._R.. For example or from chat bots which there was plenty of information on a couple of books and I just found that the perspectives that each of these people brought to the table so helpful in understanding often what not to do when you're designing experience and again some of the things ray practical applications of things you can do due to move forward in in in that some of that we've learned a lot about just the history of a- voice interfaces as well because so many of those people have worked for years in this space so who wants to jump in on this one. I got a couple of things to say about that <hes> because I I've lived in a world where the only means to interact with an application that I've written is just through voice and so there's just been some principals that I've always abided by one is <hes> if you're going to enumerate a list of items once you'd be on three. That's you gotta think about it a different way. It's too much cognitive load for the person <hes> so for sure. That's what one constraint that you would necessarily have with the screen on the other one would be internally generally. We always had a rule we call it. The one breath principle which was we would never <hes> spew the script of information that would take a human more than one breath to speak and if it was too long we would we would try to break it up in a way that was allowed the customer to the person using the application to Iraq with amy why I just always find it very interesting and helpful to hear people talk about what they've done so as an an example you know the <hes> Tim what's his last name the guy that from the Discovery Channel mccaw yeah the <hes> the the person that did the the <hes>. He's the the food network one with a new recipes. You'll yeah that was very interesting because he went into a lot of detail about <hes> some of the choices that they've made how they were leveraging the screen <hes> the philosophy behind multi-modal. I would also throw like adv eleven in there. She was another one that talked the the design concepts of her <hes> kids court so it's it's very helpful to hear what what thought process p people are going through and then what kind of decisions they make in their skills. One thing that kind of is striking me today day is when I listened to some of those older episodes about about design and and and conversation you I. I'm wondering what's going to happen now. That Alexa conversations is out there. I still don't understand it completely but there was a lot of discussion or thought around the there there are limitations to the platform now that are holding it back that people aren't really able to have these these conversational experiences the the the platform is more designed for transactional type activity but with Alexa conversations that is that going to you know. Is it going to be a completely new new paradigm so it's it'll be interesting to see what happens. Yeah Please Complete New Paradigm and this type of cross skill L. Experience where you're going to have different designers for these different skills being able to help with this fulfil this user need <hes> and how does that done without creating some discontinuity. What do you think Dave Yeah No? I I agree and I WANNA kick it over to you. What so kinda going off of Amy's question because I I was thinking the same thing? What do you think with Alexa conversations in your opinion? How does this change the whole equation in the most front from you know what they're doing today and how it's going to transform it well? Actually I've got a pull pete into this. In the second because Steve Our candidates was the first person that I heard sort of point out and said Oh this is going to move us to imperative programming model from declarative and so you know declared of being you very tightly defined all of the user interaction action scenarios whereas imperatives as to what the goal is have general constraints but then turn it over to something else to fulfil in this case the I will fulfil some of those and so there's I think there's two different things when we think about Alexa conversations one is the fact act that the I will do more and then the second is is this other things because there were two announcements by Amazon in early June one was Alexa conversations which is really this more imperative programming and leaning on the A._I.. For more capability and the second is is this crosskill Gulf film and when you have an objective that one skill may be can't fulfill you might tie several together and so I think the answer is different for both <hes> but I do believe that this is more more similar to what bixby already has in the market and Pete since you've worked in bixby. Tell me if you think that you know my high level assessments correct and what your interpretation of Alexa conversations in the cross skill Gulf filming us yeah when I heard that announced I kinda I kinda smile because I thought boy that sounds awfully similar to the way bixby's approaching the problem in you know from somebody that came into bixby last year. I looked at that platform end. It's it doesn't resemble anything like the others right so if as a developer <hes> you basically create a Web Hook to respond to really you know there's intelligence on the Google side or the Amazon side and they will give you a in some always a dumb message will just say this is the intent and please fulfill this request <hes> if you look at the way others like big spear approaching it <hes> I it's really an A._I.. Platform so you get other features as a developer like <hes> <hes> preference learning like selection learning that she really don't have on those platforms and the way they're these deconstructing it so to say you know we're GonNa Maybe get away with these invocation names and change the way that's interacting. I mean to me that's that's the way to go. That's that's the way to scale it because if you want to have an ecosystem with many third party apps you can't you know have these implication names in trained people to you know once you get beyond five of them. You're kind of dog hire pages we can only we can only remember the the icons and are the mobile APP icons on the home screen. Maybe once green to the right swipe this idea that we're gonNA remember hundreds of specific invocation names I think it's probably not likely you know certainly people be able to learn five to ten <hes> seven being a magic number for memory of these types of things but <hes> so I think that's really interesting in the idea here of really being more <hes> intention <hes> oriented in like us a term intent a lot and in invoice invoice assistance this idea that a consumer has an intention and that intention is to complete a goal. We'll might be to listen to music. It might be to get a movie ticket might be to plan a night out and if that intention involves either information or tasks that are not all contained in a single skill then the question is how do you do that without creating this choppy experience yeah. I'd how do you orchestrate all those services. It's a very difficult problem and so and so I think Amazon's a long way from getting. Marriage I think it's really interesting. They SORTA pronounced it <hes> because they look into get into a preview period. I don't expect this thing to general availability for a long time. <hes> yeah certainly into next year before it would be G.. A. would be my guess and then it's only going to be a couple oh scenarios and then there's this tie between the two but I think really what they wanNA. DO IS THEY WANNA start seeding new development into the model the imperative model because it's less work on developers alpers it is hopefully will deliver more intuitive interaction so that not every developer needs to be a sophisticated to develop a good experience because Amazon and the Alexa Platform can actually smooth over some of those parts based on what it knows <hes> and so those were important and then eventually with his crosskill goal fulfillment. They'll all be prepped for it because if the if you've developed something in the existing model the old declared of model. It's probably not gonNA be discoverable for some of these cross skill capabilities so so I think it's really fascinating. I love what's going on. There and I have no doubt there will be implications for design in fact. I think there's going to be more thought about design when this comes up because if you're doing imperative model you have to think a little bit more about what the platform might do for you <hes> and you can think about the things you definitely want to be specific and you can really hone in on those without having to deal with all of the ancillary things things that you might normally have to think about which really just more table stakes that they even just now let the I elected in this case. Take Care for you right and and how do you <hes> you know in the future scale these conversations to to last more than motor two returns. It's very difficult to go beyond that as a developer now okay so Alexa conversations for those of your engine that we did a re Mars episode just a couple ago. I think it was one. Oh One maybe <hes> and I recommend you check that out. We've got a bunch of people talking about it but is amy said. There's a lot more discussion to be had about that going forward okay so let's move on to <hes> sort of what I would call the technologists block. I've got a couple of them here. <hes> the first I is episode eighty six Fiji Balasubramaniam from pin-drop and Dave you hit list of this is one of your top three. Tell me why so along with <hes> <hes> kind of along the same vein as mere Hirsch. I look at what V._J.. Bala- Superman Ian in pin-drop are doing as another really big idea. I think it's going to be one of the core technologies to all of the voice technology that's coming in in his whole approach is around security and just listening to that discussion was absolutely fascinating. You know the fact that they started in fraud detection and then now they're moving this into basically voice biometric gene and so being able to use all of the different <hes> indicators in your voice to identify you. I think the implications there are pretty profound for the Internet of things <hes> I think that if we think about the security aspect aspect of say like a voice activated walk the two factor authentication system that he described where it's your voice coupled with the device that you have on your person I think makes a ton of sense and you know I think it moves moves way beyond that though to all kinds of different applications like I see a future where you have these Nanno payments that you're constantly paying for things and it's all being authenticated through your person through your voice ace and so that technology in general I just was actually blown away with and in really fascinated by the implications of really being able to use your voice as your own personal key and have all these different the systems be integrated with pin-drop security to identify who you are. <hes> you know in all these different kinds of settings p. This was also on your list. What stood out for you? I love this one. <hes> this was one of those I think I listened to two times in a row because we're just so much content in their <hes>. Two things that stood out was I did realize there was over a thousand attributes that are not related to voice that he was using crazy attributes. You know the switch <unk>. It just isn't something I ever thought about but while fascinating and the other was voice aging the challenge of you know you can print your voice now but two years from now it's it's changed and aren't six months through now fairly yeah wow that's right or if you're the president I guess even faster because of the stress level and there's this fascinating yeah. I love that in this idea that I mean you brought this update like this this idea of like how do you identify the voice in what I immediately went to is this concept of we should be able to go up to any device that has a voice assistant that we've interacted with in the past so let's say Alexa Google assistant which we've all tried everyone on this podcast anyway his tried and we should be able to go up to that anywhere anywhere and it should not be in our house. It could be in a hotel room. You know we're starting to see people hotels. Come out with this idea that you can sort of personalise. You're you're smart speaker in your room but anywhere we go a smart T._v.. Somewhere like in the resort at a friend's house. Why shouldn't it be able to identify US personally so that we can actually have a personal interaction with it no matter where we are as long as it's common voice assistant and it might be all of these things that the pin-drop is doing that allows us to be able to be done so maybe the two factor would be a useful thing but also being able to look at these other signals is Pete you mentioned to determine where we are in is an appropriate place for us? It makes sense as that lineup with our G._P._S. coordinates. Maybe they're being shared somewhere. Something like that Dave Yeah. No I mean to your point like that's that's the Big Aha moment was for me. Is that technology it really does allow for like true through personalized ambient computing like if we're moving toward a time where all of the like everything has his base smart assistance embedded into it and it just really it's like a commodity is world that we're living in where it's just massively proliferated in. It's everywhere <hes> I think it's really really fascinating to think about the fact that through your voice <hes> it immediately recognizes that it's your account in in there for everything becomes personalized whether it's your own owned properties your home or your car but like you said like when you when you travel you know it's going to recognize you whether you're at a resort or even at a restaurant or wherever it might be like it was a really profound conversation. The implications just had my mind go in like a million different directions all right so <hes> V._J.. Balasubramoniyan pin-drop security very interesting next up. I want to bring into the conversation Jeff Adams. I can't remember the episode number P._O.. Remind me of that but you through that on there and you said earlier that the Ron Krone episode. You're a sucker for an origin story. There was a great origin story with Jeff Adams. Yeah episode sowed fifty nine excellent origin story. I highly recommend if you guys want to really get to the roots of especially how Alexa kind of started with their platform you have to listen it episode because the one thing I took from that <hes> <hes> he certainly has a rich history with nuance places but <hes> it was you know when people think about this era with smart speakers of things they always think of the how good the speech recognition is and then how good the natural language understanding is but the third component it is really the the far field technology that allows somebody to speak across the room at understands you and sign with highly recommended up so it's just fascinating because without what they came up with we might not be talking right now yeah for those of you haven't heard it so Jeff Adams is he. He's been in the industry since the nineties. He's he's a technology. I think he's trying to think as backgrounds. He certainly has a C._S.. Background I think is electrical engineering as well but he started working at some of the the top voice speech technology companies as C._T._O.. Back in the nineties and sort of carry that through and he was with a company when he called the APPs it was acquired by Amazon in two thousand ten or eleven remember exactly but that became the foundation for the automated speech recognition for Alexa and great story about how they locked him in a room for today's him and his team and then tell them why they acquired and all these other things in eventually they said Oh we're going to create this. Thank all the smart speaker and it's going to be able to identify different people speaking from across the room and he and his team said Oh you can't do that. They said well no. We're GONNA have to do that so let's figure it out together and so he walks through that and it just comes brings up this point with whether you know what's the bigger invention over the past several years in voice assistance. Is it actually the the I that the voices isn't got better in the in the deep neural networks made the better or was it. The fact that the far field microphone actually started to work and introduced this entire new set of use cases around the hall with Fairfield are field mice in devices across the room. Yeah it's that an instance the implementation of the wake word how they do it on device. I mean it's just you know you look at it and just say well that yeah that's the model that's the way it should be but somebody had a bet that absolutely absolutely okay so I want to go to another concept. More in this is around compilation episodes so compilation episodes are generally wear eva or I will be on cited an event and we just conduct a series of back to back interviews usually somewhere between five and ten minutes long and we did the first time hundred we only did seven of those. I think the first one was certainly within the last nine months but they seem to be pretty popular in eighteen. You put these on your list or two of these on your list C._A._S.. Episode Seventy Eight and seventy nine. You know tell me what you thought about those in general around these compilation episodes yeah I mean I I love those because it it's a it's a window into what's going on. If we're not able to go to the event I went to C._A._S.. Back in two thousand seventeen <hes> it's not something I go to every year but I thought you did a fantastic job of of getting many different aspects of of the the voice industry on the podcast and you were able to talk to Dr Lee at Samsung and he talked about their their vision for the <hes> the connected life and this was sort of before the big push to get developers onto big speed he maybe alluded to that a little bit <hes> you got the <hes> Roger Zang thank from Alibaba was talking about team all genie which is not something we hear a lot about over here but it's really interesting to get his perspective and then you had add some smaller companies on their like ones that are trying to create a devices for for children children's content <hes> smart shower Microsoft so it was it's just great to get that kind of information <hes> without without having to actually you know go to the show to be honest yeah fair enough so episode seventy eight was eight interviews with small innovative companies and it was part of the showcase on Sunday night sort of before things really kickoff off in earnest their E._S. and then episode seventy nine was mostly big companies are a couple small company SORTA sprinkled in there <hes> but these were tended to be more established businesses everything from Samsung L._G.. <hes> and you know to sensory and a few others that was a that was really fun. Dave what's your perspective on these compilation episodes or the seat. Yes ones in particular. I love him. I agree with amy. I think it's great that hey you can go and do Vegas for us and Kinda report back. Everything that went went down there and <hes> it. I mean there's a lot of really interesting. Insight like I remember I think was at I fly tech. That was the Chinese company company that did all the live language translation yes yeah that company was really really interesting to me. I mean that it was a it's a Chinese company so it's the impetus in Chinese but it seemed like they were pretty far out ahead in terms of of <hes> the type of live language translation that they can do <hes> but in general. I think it's just really cool and it's Nice because what I've noticed that you tend to do is you'll get you'll interview a bunch of people which I think inspires later episodes road <hes> like I think it was either Amir Hersher B._J.. Balasubramaniam that you interviewed at one of the compilation episodes and then you brought them on for a full interview so what what's great about that. Is it kind of gives everybody a preview as to like okay so so these are really interesting people and I I again. I forgot which of the two was but I remember when okay I remember when that came out. When the actual episode came out I recognized him from the compilation episodes so it's nice as kind of almost a teaser for later later episodes yeah? Let's get our ten minute. Exchange was so good. I was like Oh Jeez. I didn't really realize what I was scratching the surface on here. We need yeah. He was very kind to give us some time. <hes> that's great. So what do you think Pete you've listened to. Most of the podcast probably episodes I mean where do you put the compilations like 'cause most of the things we do with these one on one interviews we also have the panels were easily talking about either <hes> a specific topic that we've done research on region sort of the year end episode. We don't have too many of those who probably had you know eight or ten of those. <hes> you know how do you think about these like the individual ones for the panels versus like doing today or the compilations nations which are just one interview. After another. I mean I would probably say a my favorite style is the you know the single interview but I love the calculation ones because <hes> you just get a for example you just WanNa Google I o and just I I think Nick Schwab is a or somebody and just hearing these perspectives not only from you but on really smart people giving perspective on the conference. This isn't something that would never really get A. Let's listen to something like that <hes> I i. I actually have a question for you. How many of these conferences to co two every year? It seems like you're not everyone. <hes> I've been to more than I planned is probably the best way to say that so I think yeah I've done compilations. My first one might have been at the Amazon event last year where they did their product launch in September and they did twelve products and then they did like seventy other announcements. Sousse insane and I just got some journalists on and I did that was popular. People sort of like that so I did it see yes I've done I did too at the voice of the car summit which Score Publishing Bradley Matt Wreck put on which it was really good. <hes> you know is able to talk to some big companies and small companies air that was popular <hes> I just did remorse. I think I did another one too. It's hard to remember Oh and I've got one coming out like actually by the time this publishes wishes will be another one that came out of just to meet up in in New York City because I was going to be there and was able to capture five interviews of people who attended to sort of give the perspective of what was going on at that event so <hes> I guess I guess in the end and a lot of these events I too but the intent would be to go to no more than one every other month but it seems like it's much more frequent than that yeah they're great. I think that's how we really originally met last year. It was through. It's true we did you were part of a compilation ablation fact that was probably my second compilation episode <hes> because we had you roger and Joe Murky <hes> yeah because you know we have said that Samsung developer conference and I was able to snag <unk> Adam Chair. We'll talk about a couple of minutes which was great because you know I hadn't really known that I was going to be able to do that and then as Geez you know we've got some people here who really have a perspective on this. We need to share those thoughts and so that was great two and that was yeah we met the I met you and then two hours later. You're on the PODCAST. It was great yeah and we're starting amy any thoughts to wrap this part up yeah. Just I think the hot takes at some of those events. <hes> like Pete alluded to are really helpful when you were at Mars and then you know the Alexa conversations news came out to get the the thoughts of some experienced developers on how they were digesting the information that was really useful so that's one of the things about the compilations that what I think is helpful. That's great and I think that's nice too because they tend to be more like articles and not everybody has time to read all the stuff so it's a lot easier where they can sort of get that heartache so so that's good we we definitely get a lot of good feedback. I think we'll we'll wind up. Doing it. Certainly is many I don't know we're going to do a lot more but <hes> because we tend to focus on these one on one interviews and so I want to transition away from that and I want to talk about what I would call sort of the platforms we've had we've had several <hes> but I want to talk about a couple in particular that we had that were very popular so the first two I want to talk about our Paul Cut Senior from episode seventy seven David's Bisky from ninety five live <hes> they are both from the Amazon Alexa team both of those episodes now they sort of cheating because they're both this year and we have a bear audience last year but but those are both in the top ten all time <hes> and you know Paul was able to talk in some dapper around what developers are doing on the platform with their with their not doing but should be which I thought was Great Day. We talked a little bit about developers and his background but we also talked a lot about brands and how they're taking advantage of the platform or making mistakes if he was pretty open about that so just along those lines like did anybody on the panel. Did you guys have any perspective on those Alexa conversations. I had earlier this year Dave. I thought that I mean I think it's so cool that <hes> I'm I wasn't I was in high school when the mobile era really came around so impressed like podcasting wasn't what it is today and so oh I just think it's really cool to hear someone like Dave or Paul. Be So candid about you know just like the whole developer community and transparent about where they're at with technology so for me like <hes> <hes> and I'd say the same with Adam schefter like just the access that <hes> like sharing these leaders and like chief evangelist <hes> speak to what's going on with the technology is extremely insightful invaluable and I just think it's it's really cool like that. You'll you can parse out where the technologies at ask questions about. You know like the way that the changes are being made around Alexa conversations or something like that so it's for for I think the audience it's just really really cool to kind of get that type of access to these people that really are in charge of the decisions of how this technology is being built. I totally agree with Dave I mean for me. It's just it's important to think that these people are still around there accessible in a way. It's kind of very cool that you know Dave is still doing the evangelist thing although I think maybe he said his title changed or whatever but you know the fact that he's still out there going join to conferences on his accessible is not necessarily doing the office our stuff anymore <hes> but Teesta you know he's he's a person that will really inspire you and help you get stuff. Done and Paul is the same way I mean. Paul is just very supportive advocate for the third party developer. I think that might be missing from you know some of the other platforms but for me the the that kind of the you know that kind of leadership is just really important. Yeah I was just GonNa say to Amazon's really made a huge investment under evangelism team. I think they just do a wonderful job. If you think of the last three years really really after educate on many things that are completely nude everybody and <hes> David came from other companies like Microsoft so you know this isn't their first Rodeo so they know what it's what it takes to really build a platform developers for Sonya. I will give you a little inside baseball here too. I asked Amazon for eighteen months to have someone on the pet podcast now. I didn't ask them every week. Yeah I'd ask periodically. It'd be a in Davis tapings skin. I talked about it several times but I think early on the the P._R.. Department maybe a little bit more wary of having people talk. I'm podcast and stuff like that. Particularly People Apple's sort of know what's going on but wait ask questions. It would be uncomfortable but I will say they've been great this year. I've had Amazon. I think three times now I've had those both <hes> Paul Day but also Aryan Walker Walker who runs the auto team and so you know Kudos to them for for being open and debuted one more comment before I move onto the next one. Will I just wanted to ask you. Why do you think that is I mean I've noticed that your podcast in particular color? You're getting like all the different platforms coming on here. Do you think that they're noticing that by being really transparent and open to the type of conversations in questions that you ask that that it's actually really beneficial bill to have that level of transparency. I think everyone I've talked to they. Recognize the benefit of it at after they've done it so you know whether they know there's benefit there beforehand or not. I don't know and I think part of it is we do this long form <hes> format where we talk about a lot of things and we go into some depth and that makes some people in the P._R.. Community uncomfortable because they're like Oh. What are you gonNA talk about right? You know if you're gonNA talk for an hour is am. I GONNA get in trouble or as you know something. We don't want to come out but you know frankly I think part of what you're what you're angling towards his right. This openness is really helpful. In in actually the the event that Amazon had last September September for their product launch. They had a lot more journalists there than they'd had the previous year they did something which was a little more contained was much more open. This last fall in in some of the people on the the media relations department were excited because they're like we're finally family. We're we're you know this. I won't say we find this is a change. You know for Amazon that it's trying to be more engaging more open share more information. I think it's very different because Amazon five years ago is very insular didn't really talk much the media at all and they're in the sort of learned how to do this engagement and <hes> and I will say I've had other news with them too I forgot about they set up some interviews for me remarks as well. Some of that stuff won't wind up in the PODCAST. Ask but you know for other purposes so I'll say they really want to do that now. I will say with what we do. <hes> it's not so much we're asking him opinions about sort of events of the day or the type of thing we're asking them to tell their story and we're asking thoughtful questions that we know our audience will care about and we have a pretty big audience in the people that who listen to voice by podcast Amazon Google Samsung all these platforms really care about reaching and making sure that did they hear some of this information what they have to say and even respond to the tough questions when they come up okay so great question Dave <hes> hopefully it's perspective for folks but I wanNA talk about Google and I haven't had Google on for <hes>. <hes> two years not quite over maybe twenty months because they were so good to me early on Brad Abrams was episode fourteen still one of my favorite episodes <hes> sort of the first big company that made as someone available to me and Brad Abrams feed oh no he is you know at the time and he still has similar type role but oversaw the Google assistant engineering team and so if you think about it that was about it was less than a year maybe nine ten months after they ahead formerly released Google Assistant to the marketplace and there were so many early learnings that he shared we talked a lot about search so much depth on how Google was thinking about search on voice assistance and and you know having that perspective now and seeing some of those things are the same or evolved. Maybe changed better word. <hes> is really fascinating and then also <hes> some of the things they were thinking about in terms of the way they were gonNA enable experiences across surfaces. I really excited about that and I think if for a lot of people I would say that maybe the audio quality isn't so great maybe Pete you may have mentioned that from at one and I'm not really. I don't remember exactly why wasn't great but I will tell you. The content is excellent and you'll get a perspective perspective there around Google you'll be able to see some of the things that they've done you build under <unk> you'll be able to learn <hes> what may be as a change in their understanding of what is just really a an extension of where they originally were headed. So did you guys want to comment on Brad Abrams. I don't know if you've listened to that one. Yeah I'll just say I've actually known who Brad was before. He worked for Google Super Smart Guy Obviously and I just think that Google is in a very interesting position for for many reasons <hes> certainly from their distribution. <hes> you know the quality of what they've produced <hes> so they're they're going to be around for the near future for sure <hes> certainly in the space <hes> certainly certainly <hes> if you talk about voice search I mean that makes perfect sense. Felon the focus on that I don't I don't have anything for the Brad Abrams episode just because it was so long ago. I don't really remember it but bread. I have a question for you. Attending all of these different conferences <hes> the Google I o this year really really stood out to me. I thought that <hes> Google on our duplex on the web was fascinating than also the assistant to point out so just <hes> kind of off outside tangent here are how are you thinking about Google where they stand right now. <hes> I would say Google fascinates me it the breadth and depth of what they're doing a lot of that is an obvious unless you're really looking at the stuff on a day-to-day basis and sort of understand where the limitations on the other platforms are did is Google duplex with something that everyone sort of understood stood was important in different so that was last year and that was that was a good indication of the fact that Google has a lot of assets here <hes> that maybe some of the other players don't have at their fingertips so even though they came to the market late <hes> they might be able to catch up in leapfrog in some areas now google duplexes had some issues. I think you know a year later. It's maybe not what it was portrayed to be still significant forget advanced but maybe not as spectacular as as initial is initially tried to lead on so duplex for the Web I think is interesting that goes to the idea that they've got this machine learning and they can apply in many different ways and and I liked and sort of automatically the forms and those types of things <hes> but I will tell you the thing that most impressed me was the on device this year <hes> in partly was the on device performance that the low latency the potential for security benefits fits that we've seen some of that with apple without device with Syria but we haven't seen anything that extensive and the you know having had an opportunity to speak with some of the senior folks in the Engineering Group there who are are different than the other folks that we might deal with on a day-to-day basis in the voice industry <hes> really fascinating what that could lead to over time and it starts to make you think that what we've been doing over the last couple a couple of years as my friend Jan Gilan Might Sane is soon gonNa seem quite quaint <hes> so so. I think we're I think we're we're moving to a place that <hes> we're. We're GonNa see the things that you know. The training wheels approach that we had over the last couple of years is just GonNa get blown away. You know these things really metastasized okay. I thought real sorry real quick. I was just GonNa say I thought it was ironic that the <hes> you know this whole privacy debate because I like you that the whole notion of moving it from the cloud to on the device kind of eliminates the whole well to a degree but it definitely changes the narrative round on Google and <hes> privacy it doesn't they had a great headline about facebook talk his talks about privacy Google just implemented it. <hes> and there were a number of other privacy announcements at I._O.. Were part of that story. I think it was verger tech crunch at that had headlined <hes> but I think the timing of that for them was perfect. <hes> you know we've seen <hes> F._A.. From facebook Julio. And <hes> Apples W._d._c. all with a significant focus on privacy in the middle there we had Amazon announcing some new privacy features voice <hes> deletion of Records and the the privacy center in these things so all the platforms are really really zeroed in on the fact that this is an important topic <hes> maybe a Zeitgeist of the moment that they need to be out in front of and so it's GonNa mentioned to see how much of that actually takes hold as opposed to just announcements but I think it's fascinating all right so we do need to move on. <hes> got just a couple more things to do amy. You talked about something that interested you. SORTA listening back on dozens and dozens of these episodes is the evolution of monetization and you've pointed out episode eleven with that Media C._E._O.. Pet higby <hes> episode thirty five with Stats Shinsegi from dramatic episode thirty seven was facilitation Karanko from the former storyline <hes> which became a vocal and then recently shut down and one of the few <hes> departures from the industry or demise is the we've we've had the opportunity to watch end to end and then I'll throw another there episode sixty without a chick from Alpine day I and previously a voice labs the the sale of that business to head space each of those companies had a very specific perspective around monetization. Some are continuing to implement those what was why. Did you put those on your list. Why did you think that was a really important thing to bring out in terms of the portfolio the back catalogue voice about podcasts well? I mean someone that's trying to build a business. I'm always very very intrigued by what people are looking towards in terms of a viable business model and I think because you have such a portfolio of podcasts over you know such a long period of time now <hes> we're we're kind a seeing an evolution and you know I mentioned that the first episode that I listened to was <hes> Gioja Quinta number nine and I think even in that episode he was talking about his frustration as independent developer that <hes> it was very difficult call to to have any kind of monetization. This was of course before the ice the ankle purchases were our you. He's still frustrated yeah <hes> so I think this would be a really interesting. He no longer discussion but just to mention it here for in terms of the <hes> the hundred episodes I do think that we see that there's we still haven't figured it out and there is the there was early on the thought process that we could potentially go more towards an ad ad model because that seems to be what's working on web on mobile <hes> but the add model doesn't seem to fit very well with voice and even Steve are kind of shoes having some success with it. I think he mentioned when he talked about his experience that his his <hes> his listeners who really want to get the weather but they you know they're they don't like to be sold to he's not fat so there's some issues with the I._S._p.'s <hes> and then story line is a very interesting one but that would be a longer discussion because I think <hes> <hes> the Sealy. I can't remember exactly his last name but he you know he had Shinkansen Yeah okay he and he had this vision of how content creators the voice was going to be the youtube of content so his mantra was let's provide tools to the content creators and they will be the ones that are going to develop all these these assets that will make these devices worth buying and listening listening to any kind of came to the realization that well it's harder to design a good experience than maybe originally thought and not only that but there was some skepticism about the the readiness of the technology to have a full experience <hes> and so that business model got shut down so I think we're in a place right now. Where as things have evolved? We realized it's it's hard for ended independent independent developers to make money through <unk> purchases but the ad models aren't really allowed or working so where do we go from here and I and if we look back historically <hes> it's interesting to see the evolution but I'm not so sure that we that it points to to a direction for the future yeah. I think that's fair enough so to give people perspective so that media was doing an ad model on mobile with voice before Alexa was even launched and so they had they existed in the voice space based before the skill world they actually shifted their business to more of a subscription model around helping companies the launch skills and manage networks of skills and <hes> in in that's that's been a big success for them due to have pat back on the podcast at some point <hes> stocks with instruments. They're they're actually implementing a similar ad model where they insert ads <hes> in different types of venues at Amar Marcek from voice labs originally had an ad network ran it for several months on on Alexa was quite successful. Amazon was fully aware of it than they early in early. May of twenty seventeen changed their policy Nelson said No. You can't do that anymore. <hes> so that was sort of interesting so ads have been SORTA shut down. I think what we are seeing what they say is that advertising works on the web but we've also seen subscriptions work on the web for certain types of applications <hes> certainly on mobile. There's at there's an advertising model but Framingham is is probably the most successful subscription didn't really work very well there. When we come to a voice mostly what we're doing now is trying to push subscription models or transaction models? <hes> you know I think the one time purchases of of the content or or software probably are not. I don't see very many of them so I don't know if that's an indication that they're not successful but people have tried it. There hasn't been a lot of that seems to be more of a push push towards subscription. I know that's popular. There's there's there's a premium concepts to subscription which Steve Arcana Vich implemented which seems to make sense and then there's that transaction concept which it oh best buys trying to implement a Walmart <hes> <hes> <hes> easy needs <hes> you know sort of the wholesale bulk sale side so I still think that idea of the transaction model of selling things in either taking a piece of the transaction or making your margin for selling goods you have love is a play and I think subscription is is a question right now. Whether that's going to be successful I will say that there is definitely advertising on the general purpose platforms like election Google assistant you find it and things like podcasts or flash briefings Sir <hes> other types of things at news media traditional media put out there <hes>. It's not necessarily generating a lot of income for anybody yet but it is something that that is viable in does seem to be working for a number of publishers assures Dave or do you have any comments on modernisation. You'd like to weigh in or should we move onto the next the final two episodes. I mean Miley quick comment would be <hes> you know. It's just been frustrating <hes> to try to monetize especially on some platforms because unless you get the attention <hes> and be featured on these <hes> through discovery. You'RE NOT GONNA get a lot of usage therefore it's hard to even entertain doing premium model or something like that so <hes> you know I know some people are making money from Amazon rewards but it's clear that subsidize GonNa. That's GonNa go away while Yes Amazon rewards has definitely been they most most lucrative revenue stream for anyone in her voice space so far. It's very tangible okay so that's that's. That's a great conversation on modernisation. You know amy. I'm glad you brought that up because it probably deserves its own focus at some point. I I still feel like things are maybe a little too ambiguous at this point to have a really good episode around it but <hes> that's something that will put on put on the list to focus on coming up okay so the penultimate ultimate episode. We want to talk about today. Episode Eighty with Brian Romley he made a presentation the first presentation on this topic at the Alexa Conference in Nashville Tennessee in January of this year and he outlined his concept of the intelligence amplifier and the wisdom keeper which are two separate products so dave you you're at that event you spoke at that event <hes> expensive time there and you have some specific thoughts what's on that episode and what Brian Head Yeah <hes> I thought it was. It's obviously really really fascinating. I mean we were talking a little bit about this before. We started recording here but you know Brian is he's real out in front with the technology side note the way I got involved into this whole space was I I actually stumbled upon Brian on twitter and that's kind of how I got roped into this so I've been following Brian for a while and <hes> you know it's really interesting. It's this idea that you would have I think his personal assistant that he's built using a ton of different raspberry pies's Alfred and it more or less it serves as like this your own personal intelligence amplifiers so it's like you take it and it learns everything that you are learning but it remembers everything like we can only remember so much and so it's Kinda like this own personal personal memory for you and and then it would be kept in the wisdom keeper in so I mean yeah it was it's a real <hes> a very high level macro sorta conversation and and <hes> video that he presented there so I mean you could go a lot of different directions with it but it was definitely something that if you're involved in this space just to be aware of because it's it's an interesting thought of where this final go the I mean Bryant as a visionary. He's always been a visionary. I think some of you know he he tends to be way out there in his in his thought process which is good because it pushes the envelope so you need that I think sometimes sometimes <hes> it bothers people because they don't necessarily see it as being very practical and then if it's not practical they're like you know. It seems like it's just more a very very fluffy. I I enjoy listening to him. He's always an inspiring speaker. I for me one of the issues with with his vision of recording everything this is that I'm not sure <hes> if you recorded everything that happened all all the conversations you had all of your your interactions it it <hes> it doesn't take into account the need for curation because recording everything unless you had something like <hes> you know audio bursts to go in in there and and help you curate the things that were that you were looking for it. It would just be overwhelming and it's like the person that can remember every single thing from the day they were born. I mean that's really not a good capability to have so <hes> I liked the discussion. I think that there would be more morineath to kind of figure out how to make some of this stuff happened practically and and <hes> visionary is good but you you need to bring a product to market at some point. Yes Brian bring a product to market Pete. What do you have to say the BRYANS grade? <hes> in a lot of your guests will talk past and present he certainly very future tense and who referred to scientific scientific things that I just don't hear a lot of guests <unk> towards like <hes> you know the through print of your brain how many bits per second you can comprehend. This is kind of interesting stuff so I I like him as a guest yeah bring something to market by unless you do yeah so the in in fact I I keep seeing things I think like a I don't watch Black Mirror <hes> but I think it is saw something that's rely that similar to this idea wisdom keeper or no the intelligence amplifier and like this trailer that just showed up on my netflix because I was trying to scroll through when the other day <hes> but it is really cool. I think amy that the concept air is just as Google curates the web and tries to give you using agents tries to give you the best answer <hes> it's for what you want. The the idea there is the curation is flat of that in any given time by asking the question or maybe eventually thinking the question but in his world asking the question that it'll surface the right memory Marie and and then you can then stitch together the context. I think one of the most interesting things about it. Is this idea that the wisdom keeper I think is really cool in that. <hes> some other people have tried to do this or have done facsimiles of this us with chat bots around this idea of compiling all the information of that someone is said or thought most types of things and then be able to query it later after they've passed away or you know potentially in this case if it's if it's a real time learning solution it could actually it could actually be you somewhere else or affects emily few somewhere else so you could be in multiple places at the same time. This actually does not seem to be that far out there to me. I think it's thanks reason why from a practical standpoint from an applied innovation standpoint I don't see anybody doing this soon but it seems like a logical progression that someone will want this type of tool and whether it looks exactly the way Brian talking about it or not I don't know but at least portions of what he because he's got a very holistic vision portions of it. I think are definitely gonNA GONNA come. Amyot to final comment there. Yeah actually did see the Black Mirror episode. I forgot what it's called but molly Cyrus is in it and she's a a singer and she has this this <hes> Alexa type doll that she sells and her entire brain and personality is uploaded into the doll and then when something happens to her that you know the the the the information the doll is is very useful and I won't say anymore because there's a spoiler but it is. It's exactly the same concept so it's pretty cool to watch yeah. It's it's funny. I mean there's a lot of these things we just did. <hes> did a thing invoice insider a few weeks ago about the knowledge navigator does around W._W._e.. See this is something from. I think nineteen eighty six from. From apple which is basically a smart display concept that everybody wants today with really smart assistant that can just coordinate and do all sorts of really complex for we're nowhere near the knowledge navigator from nineteen eighteen eighty six that apple is trying to sell apples probably the furthest from it but a lot of these ideas have been out there. It's a matter of putting it together and getting the right momentum behind the idea to bring it to market just like you know the the momentum behind the smart speaker probably wouldn't have happened without sort of the the strong conviction of someone like Jeff Bezos Okay so final story our final episode one of the favorites. I certainly got a tie search social media probably the favorite episode last fall. I sat down with Shire episode sixty nine Dave and Pete. You both put this on your list. <hes> tell me why start with you. Dave well like I said earlier. It's just unbelievable like the Times that we live in today to where you can get one of the founding fathers of technology like Adam Shire like hearing. His story was really really fascinating another S._R._I.. Guy <hes> who's been I think he said he's kind of at this for twenty five years and what was really fascinating about it was that he sort of envisioned. It's like he didn't envision the web or mobile but he envisioned voice in the Smart Assistant Listen epoch and <hes> so I just thought that was really interesting and the way that you know he has his trajectory has gone you know starting with Siri and then leaving apple in moving to give the now at bixby but the thing that arms are Samsung <hes> the thing that really stood out to me is when he's describing the idea behind a capsule in he's saying you know it's the <hes> it's not necessarily about like voice as much as it's it's about contextual assistance <hes> and then he kind of walked through the different <hes> examples of how that applies so he says you know you in this is something I've thought a lot about but it's this idea where you just say. I'm really the job to be done so you say okay. I want to plan a vacation or I'm I yes so help me out in plan this vacation for me and the way that the assistant works on the back end is it's brilliant navigating. All of the different APPS and edge was crazy to me listening to that story. Is that it it's very very obvious. That apple looked at that at one time. Maybe when Steve Jobs was still around in said this is the evolution of the APP store because has it you you hear him. Describe in really what it is. It's interconnecting all of the different bits within the APPS <hes> to make a ultimately fulfill the job that you're looking to do and now that you know they're doing it over at Samsung <hes> and I think Alexa conversations was probably pretty inspired by what bixby's rolling out <hes> or at least maybe to a degree <hes> so I just thought it was really interesting to hear his vision and understand understand this the bigger picture of kind of where the guy that really kind of lead things off in this space <hes> his ultimate vision for it yeah <hes> same sentiments <hes> in order to understand bixby Samsung you have to know who had entire isn't where he's coming from. You can look on its website and see demos from the early nineties where he's really thinking about how to solve for these things and no his vision of a assistant if you hear him speak it's it's a little different than other people. You know when you think about the stack there's really in my mind. There's three components. There's the speech recognition there's the understanding but then the third apart which I think we're still working out is <hes> you know the fulfillment of the do engine right. How do you fulfill those requests and I think if you look at his approach it's so dramatically different because what he will his team him what they do is the last few to model data in so describe exactly this universe of data that you want a hopefully talk to manipulate in sign language to it and then let an A._I.? Engine figure out how rounded service at requests I mean it's dramatically different and that's why I think it's superior platform. It's it's a learning curve now because it's hard because it's just so different <hes> but you know you got to think in terms of of context I I think he knew he understands as personal assistant so all these all these attributes like <hes> you know your previous interactions are important. I'll just give you one really simple antidote where I think some people miss when they're developing these. It's <hes> if you've ever I've seen the movie double wears Prada. There's this girl and she's really intelligent but she doesn't have any context you. She doesn't work in that industry and she's the assistant to this. This woman is very demanding and she's terrible at it <hes> but over the course Chris of six months she's probably the destin she's ever had. She knows what kind of coffee she wants. She knows the look that she gives remains a certain thing in all these things collectively allow her to service those requests in a better way because the context and I think you know <hes> certainly if you look at <hes> atoms approach. I think you know he certainly approach is it a different way in his is thinking kind of the next generation of how we got you know fulfill the promise of these assistance yeah. It's it's interesting to hear his his view of the assistant as <hes>. You know you're one contact with like the artificial intelligence world so he doesn't think there's going to be multiple assistance with there's going to be one that really he knows you that knows the context understands you and that that can then get you the services that you need and when I when I hear him talk in a way it's kind of it's a little bit <hes> worrying because if the if the virtual assistant listen is going to become so important then you see why this battle between you know Samsung and Amazon and Google and apple is is so critical because you know which one are we gonNa pick and I guess for Adam I'd have to think it must be frustrating for him because he you know is out there on the forefront so so long ago but just based on timing and different you know things that happened his his view of it didn't wasn't the one that got out there and got the most traction early on so it makes me. Wonder you know what's going to happen. I think is really interesting. If you think about that episode he talks about the original Siri App so Siri was first rolled out publicly although they'd had some implementations sort of more private enterprise is situations rolled out publicly as an APP in the Iowa's APP store and he he makes his comment where if you talk to Steve Wozniak one of the original founders of Apple <hes> you know even last year would say his favorite APP of all time was a seri- It was not the theory that was implemented but this area APP because it was more this this full concept of it would asthma talk about is knowing and doing right so you'll be able to to respond onto queries from users by <hes> either accessing knowledge and he sort of talked about this concept of there's knowing companies that aggregate all the data like Google and his voice system just needs to plug into those is knowing companies and then is pete you brought up this idea the do engine which was core to the the Siri roll out in Bixby's really focused on this whole doing things like how do you put all the things together to create these transactions and capabilities abilities that either pull information or or pulling services web services <hes> in order to fulfil need and I think that was I think that's really fascinating and I do recommend people listen to that. Anybody who's in the industry that that might be the most instructive of all of the of all the podcasts because we do actually walk through twenty five years you know he's hand dot keet Laos his Co.. Kander were known for not just creating Siri this sort of the first modern voice assistant whatever you might think of that today <hes> but then also the labs was become bixby's so they've got to under their belt. It's pretty impressive Dave yeah one of the things that he said in that interview that was really interesting was about the fact that <hes> you know he has created fifty different assistance and after Siri came out it was as if everyone assumed that was the only way that assistant could be and he was describing all of these different environments in which you might have a different assistant so he described the one where <hes> you know for like a surgeon <hes> the way that that assistant was built so I thought that was really interesting because again it goes back to this notion that we've it's like he brought to market Siri and that was what everybody assumed to be in just to hear him speak about how it doesn't necessarily need you too all fit the same mold <hes> it. It made me realize that you know there probably will be a whole lot of variety in terms of their ways in which people use their assistance and it won't necessarily just be everybody uses it for the weather and everybody uses it for this use case. I it just made me think a lot of different things around that yeah. I think that's a great point that you brought up. I mean he was putting a finer. I guess a finer point on the whole concept of what Syria was supposed to be what it became mm-hmm and one of the challenges they had with serious. They didn't roll out the full when when they rolled out in the iphone four S. it didn't roll out the full feature set of the original Syria because original Syria could not have rolled out in that many countries at the scale that apple was going to roll it out and so they didn't really have enough time to build out all the feature sets and then when they did roll it out they actually way underestimated the amount a usage that was going to happen and so again just like it was all back office stuff that they had to worry about in terms of the infrastructure and that really narrowed the the functional set and then by the time they could've got around expanding going back to the original idea the due engine dog and we're gone and they found it and I think the the other leadership that was left with that <hes> Tom Gruber was still there. He was a CO founder series well <hes> but I think the rest of the leadership was a minor diner expansion really making it work. Well was really their focus. I thought that was that was really really interesting. <hes> any other comments <hes> Pete Dave or amy on Adam. Yeah I mean really quickly I I would just say in one of his other thesis we every ten years or so we move to these different paradyne's and we seem to be moving away from the mobile APP download update. Your apps every seems like two weeks. I'd have to update my up <hes> to this assistant model that what it's doing things on your behalf and is a company or developer. You're you're really writing to the platform and and this assistant is kind of interceding in that that's how he explained it in. That's Kinda degree. That's the first time I heard somebody express arrested that way but yeah absolutely yeah so every ten years you haven't user interface change in our latest. One is voice as an interface but also the voice assistance is platforms in so that that is important and I'll say this one last thing in closing about that interview with Adam that stuck with me he worked with his mentor was a gentleman that had actually rolled out a lot of concepts. He was the inventor of the mouse and all these other things he'd rolled out some things and <hes> nine hundred sixty eight that that he thought would be adopted in five years <hes> twenty years later they started to come about he had a conversation he remembers sitting down in Nineteen Ninety nine when the web was taking off in looking at a video interview of his mentor who people are asking why wire hyperlinked multimedia taking off you know the way you thought it would interconnected on the Internet and the ten years later it in one thousand nine hundred eighty didn't know why Jesus seemed like such a good idea. Why aren't people doing it? Well just it took a while later in just talking about that and so twenty five years and I said so what have you learned in one of the things he said is I've learned patients the light of these things if the right idea will come through and I think when we're in the midst of it and and we see where this is going you know whether it's the intelligence amplifier or security or on device processing all those things we can see that that's the right thing to happen but it might take a while <hes> and sometimes as a technical barrier sometimes it's just adoption and people are busy with other things but eventually they get around to it so I that's a that's a great episode. There's a lot of really good ones. I'm I'm glad we talked about this. I don't know how many we talked about your prior fifteen or more. Maybe twenty twenty two but we we've gone way over time. You guys have been so kind about this <hes> Amy why don't you tell the voice about listeners how they can keep track of you and check out the work that you're doing yeah you can follow me on the Taliban's website <unk> dot com or on twitter at I am at talks to bots and you can also check us out on instagram at Tel lables excellent Pete Yeah pitas dot com twitter conversation <unk> curve dot Com is by agency where do some development and I'll be speaking of voice <hes> Voice Summit next month I think along with Damian insult and David Camp sure you can find me on twitter at oak tree underscore Dave's <hes> you can also check out my blog future ear dot co <hes> yeah looking forward to meeting Pete at a the voice summit seen Amy Umbrella Gin Seniors at Boy Summit all K- that is great <hes> thank you so much <hes> Dave Pete Namie and thank you voiced by listeners those of you who stuck around listened all the way to the end. We got some great content in there in the last fifteen minutes so you were richly rewarded for that you can follow me on twitter at Brechin Sela. Check out what we're doing at voiced by that day I and if you really want to know what's going on in the industry. If you really want to know what's important what the people were the insiders care about uh subscribed voice insider. You can just go to voice about that A._I.. Force last insider you'll find out more information there that is subscription newsletter in it's stuff that you just not gonna see elsewhere so check that out. Thank you once again David Pete and amy.

Alexa amy stapleton developer Dave Steve Kemp Dave pete Google Amazon Ron Krone Samsung Dave Yeah Trinity Audio T. R. Pete David Amir Hirsch Clinical Oh founder Dave Camp John Kelly bixby
Amy Stapleton, Dave Kemp, and Pete Haas Discuss First 100 Episodes - Voicebot Podcast Ep 105

The Voicebot Podcast

00:00 sec | 1 year ago

Amy Stapleton, Dave Kemp, and Pete Haas Discuss First 100 Episodes - Voicebot Podcast Ep 105

"This is episode one zero five of the voice by podcast Amy Stapleton Pete Haas and Dave Kemp join me today so we can discuss some of the most important insight shared by voice by podcast guests in our first one hundred episodes. Hello everyone out there and voiced by Land Brechin seller your host for the voice by podcast. We're having a little fun this week with a panel discussion about the first one hundred voice by podcast episodes so no this is not Meta voiced by podcast. There's been so many rich insights offered by voiced by podcast guests that it wanted to figure out a simple way to unlock if you those nuggets in case you missed the episodes I also wanted a vehicle to remind listeners about what has been said you may have heard the insight right the first time but if forgotten it's not top of mind this is a way to bring it back up. I also wanted to get some discussion going on some dedicated and erudite listeners about what was said by pass gas and what they thought mattered most so with one hundred episodes to choose from we narrowed it down by asking listeners like you to let me know their favorite episodes why our panelists also weighed in and by my count we discussed today parts of at least twenty three episodes that means it over three quarters. Were not even touched I will any of your favorites included. You're about to find out is Amy Stapleton Tells Pete Haase from bear escape and Dave capital Oaktree join me on the one hundred episode retrospective. What a long fruitful trip? It's Ben then but before we get started. I have to do a quick shout out to this week sponsor Trinity Audio. Do you love voiced by podcast. Do you get a lot of value in insight shared by my guests each week. Thank our sponsors on social media and visit their websites if you do I'd appreciate it so twenty audio enables publishers to turn readers into listeners by transforming their written content into lifelike speech. We have voiced first interfaces today and that's leading to more demand for audio content Trinity Audio's helping hoping fill that demand with its contact. That's content technology solution. They characterize it is audio fighing your entire website or contact catalog studies by the company show the adding audio increases website engagement by five live acts and it's free so to learn more good at Trinity Audio Dot A._I.. That's Trinity Audio Dot A._I.. T. R. I N.. T. Y. Audio Dot A._i.. Check out the trinity player for your website. I trinity brief for creating a type of flash briefing from your website content in trinity skills that make your website content accessible through Alexa who even know you could do that so checkout Trinity Audio Dot A._I.. And if you get a chance send them a tweet tweet at the Trinity Audio you can tweet at them at the Trinity Audio and thank them for supporting independent media like the voice podcast all right now for some insights mind from the commentary of people like Adam Chair Cathy Pearl Brad Abrams. David's Bisky Liza Foulkes V._J.. Balasubramaniam Ron Krone and many many more. Let's get started okay voiced by nation. We have a special episode for you today. We are reviewing the first one hundred podcast episodes of the voice about podcast. I know many of you joined recently. Let's say within the last year and you've not had a chance to listen to all of the episodes episodes so I have three guests here that have listened to a lot of them. I won't say they've listened to all them. Maybe pizzas and all of them but I I don't think anybody is listened to all of them but they've listened to a lot of them almost all of them and they're I'm GonNa tell you some of their favorites some of the things that they took away and then allow you some of you will be able to skip over because you'll get the main point others. You'll say oh that's a topic I wanna go much deeper on so my guest today our first up Dave Steve Kemp Dave. Can you introduce yourself to the voice about audience. Yeah absolutely thanks for having me on Brett <hes> so like you said my name's Dave Camp on the business development manager at ultra products. We are a distributor of Clinical Oh supplies in assistive listening devices to the hearing healthcare professional <hes> so that's my day to day job and then I got real immersed into this space the whole voice space and started a blog called future ear fugier dot co <hes>. I'm started to write a little bit about what's going on in the voice base how it impacts the in the year device computers that are starting to kind of emerge herbals <hes> wrote some articles for you over at voice about as well so really excited to a chat about all things always spot today great. Yes I think we have maybe a half dozen here. ables articles from you and those are very popular. It is a topic that people really care about is tightly integrated with the way we think of voice and mobility going forward. Take a next up. We have amy stapleton. Amy Is not new to the PODCAST. She was with us back. In episode eighty four. I think <hes> amy introduce yourself place to the listeners or reintroduce yourself hi. I'm Amy Stapleton. I'm the founder of tells and we create conversational story experiences and we also provide a platform that lets independent authors contribute their original stories to our existing voice apps so example of that is our my box of chocolates skill and we try to publish a new box of stories every month and those stories are by different different authors each time and <hes>. I think you said you wanted us to perhaps mention the first podcast you're listening. Tell me the first one you heard. The first one that I remember very definitely is episode number nine with Joe Do Quinta yes yup. Good Ride Jos Jos a well-known developer in the community and he <hes> he really was very helpful and I think for a lot of the early listeners talking about his experiences building games M._C._i.. That was a good one dave. I forgot to ask you that question. Do you remember the first one you listen to. I was just going through my podcast APP to look and see looks like the first one was <hes> Yon Caning with job. Oh Oh yeah that's that's back in the day to. I think that would that would probably are I tend somewhere in that range. You guys get you guys definitely go. You started early so that that's a good thing yeah yawn <hes> build a Jovo which is a like a cross platform. I'm open source publishing tool <hes> which is really cool and they've done really well since then. I've been happy to see the growth there okay third up. We have Pitas Pete. Please reintroduce yourself to the voice by podcast audience. Pete may be familiar to <hes> some of you. He was on as a guest in episode seventy where we talked about the bixby launched last fall pete say Hello Hey guys this is Pete House and <hes> I will say. This is my favorite podcast. I listen to a lot of them so I am very glad to be here. <hes> I bid. I was thinking about this because you're gonNa have to introduce myself doing account. I think I've probably since two thousand five written probably a couple hundred absent only use voice start a Ni- V._R.. Space but more recently maybe two or three years writing for Alexa for Google and most recently for Samsung big outstanding okay wait am i. I they may first episode was episode six with John Kelly Oh John Calvi yes of bespoke in talking about acting and some of their development tools way back then that he was really focused on yes Johnson on a couple of times and I think it might have again on on this this summer. I gotTA reach out to them because what we've done is I think I've got him the last two years and so it's for a few the folks that I've come back to. It's nice to see their evolution over time and what they've been able to do each year. Okay so that's great. I think what we're GONNA DO <hes> folks for the listeners will follow along. We have a number of episodes that have been submitted by the panelists is their favorites. I think everyone submitted three to five. I threw a couple extras in there just for are fun to talk about so we've got a few different formats but we're going to start going through them and I up is is is a podcast episode that it is distinguished by being the longest we ever did. I can't remember if it's an hour forty five or an hour fifty but it's with Ron Krone. It's episode ninety four Ron for those of you are familiar and haven't listened to that was the original founding C._E._O.. Of Nuance and it took so long it was such a long episode for a few reasons one is interesting listing insightful guy but also we talked about nuance the biggest voice application company around three billion dollars in annual revenue. Just in you know selling voice software <hes> talked about his origin of identifying the technology building the team getting spun out of S._R._I.. All the way up through all the different milestones they did and then how more recently he's come back into the industry and has been participating as we're hitting sort of a new evolution and and after being out for ten years he took the company public chairman for a couple years and then started doing other things and so it's really interesting to have him come back and to give his perspective and how voice had changed and that's that's why I was so long and I really really enjoyed sitting down with Ron but I'm really interested P. Why did this one stand out for you? I am a sucker for good origin story so anytime those those come on for somebody like Him who's certainly industry veteran <hes> I was all over that one I actually back in the day was using new onset rhyme V._R.. Platform and scans off in this thing so I I knew about that world <hes> so this is fascinating to see him for from really nine technology background come into the space identified big opportunities and was really laying the groundwork and I know one of the questions they ask is. How do you take this from kind of being a science experiment to a commercial product in at the time? This is unthinking ninety four for that was in question never really been done before yeah they they were involved in a lot of firsts and one of the things I do like about that is there's a lot of people who are new to the industry and that's fine because that's that's what things have changed the platforms different capabilities different but it's always nice for people ever perspective on where this all came from amy. Did you have a comment yeah. I just wanted to say I definitely remember listening to that episode. What is well I know is I had to visit my mom at the time and I was listening to it as walking through the park in it <hes> was a was a really interesting long episode? I like the fact that he talked about. You know the the evolution of technology and then how nuance aunts was not the company that had bought what was the scans off scans off was the one that had actually bought nuance but they decided to take the nuance name which was interesting <hes> but I don't really remember exactly what <hes> what Ron's take on things today is did you can you kind of sum that up because I'm trying to remember like what is he doing with the Lord Mr Jones stuff whether yeah so first of all he's an investor now and he he uses Alexa he was intrigued by when it first came out more intrigued he said than he was with Siri when that came out <hes> and he's never seen a reason to use another smart speaker which is interesting because you think about some of the things that are out there now you know he probably would have been playing with all of them fifteen years ago by intrigued by them <hes> but one of the things he's really interested right now is this concept of artificial humans and and and now his I think the last twenty minutes of the podcast we talked about that as well and that's this idea of it. It's not it's not unrelated to voice assistance. It's this idea that you know we take someone like a celebrity. Maybe we we have an example of this like John Legend to voice for some interactions on Google assistant not too long ago but it could be this type of thing where people like interact with with that assistant in it sounds like John Legend or it sounds like some you know <unk> perfectly newly created solution and it has a personality and you can actually have a full interaction with them and some of the things he's investing in not just the voice assistant but that's part of it right. That's the brains behind it and the voice but then there's also avatars and other types of manifestations that can that can come out so I thought that was really interesting. A lot of people have commented on that part in particular. I thought you know be beat. Beat starts out with the origin story and I love. I love that part of it but a number of people on Twitter said Oh Jeez I love that part about the artificial humans at the ants glad we got to that okay fine Ron Chrome. We gotta keep moving along number. Two on our list. Today is episode ninety so again. We're still in the nineties. A couple recent ones in this was from both amy and Pete and it was about was Steve Arcana Vich so amy. Why don't you comment on why you put Steve Arcana Vich on who is best known for the Alexis Skill Big Sky? Yeah I thought that was just a really great <hes> introduction to to an indie developer and he was <hes> you know I know he's a very interesting. I mean he's very accomplished. He's a philosophy professor. He talked a lot about how he uses he thanks through precise language and using using his background as having to write clearly to help him articulate things in his in his skill experience which was kind of cool <hes> but you kind of see in him an into an indie developer that started from the very beginning did not have a background in coding taught himself how to Code and then found the perfect use case which is weather and at the time and even today the native weather voice APP is not that great the native experience so he got in there he did something and he was able to really leverage that in monetize ties it and I think you know one of the few that found a use case as an independent developer. That's really got something going there. I mean you know whether an an ambient sounds those are the those are the places where I think an independent developer right now can still <hes> or could in the early days gained some traction but you know now. The question is what what's going to happen. If someone for other ended in India developers now it's kind of harder to get in and he he was very forthright. About how many people are using his experience <hes> the the <hes> the ability to monetize it what he's doing to monetize it how much it's costing him to get like some of the radar information so it it was just very interesting to hear his perspective yeah. He was definitely Kelly an open book. I love that about it and he is. He is a very interesting guy. We had a really good conversation because we're both interested in the space. You know a lot of that was just sort of normal back and forth when you're doing some joint discovery together I love the conversation about him. Competing Against First Party <hes> the native weather weather APP and what that meant for him and how you can still differentiate even a first parties got some solutions in demonisation and some of Viz user numbers. I heard a why of a lot of feedback from listeners really thankful that someone was saying okay. This is what a successful skill looks like on the platform now pete. Did you have something to say about that. Yeah no oh that was a very interesting one especially as a developer in coming at it from probably a similar point of view but it was a little discouraged because he says he's not even a developer and he's writing in Ruby which is some developers are listening using now will will will understand. It's typically not how it's done but they have such a you know the hit hit skill and and to be so successful is really fascinating <hes> and he was really open about sharing that so that was just great here yet is funny. He built that ruby framework. I forgot about that <hes> because his the first developer he hired new Ruby and basically did some sort of weird translation with probably Java or python at the time yeah I mean he probably hired a web developer at the day at the time that was ruby yeah and then so then when he wanted to maintain it he had to teach himself Ruby Yeah. That's the not the conventional way to do it but I guess he's sort of an unconventional developer. Okay third up. We have episode ninety seven Amir Hirsch a he is the founder and C._E._o.. Or Co Founder C._E._O.. Of audio burst and Dave camp this was on your list at the top yes I I thought this was one of the most fascinating episodes that you had had <hes>. I mean this whole concept of <hes> cataloging all of the different <unk> audio <hes> sources out there starting being with podcast and radio. I think is it's paramount to the future of really this whole movement in the reason. I say that is really. I think the killer use case we keep saying what's the killer use case for this and what's the use case but I think it's pretty much staring us in the face right now. <hes> it's audio. It's audio consumption and <hes> you know I think that you look at the the rate at which <hes> just audio contents being produced with all the new I think I heard on that episode. He said something like seven million episodes are currently out there across hundreds of thousands of different <hes> podcasts and so this idea of being able to say <hes> to your assistant. Dan You know Alexa or Google Assistant Tommy about Tesla for the day Gimme Gimme an update on what's going on the Tesla and rather than just feeding you like what's what's trending on business business insider or something like that. It's actually feeding you. Snippets of different podcasts that people are discussing and I just think that's such an important component to this idea of ultimately getting to the point where you have this curated feet of audio content. <hes> that is real personalized you where you have people that are <hes> you're not even exposed to but your assistant is starting to learn based on your behavior that you might like it so you're finding all kinds of different content and I just I think that what Amir's doing audio burst is going to be the foundation in the backbone by cataloging hall yeah so I really like Amir story at a couple of different things I like about that. One is it's not just about Alexa skills and Google actions. It's a different way to use the technology in this idea of making audio content discoverable also as someone who creates audio content. We both do that. <hes> you know that's obviously attractive but I think if I think about the value proposition for voice assistant users. I think it's really important that we think about how we can how we can generate content. It's not just reading off a wikipedia page just a bunch of research on this like yes so many of these queries you get some sort of Wikipedia <hes> information or maybe news from like a print new source when there's all this rich rich audio content out there that could be a better could be a better solution. The only thing I like about is like a big idea. It's like to try to the scale of Google did exactly it's Google in this space is kind of a takeaway. I had different amy what what did you have to think about that. I just wanted to say that on that episode to me. That was a really good example. Have how you do such a fantastic job of like breaking down of very difficult technical subject so people can understand it because I remember there. Were you kept like drilling down and asking specific questions about how the technology worked and he would explain it and that's just you do a really good job of like ferreting out information so I just wanted to kind of bring that up because that was that was a great example of it okay well. I'm glad I'm glad that came across we. I'm I'm basically asking questions I want answered to the extent that that's helpful to the users our listeners so much the better okay so I want to talk dog next about what I call the design block and this is several episodes so let's see what I have on my list here. I've got episode thirty with Kathy Pearl Episode Thirty Three with Mark Webster Episode Twenty eight with Tim mcilwraith. Oh I'm sorry Kathy Pearl. She's now with Google was essentially at the time Mark Webster founded. A company called say sprang which was sold to adobe last year may was two years ago now it was it was a while ago <hes> Tim Maccari <hes> is search in the leadership over at discovery communications so discovery network those types of things episode forty as well Karen Kosinski who is an independent voice in solution <hes> user experience design consultant <hes> I would put Amy Stapleton Episode Eighty four in this group as well as Lisa Falcon episode three and Emerson sklar episode sixty six. I may have forgotten <hes> some of these episodes. Sixty six was actually Schley a compilation around design. We've really done a lot of voice you acts or voice design episode sense and because we hit so much content <hes> the one I did with Emerson where we I think we we went through like twelve different episodes will snippets snippets of twelve different episodes and in feedback people had around design but I want to throw this out there <hes> in and get your feedback on it because early on design was so important <hes> and there there were just so little information about design out there and voice user experience design in terms of the general conversation of how voice assistance we're going to be different than I._V._R.. For example or from chat bots which there was plenty of information on a couple of books and I just found that the perspectives that each of these people brought to the table so helpful in understanding often what not to do when you're designing experience and again some of the things ray practical applications of things you can do due to move forward in in in that some of that we've learned a lot about just the history of a- voice interfaces as well because so many of those people have worked for years in this space so who wants to jump in on this one. I got a couple of things to say about that <hes> because I I've lived in a world where the only means to interact with an application that I've written is just through voice and so there's just been some principals that I've always abided by one is <hes> if you're going to enumerate a list of items once you'd be on three. That's you gotta think about it a different way. It's too much cognitive load for the person <hes> so for sure. That's what one constraint that you would necessarily have with the screen on the other one would be internally generally. We always had a rule we call it. The one breath principle which was we would never <hes> spew the script of information that would take a human more than one breath to speak and if it was too long we would we would try to break it up in a way that was allowed the customer to the person using the application to Iraq with amy why I just always find it very interesting and helpful to hear people talk about what they've done so as an an example you know the <hes> Tim what's his last name the guy that from the Discovery Channel mccaw yeah the <hes> the the person that did the the <hes>. He's the the food network one with a new recipes. You'll yeah that was very interesting because he went into a lot of detail about <hes> some of the choices that they've made how they were leveraging the screen <hes> the philosophy behind multi-modal. I would also throw like adv eleven in there. She was another one that talked the the design concepts of her <hes> kids court so it's it's very helpful to hear what what thought process p people are going through and then what kind of decisions they make in their skills. One thing that kind of is striking me today day is when I listened to some of those older episodes about about design and and and conversation you I. I'm wondering what's going to happen now. That Alexa conversations is out there. I still don't understand it completely but there was a lot of discussion or thought around the there there are limitations to the platform now that are holding it back that people aren't really able to have these these conversational experiences the the the platform is more designed for transactional type activity but with Alexa conversations that is that going to you know. Is it going to be a completely new new paradigm so it's it'll be interesting to see what happens. Yeah Please Complete New Paradigm and this type of cross skill L. Experience where you're going to have different designers for these different skills being able to help with this fulfil this user need <hes> and how does that done without creating some discontinuity. What do you think Dave Yeah No? I I agree and I WANNA kick it over to you. What so kinda going off of Amy's question because I I was thinking the same thing? What do you think with Alexa conversations in your opinion? How does this change the whole equation in the most front from you know what they're doing today and how it's going to transform it well? Actually I've got a pull pete into this. In the second because Steve Our candidates was the first person that I heard sort of point out and said Oh this is going to move us to imperative programming model from declarative and so you know declared of being you very tightly defined all of the user interaction action scenarios whereas imperatives as to what the goal is have general constraints but then turn it over to something else to fulfil in this case the I will fulfil some of those and so there's I think there's two different things when we think about Alexa conversations one is the fact act that the I will do more and then the second is is this other things because there were two announcements by Amazon in early June one was Alexa conversations which is really this more imperative programming and leaning on the A._I.. For more capability and the second is is this crosskill Gulf film and when you have an objective that one skill may be can't fulfill you might tie several together and so I think the answer is different for both <hes> but I do believe that this is more more similar to what bixby already has in the market and Pete since you've worked in bixby. Tell me if you think that you know my high level assessments correct and what your interpretation of Alexa conversations in the cross skill Gulf filming us yeah when I heard that announced I kinda I kinda smile because I thought boy that sounds awfully similar to the way bixby's approaching the problem in you know from somebody that came into bixby last year. I looked at that platform end. It's it doesn't resemble anything like the others right so if as a developer <hes> you basically create a Web Hook to respond to really you know there's intelligence on the Google side or the Amazon side and they will give you a in some always a dumb message will just say this is the intent and please fulfill this request <hes> if you look at the way others like big spear approaching it <hes> I it's really an A._I.. Platform so you get other features as a developer like <hes> <hes> preference learning like selection learning that she really don't have on those platforms and the way they're these deconstructing it so to say you know we're GonNa Maybe get away with these invocation names and change the way that's interacting. I mean to me that's that's the way to go. That's that's the way to scale it because if you want to have an ecosystem with many third party apps you can't you know have these implication names in trained people to you know once you get beyond five of them. You're kind of dog hire pages we can only we can only remember the the icons and are the mobile APP icons on the home screen. Maybe once green to the right swipe this idea that we're gonNA remember hundreds of specific invocation names I think it's probably not likely you know certainly people be able to learn five to ten <hes> seven being a magic number for memory of these types of things but <hes> so I think that's really interesting in the idea here of really being more <hes> intention <hes> oriented in like us a term intent a lot and in invoice invoice assistance this idea that a consumer has an intention and that intention is to complete a goal. We'll might be to listen to music. It might be to get a movie ticket might be to plan a night out and if that intention involves either information or tasks that are not all contained in a single skill then the question is how do you do that without creating this choppy experience yeah. I'd how do you orchestrate all those services. It's a very difficult problem and so and so I think Amazon's a long way from getting. Marriage I think it's really interesting. They SORTA pronounced it <hes> because they look into get into a preview period. I don't expect this thing to general availability for a long time. <hes> yeah certainly into next year before it would be G.. A. would be my guess and then it's only going to be a couple oh scenarios and then there's this tie between the two but I think really what they wanNA. DO IS THEY WANNA start seeding new development into the model the imperative model because it's less work on developers alpers it is hopefully will deliver more intuitive interaction so that not every developer needs to be a sophisticated to develop a good experience because Amazon and the Alexa Platform can actually smooth over some of those parts based on what it knows <hes> and so those were important and then eventually with his crosskill goal fulfillment. They'll all be prepped for it because if the if you've developed something in the existing model the old declared of model. It's probably not gonNA be discoverable for some of these cross skill capabilities so so I think it's really fascinating. I love what's going on. There and I have no doubt there will be implications for design in fact. I think there's going to be more thought about design when this comes up because if you're doing imperative model you have to think a little bit more about what the platform might do for you <hes> and you can think about the things you definitely want to be specific and you can really hone in on those without having to deal with all of the ancillary things things that you might normally have to think about which really just more table stakes that they even just now let the I elected in this case. Take Care for you right and and how do you <hes> you know in the future scale these conversations to to last more than motor two returns. It's very difficult to go beyond that as a developer now okay so Alexa conversations for those of your engine that we did a re Mars episode just a couple ago. I think it was one. Oh One maybe <hes> and I recommend you check that out. We've got a bunch of people talking about it but is amy said. There's a lot more discussion to be had about that going forward okay so let's move on to <hes> sort of what I would call the technologists block. I've got a couple of them here. <hes> the first I is episode eighty six Fiji Balasubramaniam from pin-drop and Dave you hit list of this is one of your top three. Tell me why so along with <hes> <hes> kind of along the same vein as mere Hirsch. I look at what V._J.. Bala- Superman Ian in pin-drop are doing as another really big idea. I think it's going to be one of the core technologies to all of the voice technology that's coming in in his whole approach is around security and just listening to that discussion was absolutely fascinating. You know the fact that they started in fraud detection and then now they're moving this into basically voice biometric gene and so being able to use all of the different <hes> indicators in your voice to identify you. I think the implications there are pretty profound for the Internet of things <hes> I think that if we think about the security aspect aspect of say like a voice activated walk the two factor authentication system that he described where it's your voice coupled with the device that you have on your person I think makes a ton of sense and you know I think it moves moves way beyond that though to all kinds of different applications like I see a future where you have these Nanno payments that you're constantly paying for things and it's all being authenticated through your person through your voice ace and so that technology in general I just was actually blown away with and in really fascinated by the implications of really being able to use your voice as your own personal key and have all these different the systems be integrated with pin-drop security to identify who you are. <hes> you know in all these different kinds of settings p. This was also on your list. What stood out for you? I love this one. <hes> this was one of those I think I listened to two times in a row because we're just so much content in their <hes>. Two things that stood out was I did realize there was over a thousand attributes that are not related to voice that he was using crazy attributes. You know the switch <unk>. It just isn't something I ever thought about but while fascinating and the other was voice aging the challenge of you know you can print your voice now but two years from now it's it's changed and aren't six months through now fairly yeah wow that's right or if you're the president I guess even faster because of the stress level and there's this fascinating yeah. I love that in this idea that I mean you brought this update like this this idea of like how do you identify the voice in what I immediately went to is this concept of we should be able to go up to any device that has a voice assistant that we've interacted with in the past so let's say Alexa Google assistant which we've all tried everyone on this podcast anyway his tried and we should be able to go up to that anywhere anywhere and it should not be in our house. It could be in a hotel room. You know we're starting to see people hotels. Come out with this idea that you can sort of personalise. You're you're smart speaker in your room but anywhere we go a smart T._v.. Somewhere like in the resort at a friend's house. Why shouldn't it be able to identify US personally so that we can actually have a personal interaction with it no matter where we are as long as it's common voice assistant and it might be all of these things that the pin-drop is doing that allows us to be able to be done so maybe the two factor would be a useful thing but also being able to look at these other signals is Pete you mentioned to determine where we are in is an appropriate place for us? It makes sense as that lineup with our G._P._S. coordinates. Maybe they're being shared somewhere. Something like that Dave Yeah. No I mean to your point like that's that's the Big Aha moment was for me. Is that technology it really does allow for like true through personalized ambient computing like if we're moving toward a time where all of the like everything has his base smart assistance embedded into it and it just really it's like a commodity is world that we're living in where it's just massively proliferated in. It's everywhere <hes> I think it's really really fascinating to think about the fact that through your voice <hes> it immediately recognizes that it's your account in in there for everything becomes personalized whether it's your own owned properties your home or your car but like you said like when you when you travel you know it's going to recognize you whether you're at a resort or even at a restaurant or wherever it might be like it was a really profound conversation. The implications just had my mind go in like a million different directions all right so <hes> V._J.. Balasubramoniyan pin-drop security very interesting next up. I want to bring into the conversation Jeff Adams. I can't remember the episode number P._O.. Remind me of that but you through that on there and you said earlier that the Ron Krone episode. You're a sucker for an origin story. There was a great origin story with Jeff Adams. Yeah episode sowed fifty nine excellent origin story. I highly recommend if you guys want to really get to the roots of especially how Alexa kind of started with their platform you have to listen it episode because the one thing I took from that <hes> <hes> he certainly has a rich history with nuance places but <hes> it was you know when people think about this era with smart speakers of things they always think of the how good the speech recognition is and then how good the natural language understanding is but the third component it is really the the far field technology that allows somebody to speak across the room at understands you and sign with highly recommended up so it's just fascinating because without what they came up with we might not be talking right now yeah for those of you haven't heard it so Jeff Adams is he. He's been in the industry since the nineties. He's he's a technology. I think he's trying to think as backgrounds. He certainly has a C._S.. Background I think is electrical engineering as well but he started working at some of the the top voice speech technology companies as C._T._O.. Back in the nineties and sort of carry that through and he was with a company when he called the APPs it was acquired by Amazon in two thousand ten or eleven remember exactly but that became the foundation for the automated speech recognition for Alexa and great story about how they locked him in a room for today's him and his team and then tell them why they acquired and all these other things in eventually they said Oh we're going to create this. Thank all the smart speaker and it's going to be able to identify different people speaking from across the room and he and his team said Oh you can't do that. They said well no. We're GONNA have to do that so let's figure it out together and so he walks through that and it just comes brings up this point with whether you know what's the bigger invention over the past several years in voice assistance. Is it actually the the I that the voices isn't got better in the in the deep neural networks made the better or was it. The fact that the far field microphone actually started to work and introduced this entire new set of use cases around the hall with Fairfield are field mice in devices across the room. Yeah it's that an instance the implementation of the wake word how they do it on device. I mean it's just you know you look at it and just say well that yeah that's the model that's the way it should be but somebody had a bet that absolutely absolutely okay so I want to go to another concept. More in this is around compilation episodes so compilation episodes are generally wear eva or I will be on cited an event and we just conduct a series of back to back interviews usually somewhere between five and ten minutes long and we did the first time hundred we only did seven of those. I think the first one was certainly within the last nine months but they seem to be pretty popular in eighteen. You put these on your list or two of these on your list C._A._S.. Episode Seventy Eight and seventy nine. You know tell me what you thought about those in general around these compilation episodes yeah I mean I I love those because it it's a it's a window into what's going on. If we're not able to go to the event I went to C._A._S.. Back in two thousand seventeen <hes> it's not something I go to every year but I thought you did a fantastic job of of getting many different aspects of of the the voice industry on the podcast and you were able to talk to Dr Lee at Samsung and he talked about their their vision for the <hes> the connected life and this was sort of before the big push to get developers onto big speed he maybe alluded to that a little bit <hes> you got the <hes> Roger Zang thank from Alibaba was talking about team all genie which is not something we hear a lot about over here but it's really interesting to get his perspective and then you had add some smaller companies on their like ones that are trying to create a devices for for children children's content <hes> smart shower Microsoft so it was it's just great to get that kind of information <hes> without without having to actually you know go to the show to be honest yeah fair enough so episode seventy eight was eight interviews with small innovative companies and it was part of the showcase on Sunday night sort of before things really kickoff off in earnest their E._S. and then episode seventy nine was mostly big companies are a couple small company SORTA sprinkled in there <hes> but these were tended to be more established businesses everything from Samsung L._G.. <hes> and you know to sensory and a few others that was a that was really fun. Dave what's your perspective on these compilation episodes or the seat. Yes ones in particular. I love him. I agree with amy. I think it's great that hey you can go and do Vegas for us and Kinda report back. Everything that went went down there and <hes> it. I mean there's a lot of really interesting. Insight like I remember I think was at I fly tech. That was the Chinese company company that did all the live language translation yes yeah that company was really really interesting to me. I mean that it was a it's a Chinese company so it's the impetus in Chinese but it seemed like they were pretty far out ahead in terms of of <hes> the type of live language translation that they can do <hes> but in general. I think it's just really cool and it's Nice because what I've noticed that you tend to do is you'll get you'll interview a bunch of people which I think inspires later episodes road <hes> like I think it was either Amir Hersher B._J.. Balasubramaniam that you interviewed at one of the compilation episodes and then you brought them on for a full interview so what what's great about that. Is it kind of gives everybody a preview as to like okay so so these are really interesting people and I I again. I forgot which of the two was but I remember when okay I remember when that came out. When the actual episode came out I recognized him from the compilation episodes so it's nice as kind of almost a teaser for later later episodes yeah? Let's get our ten minute. Exchange was so good. I was like Oh Jeez. I didn't really realize what I was scratching the surface on here. We need yeah. He was very kind to give us some time. <hes> that's great. So what do you think Pete you've listened to. Most of the podcast probably episodes I mean where do you put the compilations like 'cause most of the things we do with these one on one interviews we also have the panels were easily talking about either <hes> a specific topic that we've done research on region sort of the year end episode. We don't have too many of those who probably had you know eight or ten of those. <hes> you know how do you think about these like the individual ones for the panels versus like doing today or the compilations nations which are just one interview. After another. I mean I would probably say a my favorite style is the you know the single interview but I love the calculation ones because <hes> you just get a for example you just WanNa Google I o and just I I think Nick Schwab is a or somebody and just hearing these perspectives not only from you but on really smart people giving perspective on the conference. This isn't something that would never really get A. Let's listen to something like that <hes> I i. I actually have a question for you. How many of these conferences to co two every year? It seems like you're not everyone. <hes> I've been to more than I planned is probably the best way to say that so I think yeah I've done compilations. My first one might have been at the Amazon event last year where they did their product launch in September and they did twelve products and then they did like seventy other announcements. Sousse insane and I just got some journalists on and I did that was popular. People sort of like that so I did it see yes I've done I did too at the voice of the car summit which Score Publishing Bradley Matt Wreck put on which it was really good. <hes> you know is able to talk to some big companies and small companies air that was popular <hes> I just did remorse. I think I did another one too. It's hard to remember Oh and I've got one coming out like actually by the time this publishes wishes will be another one that came out of just to meet up in in New York City because I was going to be there and was able to capture five interviews of people who attended to sort of give the perspective of what was going on at that event so <hes> I guess I guess in the end and a lot of these events I too but the intent would be to go to no more than one every other month but it seems like it's much more frequent than that yeah they're great. I think that's how we really originally met last year. It was through. It's true we did you were part of a compilation ablation fact that was probably my second compilation episode <hes> because we had you roger and Joe Murky <hes> yeah because you know we have said that Samsung developer conference and I was able to snag <unk> Adam Chair. We'll talk about a couple of minutes which was great because you know I hadn't really known that I was going to be able to do that and then as Geez you know we've got some people here who really have a perspective on this. We need to share those thoughts and so that was great two and that was yeah we met the I met you and then two hours later. You're on the PODCAST. It was great yeah and we're starting amy any thoughts to wrap this part up yeah. Just I think the hot takes at some of those events. <hes> like Pete alluded to are really helpful when you were at Mars and then you know the Alexa conversations news came out to get the the thoughts of some experienced developers on how they were digesting the information that was really useful so that's one of the things about the compilations that what I think is helpful. That's great and I think that's nice too because they tend to be more like articles and not everybody has time to read all the stuff so it's a lot easier where they can sort of get that heartache so so that's good we we definitely get a lot of good feedback. I think we'll we'll wind up. Doing it. Certainly is many I don't know we're going to do a lot more but <hes> because we tend to focus on these one on one interviews and so I want to transition away from that and I want to talk about what I would call sort of the platforms we've had we've had several <hes> but I want to talk about a couple in particular that we had that were very popular so the first two I want to talk about our Paul Cut Senior from episode seventy seven David's Bisky from ninety five live <hes> they are both from the Amazon Alexa team both of those episodes now they sort of cheating because they're both this year and we have a bear audience last year but but those are both in the top ten all time <hes> and you know Paul was able to talk in some dapper around what developers are doing on the platform with their with their not doing but should be which I thought was Great Day. We talked a little bit about developers and his background but we also talked a lot about brands and how they're taking advantage of the platform or making mistakes if he was pretty open about that so just along those lines like did anybody on the panel. Did you guys have any perspective on those Alexa conversations. I had earlier this year Dave. I thought that I mean I think it's so cool that <hes> I'm I wasn't I was in high school when the mobile era really came around so impressed like podcasting wasn't what it is today and so oh I just think it's really cool to hear someone like Dave or Paul. Be So candid about you know just like the whole developer community and transparent about where they're at with technology so for me like <hes> <hes> and I'd say the same with Adam schefter like just the access that <hes> like sharing these leaders and like chief evangelist <hes> speak to what's going on with the technology is extremely insightful invaluable and I just think it's it's really cool like that. You'll you can parse out where the technologies at ask questions about. You know like the way that the changes are being made around Alexa conversations or something like that so it's for for I think the audience it's just really really cool to kind of get that type of access to these people that really are in charge of the decisions of how this technology is being built. I totally agree with Dave I mean for me. It's just it's important to think that these people are still around there accessible in a way. It's kind of very cool that you know Dave is still doing the evangelist thing although I think maybe he said his title changed or whatever but you know the fact that he's still out there going join to conferences on his accessible is not necessarily doing the office our stuff anymore <hes> but Teesta you know he's he's a person that will really inspire you and help you get stuff. Done and Paul is the same way I mean. Paul is just very supportive advocate for the third party developer. I think that might be missing from you know some of the other platforms but for me the the that kind of the you know that kind of leadership is just really important. Yeah I was just GonNa say to Amazon's really made a huge investment under evangelism team. I think they just do a wonderful job. If you think of the last three years really really after educate on many things that are completely nude everybody and <hes> David came from other companies like Microsoft so you know this isn't their first Rodeo so they know what it's what it takes to really build a platform developers for Sonya. I will give you a little inside baseball here too. I asked Amazon for eighteen months to have someone on the pet podcast now. I didn't ask them every week. Yeah I'd ask periodically. It'd be a in Davis tapings skin. I talked about it several times but I think early on the the P._R.. Department maybe a little bit more wary of having people talk. I'm podcast and stuff like that. Particularly People Apple's sort of know what's going on but wait ask questions. It would be uncomfortable but I will say they've been great this year. I've had Amazon. I think three times now I've had those both <hes> Paul Day but also Aryan Walker Walker who runs the auto team and so you know Kudos to them for for being open and debuted one more comment before I move onto the next one. Will I just wanted to ask you. Why do you think that is I mean I've noticed that your podcast in particular color? You're getting like all the different platforms coming on here. Do you think that they're noticing that by being really transparent and open to the type of conversations in questions that you ask that that it's actually really beneficial bill to have that level of transparency. I think everyone I've talked to they. Recognize the benefit of it at after they've done it so you know whether they know there's benefit there beforehand or not. I don't know and I think part of it is we do this long form <hes> format where we talk about a lot of things and we go into some depth and that makes some people in the P._R.. Community uncomfortable because they're like Oh. What are you gonNA talk about right? You know if you're gonNA talk for an hour is am. I GONNA get in trouble or as you know something. We don't want to come out but you know frankly I think part of what you're what you're angling towards his right. This openness is really helpful. In in actually the the event that Amazon had last September September for their product launch. They had a lot more journalists there than they'd had the previous year they did something which was a little more contained was much more open. This last fall in in some of the people on the the media relations department were excited because they're like we're finally family. We're we're you know this. I won't say we find this is a change. You know for Amazon that it's trying to be more engaging more open share more information. I think it's very different because Amazon five years ago is very insular didn't really talk much the media at all and they're in the sort of learned how to do this engagement and <hes> and I will say I've had other news with them too I forgot about they set up some interviews for me remarks as well. Some of that stuff won't wind up in the PODCAST. Ask but you know for other purposes so I'll say they really want to do that now. I will say with what we do. <hes> it's not so much we're asking him opinions about sort of events of the day or the type of thing we're asking them to tell their story and we're asking thoughtful questions that we know our audience will care about and we have a pretty big audience in the people that who listen to voice by podcast Amazon Google Samsung all these platforms really care about reaching and making sure that did they hear some of this information what they have to say and even respond to the tough questions when they come up okay so great question Dave <hes> hopefully it's perspective for folks but I wanNA talk about Google and I haven't had Google on for <hes>. <hes> two years not quite over maybe twenty months because they were so good to me early on Brad Abrams was episode fourteen still one of my favorite episodes <hes> sort of the first big company that made as someone available to me and Brad Abrams feed oh no he is you know at the time and he still has similar type role but oversaw the Google assistant engineering team and so if you think about it that was about it was less than a year maybe nine ten months after they ahead formerly released Google Assistant to the marketplace and there were so many early learnings that he shared we talked a lot about search so much depth on how Google was thinking about search on voice assistance and and you know having that perspective now and seeing some of those things are the same or evolved. Maybe changed better word. <hes> is really fascinating and then also <hes> some of the things they were thinking about in terms of the way they were gonNA enable experiences across surfaces. I really excited about that and I think if for a lot of people I would say that maybe the audio quality isn't so great maybe Pete you may have mentioned that from at one and I'm not really. I don't remember exactly why wasn't great but I will tell you. The content is excellent and you'll get a perspective perspective there around Google you'll be able to see some of the things that they've done you build under <unk> you'll be able to learn <hes> what may be as a change in their understanding of what is just really a an extension of where they originally were headed. So did you guys want to comment on Brad Abrams. I don't know if you've listened to that one. Yeah I'll just say I've actually known who Brad was before. He worked for Google Super Smart Guy Obviously and I just think that Google is in a very interesting position for for many reasons <hes> certainly from their distribution. <hes> you know the quality of what they've produced <hes> so they're they're going to be around for the near future for sure <hes> certainly in the space <hes> certainly certainly <hes> if you talk about voice search I mean that makes perfect sense. Felon the focus on that I don't I don't have anything for the Brad Abrams episode just because it was so long ago. I don't really remember it but bread. I have a question for you. Attending all of these different conferences <hes> the Google I o this year really really stood out to me. I thought that <hes> Google on our duplex on the web was fascinating than also the assistant to point out so just <hes> kind of off outside tangent here are how are you thinking about Google where they stand right now. <hes> I would say Google fascinates me it the breadth and depth of what they're doing a lot of that is an obvious unless you're really looking at the stuff on a day-to-day basis and sort of understand where the limitations on the other platforms are did is Google duplex with something that everyone sort of understood stood was important in different so that was last year and that was that was a good indication of the fact that Google has a lot of assets here <hes> that maybe some of the other players don't have at their fingertips so even though they came to the market late <hes> they might be able to catch up in leapfrog in some areas now google duplexes had some issues. I think you know a year later. It's maybe not what it was portrayed to be still significant forget advanced but maybe not as spectacular as as initial is initially tried to lead on so duplex for the Web I think is interesting that goes to the idea that they've got this machine learning and they can apply in many different ways and and I liked and sort of automatically the forms and those types of things <hes> but I will tell you the thing that most impressed me was the on device this year <hes> in partly was the on device performance that the low latency the potential for security benefits fits that we've seen some of that with apple without device with Syria but we haven't seen anything that extensive and the you know having had an opportunity to speak with some of the senior folks in the Engineering Group there who are are different than the other folks that we might deal with on a day-to-day basis in the voice industry <hes> really fascinating what that could lead to over time and it starts to make you think that what we've been doing over the last couple a couple of years as my friend Jan Gilan Might Sane is soon gonNa seem quite quaint <hes> so so. I think we're I think we're we're moving to a place that <hes> we're. We're GonNa see the things that you know. The training wheels approach that we had over the last couple of years is just GonNa get blown away. You know these things really metastasized okay. I thought real sorry real quick. I was just GonNa say I thought it was ironic that the <hes> you know this whole privacy debate because I like you that the whole notion of moving it from the cloud to on the device kind of eliminates the whole well to a degree but it definitely changes the narrative round on Google and <hes> privacy it doesn't they had a great headline about facebook talk his talks about privacy Google just implemented it. <hes> and there were a number of other privacy announcements at I._O.. Were part of that story. I think it was verger tech crunch at that had headlined <hes> but I think the timing of that for them was perfect. <hes> you know we've seen <hes> F._A.. From facebook Julio. And <hes> Apples W._d._c. all with a significant focus on privacy in the middle there we had Amazon announcing some new privacy features voice <hes> deletion of Records and the the privacy center in these things so all the platforms are really really zeroed in on the fact that this is an important topic <hes> maybe a Zeitgeist of the moment that they need to be out in front of and so it's GonNa mentioned to see how much of that actually takes hold as opposed to just announcements but I think it's fascinating all right so we do need to move on. <hes> got just a couple more things to do amy. You talked about something that interested you. SORTA listening back on dozens and dozens of these episodes is the evolution of monetization and you've pointed out episode eleven with that Media C._E._O.. Pet higby <hes> episode thirty five with Stats Shinsegi from dramatic episode thirty seven was facilitation Karanko from the former storyline <hes> which became a vocal and then recently shut down and one of the few <hes> departures from the industry or demise is the we've we've had the opportunity to watch end to end and then I'll throw another there episode sixty without a chick from Alpine day I and previously a voice labs the the sale of that business to head space each of those companies had a very specific perspective around monetization. Some are continuing to implement those what was why. Did you put those on your list. Why did you think that was a really important thing to bring out in terms of the portfolio the back catalogue voice about podcasts well? I mean someone that's trying to build a business. I'm always very very intrigued by what people are looking towards in terms of a viable business model and I think because you have such a portfolio of podcasts over you know such a long period of time now <hes> we're we're kind a seeing an evolution and you know I mentioned that the first episode that I listened to was <hes> Gioja Quinta number nine and I think even in that episode he was talking about his frustration as independent developer that <hes> it was very difficult call to to have any kind of monetization. This was of course before the ice the ankle purchases were our you. He's still frustrated yeah <hes> so I think this would be a really interesting. He no longer discussion but just to mention it here for in terms of the <hes> the hundred episodes I do think that we see that there's we still haven't figured it out and there is the there was early on the thought process that we could potentially go more towards an ad ad model because that seems to be what's working on web on mobile <hes> but the add model doesn't seem to fit very well with voice and even Steve are kind of shoes having some success with it. I think he mentioned when he talked about his experience that his his <hes> his listeners who really want to get the weather but they you know they're they don't like to be sold to he's not fat so there's some issues with the I._S._p.'s <hes> and then story line is a very interesting one but that would be a longer discussion because I think <hes> <hes> the Sealy. I can't remember exactly his last name but he you know he had Shinkansen Yeah okay he and he had this vision of how content creators the voice was going to be the youtube of content so his mantra was let's provide tools to the content creators and they will be the ones that are going to develop all these these assets that will make these devices worth buying and listening listening to any kind of came to the realization that well it's harder to design a good experience than maybe originally thought and not only that but there was some skepticism about the the readiness of the technology to have a full experience <hes> and so that business model got shut down so I think we're in a place right now. Where as things have evolved? We realized it's it's hard for ended independent independent developers to make money through <unk> purchases but the ad models aren't really allowed or working so where do we go from here and I and if we look back historically <hes> it's interesting to see the evolution but I'm not so sure that we that it points to to a direction for the future yeah. I think that's fair enough so to give people perspective so that media was doing an ad model on mobile with voice before Alexa was even launched and so they had they existed in the voice space based before the skill world they actually shifted their business to more of a subscription model around helping companies the launch skills and manage networks of skills and <hes> in in that's that's been a big success for them due to have pat back on the podcast at some point <hes> stocks with instruments. They're they're actually implementing a similar ad model where they insert ads <hes> in different types of venues at Amar Marcek from voice labs originally had an ad network ran it for several months on on Alexa was quite successful. Amazon was fully aware of it than they early in early. May of twenty seventeen changed their policy Nelson said No. You can't do that anymore. <hes> so that was sort of interesting so ads have been SORTA shut down. I think what we are seeing what they say is that advertising works on the web but we've also seen subscriptions work on the web for certain types of applications <hes> certainly on mobile. There's at there's an advertising model but Framingham is is probably the most successful subscription didn't really work very well there. When we come to a voice mostly what we're doing now is trying to push subscription models or transaction models? <hes> you know I think the one time purchases of of the content or or software probably are not. I don't see very many of them so I don't know if that's an indication that they're not successful but people have tried it. There hasn't been a lot of that seems to be more of a push push towards subscription. I know that's popular. There's there's there's a premium concepts to subscription which Steve Arcana Vich implemented which seems to make sense and then there's that transaction concept which it oh best buys trying to implement a Walmart <hes> <hes> <hes> easy needs <hes> you know sort of the wholesale bulk sale side so I still think that idea of the transaction model of selling things in either taking a piece of the transaction or making your margin for selling goods you have love is a play and I think subscription is is a question right now. Whether that's going to be successful I will say that there is definitely advertising on the general purpose platforms like election Google assistant you find it and things like podcasts or flash briefings Sir <hes> other types of things at news media traditional media put out there <hes>. It's not necessarily generating a lot of income for anybody yet but it is something that that is viable in does seem to be working for a number of publishers assures Dave or do you have any comments on modernisation. You'd like to weigh in or should we move onto the next the final two episodes. I mean Miley quick comment would be <hes> you know. It's just been frustrating <hes> to try to monetize especially on some platforms because unless you get the attention <hes> and be featured on these <hes> through discovery. You'RE NOT GONNA get a lot of usage therefore it's hard to even entertain doing premium model or something like that so <hes> you know I know some people are making money from Amazon rewards but it's clear that subsidize GonNa. That's GonNa go away while Yes Amazon rewards has definitely been they most most lucrative revenue stream for anyone in her voice space so far. It's very tangible okay so that's that's. That's a great conversation on modernisation. You know amy. I'm glad you brought that up because it probably deserves its own focus at some point. I I still feel like things are maybe a little too ambiguous at this point to have a really good episode around it but <hes> that's something that will put on put on the list to focus on coming up okay so the penultimate ultimate episode. We want to talk about today. Episode Eighty with Brian Romley he made a presentation the first presentation on this topic at the Alexa Conference in Nashville Tennessee in January of this year and he outlined his concept of the intelligence amplifier and the wisdom keeper which are two separate products so dave you you're at that event you spoke at that event <hes> expensive time there and you have some specific thoughts what's on that episode and what Brian Head Yeah <hes> I thought it was. It's obviously really really fascinating. I mean we were talking a little bit about this before. We started recording here but you know Brian is he's real out in front with the technology side note the way I got involved into this whole space was I I actually stumbled upon Brian on twitter and that's kind of how I got roped into this so I've been following Brian for a while and <hes> you know it's really interesting. It's this idea that you would have I think his personal assistant that he's built using a ton of different raspberry pies's Alfred and it more or less it serves as like this your own personal intelligence amplifiers so it's like you take it and it learns everything that you are learning but it remembers everything like we can only remember so much and so it's Kinda like this own personal personal memory for you and and then it would be kept in the wisdom keeper in so I mean yeah it was it's a real <hes> a very high level macro sorta conversation and and <hes> video that he presented there so I mean you could go a lot of different directions with it but it was definitely something that if you're involved in this space just to be aware of because it's it's an interesting thought of where this final go the I mean Bryant as a visionary. He's always been a visionary. I think some of you know he he tends to be way out there in his in his thought process which is good because it pushes the envelope so you need that I think sometimes sometimes <hes> it bothers people because they don't necessarily see it as being very practical and then if it's not practical they're like you know. It seems like it's just more a very very fluffy. I I enjoy listening to him. He's always an inspiring speaker. I for me one of the issues with with his vision of recording everything this is that I'm not sure <hes> if you recorded everything that happened all all the conversations you had all of your your interactions it it <hes> it doesn't take into account the need for curation because recording everything unless you had something like <hes> you know audio bursts to go in in there and and help you curate the things that were that you were looking for it. It would just be overwhelming and it's like the person that can remember every single thing from the day they were born. I mean that's really not a good capability to have so <hes> I liked the discussion. I think that there would be more morineath to kind of figure out how to make some of this stuff happened practically and and <hes> visionary is good but you you need to bring a product to market at some point. Yes Brian bring a product to market Pete. What do you have to say the BRYANS grade? <hes> in a lot of your guests will talk past and present he certainly very future tense and who referred to scientific scientific things that I just don't hear a lot of guests <unk> towards like <hes> you know the through print of your brain how many bits per second you can comprehend. This is kind of interesting stuff so I I like him as a guest yeah bring something to market by unless you do yeah so the in in fact I I keep seeing things I think like a I don't watch Black Mirror <hes> but I think it is saw something that's rely that similar to this idea wisdom keeper or no the intelligence amplifier and like this trailer that just showed up on my netflix because I was trying to scroll through when the other day <hes> but it is really cool. I think amy that the concept air is just as Google curates the web and tries to give you using agents tries to give you the best answer <hes> it's for what you want. The the idea there is the curation is flat of that in any given time by asking the question or maybe eventually thinking the question but in his world asking the question that it'll surface the right memory Marie and and then you can then stitch together the context. I think one of the most interesting things about it. Is this idea that the wisdom keeper I think is really cool in that. <hes> some other people have tried to do this or have done facsimiles of this us with chat bots around this idea of compiling all the information of that someone is said or thought most types of things and then be able to query it later after they've passed away or you know potentially in this case if it's if it's a real time learning solution it could actually it could actually be you somewhere else or affects emily few somewhere else so you could be in multiple places at the same time. This actually does not seem to be that far out there to me. I think it's thanks reason why from a practical standpoint from an applied innovation standpoint I don't see anybody doing this soon but it seems like a logical progression that someone will want this type of tool and whether it looks exactly the way Brian talking about it or not I don't know but at least portions of what he because he's got a very holistic vision portions of it. I think are definitely gonNA GONNA come. Amyot to final comment there. Yeah actually did see the Black Mirror episode. I forgot what it's called but molly Cyrus is in it and she's a a singer and she has this this <hes> Alexa type doll that she sells and her entire brain and personality is uploaded into the doll and then when something happens to her that you know the the the the information the doll is is very useful and I won't say anymore because there's a spoiler but it is. It's exactly the same concept so it's pretty cool to watch yeah. It's it's funny. I mean there's a lot of these things we just did. <hes> did a thing invoice insider a few weeks ago about the knowledge navigator does around W._W._e.. See this is something from. I think nineteen eighty six from. From apple which is basically a smart display concept that everybody wants today with really smart assistant that can just coordinate and do all sorts of really complex for we're nowhere near the knowledge navigator from nineteen eighteen eighty six that apple is trying to sell apples probably the furthest from it but a lot of these ideas have been out there. It's a matter of putting it together and getting the right momentum behind the idea to bring it to market just like you know the the momentum behind the smart speaker probably wouldn't have happened without sort of the the strong conviction of someone like Jeff Bezos Okay so final story our final episode one of the favorites. I certainly got a tie search social media probably the favorite episode last fall. I sat down with Shire episode sixty nine Dave and Pete. You both put this on your list. <hes> tell me why start with you. Dave well like I said earlier. It's just unbelievable like the Times that we live in today to where you can get one of the founding fathers of technology like Adam Shire like hearing. His story was really really fascinating another S._R._I.. Guy <hes> who's been I think he said he's kind of at this for twenty five years and what was really fascinating about it was that he sort of envisioned. It's like he didn't envision the web or mobile but he envisioned voice in the Smart Assistant Listen epoch and <hes> so I just thought that was really interesting and the way that you know he has his trajectory has gone you know starting with Siri and then leaving apple in moving to give the now at bixby but the thing that arms are Samsung <hes> the thing that really stood out to me is when he's describing the idea behind a capsule in he's saying you know it's the <hes> it's not necessarily about like voice as much as it's it's about contextual assistance <hes> and then he kind of walked through the different <hes> examples of how that applies so he says you know you in this is something I've thought a lot about but it's this idea where you just say. I'm really the job to be done so you say okay. I want to plan a vacation or I'm I yes so help me out in plan this vacation for me and the way that the assistant works on the back end is it's brilliant navigating. All of the different APPS and edge was crazy to me listening to that story. Is that it it's very very obvious. That apple looked at that at one time. Maybe when Steve Jobs was still around in said this is the evolution of the APP store because has it you you hear him. Describe in really what it is. It's interconnecting all of the different bits within the APPS <hes> to make a ultimately fulfill the job that you're looking to do and now that you know they're doing it over at Samsung <hes> and I think Alexa conversations was probably pretty inspired by what bixby's rolling out <hes> or at least maybe to a degree <hes> so I just thought it was really interesting to hear his vision and understand understand this the bigger picture of kind of where the guy that really kind of lead things off in this space <hes> his ultimate vision for it yeah <hes> same sentiments <hes> in order to understand bixby Samsung you have to know who had entire isn't where he's coming from. You can look on its website and see demos from the early nineties where he's really thinking about how to solve for these things and no his vision of a assistant if you hear him speak it's it's a little different than other people. You know when you think about the stack there's really in my mind. There's three components. There's the speech recognition there's the understanding but then the third apart which I think we're still working out is <hes> you know the fulfillment of the do engine right. How do you fulfill those requests and I think if you look at his approach it's so dramatically different because what he will his team him what they do is the last few to model data in so describe exactly this universe of data that you want a hopefully talk to manipulate in sign language to it and then let an A._I.? Engine figure out how rounded service at requests I mean it's dramatically different and that's why I think it's superior platform. It's it's a learning curve now because it's hard because it's just so different <hes> but you know you got to think in terms of of context I I think he knew he understands as personal assistant so all these all these attributes like <hes> you know your previous interactions are important. I'll just give you one really simple antidote where I think some people miss when they're developing these. It's <hes> if you've ever I've seen the movie double wears Prada. There's this girl and she's really intelligent but she doesn't have any context you. She doesn't work in that industry and she's the assistant to this. This woman is very demanding and she's terrible at it <hes> but over the course Chris of six months she's probably the destin she's ever had. She knows what kind of coffee she wants. She knows the look that she gives remains a certain thing in all these things collectively allow her to service those requests in a better way because the context and I think you know <hes> certainly if you look at <hes> atoms approach. I think you know he certainly approach is it a different way in his is thinking kind of the next generation of how we got you know fulfill the promise of these assistance yeah. It's it's interesting to hear his his view of the assistant as <hes>. You know you're one contact with like the artificial intelligence world so he doesn't think there's going to be multiple assistance with there's going to be one that really he knows you that knows the context understands you and that that can then get you the services that you need and when I when I hear him talk in a way it's kind of it's a little bit <hes> worrying because if the if the virtual assistant listen is going to become so important then you see why this battle between you know Samsung and Amazon and Google and apple is is so critical because you know which one are we gonNa pick and I guess for Adam I'd have to think it must be frustrating for him because he you know is out there on the forefront so so long ago but just based on timing and different you know things that happened his his view of it didn't wasn't the one that got out there and got the most traction early on so it makes me. Wonder you know what's going to happen. I think is really interesting. If you think about that episode he talks about the original Siri App so Siri was first rolled out publicly although they'd had some implementations sort of more private enterprise is situations rolled out publicly as an APP in the Iowa's APP store and he he makes his comment where if you talk to Steve Wozniak one of the original founders of Apple <hes> you know even last year would say his favorite APP of all time was a seri- It was not the theory that was implemented but this area APP because it was more this this full concept of it would asthma talk about is knowing and doing right so you'll be able to to respond onto queries from users by <hes> either accessing knowledge and he sort of talked about this concept of there's knowing companies that aggregate all the data like Google and his voice system just needs to plug into those is knowing companies and then is pete you brought up this idea the do engine which was core to the the Siri roll out in Bixby's really focused on this whole doing things like how do you put all the things together to create these transactions and capabilities abilities that either pull information or or pulling services web services <hes> in order to fulfil need and I think that was I think that's really fascinating and I do recommend people listen to that. Anybody who's in the industry that that might be the most instructive of all of the of all the podcasts because we do actually walk through twenty five years you know he's hand dot keet Laos his Co.. Kander were known for not just creating Siri this sort of the first modern voice assistant whatever you might think of that today <hes> but then also the labs was become bixby's so they've got to under their belt. It's pretty impressive Dave yeah one of the things that he said in that interview that was really interesting was about the fact that <hes> you know he has created fifty different assistance and after Siri came out it was as if everyone assumed that was the only way that assistant could be and he was describing all of these different environments in which you might have a different assistant so he described the one where <hes> you know for like a surgeon <hes> the way that that assistant was built so I thought that was really interesting because again it goes back to this notion that we've it's like he brought to market Siri and that was what everybody assumed to be in just to hear him speak about how it doesn't necessarily need you too all fit the same mold <hes> it. It made me realize that you know there probably will be a whole lot of variety in terms of their ways in which people use their assistance and it won't necessarily just be everybody uses it for the weather and everybody uses it for this use case. I it just made me think a lot of different things around that yeah. I think that's a great point that you brought up. I mean he was putting a finer. I guess a finer point on the whole concept of what Syria was supposed to be what it became mm-hmm and one of the challenges they had with serious. They didn't roll out the full when when they rolled out in the iphone four S. it didn't roll out the full feature set of the original Syria because original Syria could not have rolled out in that many countries at the scale that apple was going to roll it out and so they didn't really have enough time to build out all the feature sets and then when they did roll it out they actually way underestimated the amount a usage that was going to happen and so again just like it was all back office stuff that they had to worry about in terms of the infrastructure and that really narrowed the the functional set and then by the time they could've got around expanding going back to the original idea the due engine dog and we're gone and they found it and I think the the other leadership that was left with that <hes> Tom Gruber was still there. He was a CO founder series well <hes> but I think the rest of the leadership was a minor diner expansion really making it work. Well was really their focus. I thought that was that was really really interesting. <hes> any other comments <hes> Pete Dave or amy on Adam. Yeah I mean really quickly I I would just say in one of his other thesis we every ten years or so we move to these different paradyne's and we seem to be moving away from the mobile APP download update. Your apps every seems like two weeks. I'd have to update my up <hes> to this assistant model that what it's doing things on your behalf and is a company or developer. You're you're really writing to the platform and and this assistant is kind of interceding in that that's how he explained it in. That's Kinda degree. That's the first time I heard somebody express arrested that way but yeah absolutely yeah so every ten years you haven't user interface change in our latest. One is voice as an interface but also the voice assistance is platforms in so that that is important and I'll say this one last thing in closing about that interview with Adam that stuck with me he worked with his mentor was a gentleman that had actually rolled out a lot of concepts. He was the inventor of the mouse and all these other things he'd rolled out some things and <hes> nine hundred sixty eight that that he thought would be adopted in five years <hes> twenty years later they started to come about he had a conversation he remembers sitting down in Nineteen Ninety nine when the web was taking off in looking at a video interview of his mentor who people are asking why wire hyperlinked multimedia taking off you know the way you thought it would interconnected on the Internet and the ten years later it in one thousand nine hundred eighty didn't know why Jesus seemed like such a good idea. Why aren't people doing it? Well just it took a while later in just talking about that and so twenty five years and I said so what have you learned in one of the things he said is I've learned patients the light of these things if the right idea will come through and I think when we're in the midst of it and and we see where this is going you know whether it's the intelligence amplifier or security or on device processing all those things we can see that that's the right thing to happen but it might take a while <hes> and sometimes as a technical barrier sometimes it's just adoption and people are busy with other things but eventually they get around to it so I that's a that's a great episode. There's a lot of really good ones. I'm I'm glad we talked about this. I don't know how many we talked about your prior fifteen or more. Maybe twenty twenty two but we we've gone way over time. You guys have been so kind about this <hes> Amy why don't you tell the voice about listeners how they can keep track of you and check out the work that you're doing yeah you can follow me on the Taliban's website <unk> dot com or on twitter at I am at talks to bots and you can also check us out on instagram at Tel lables excellent Pete Yeah pitas dot com twitter conversation <unk> curve dot Com is by agency where do some development and I'll be speaking of voice <hes> Voice Summit next month I think along with Damian insult and David Camp sure you can find me on twitter at oak tree underscore Dave's <hes> you can also check out my blog future ear dot co <hes> yeah looking forward to meeting Pete at a the voice summit seen Amy Umbrella Gin Seniors at Boy Summit all K- that is great <hes> thank you so much <hes> Dave Pete Namie and thank you voiced by listeners those of you who stuck around listened all the way to the end. We got some great content in there in the last fifteen minutes so you were richly rewarded for that you can follow me on twitter at Brechin Sela. Check out what we're doing at voiced by that day I and if you really want to know what's going on in the industry. If you really want to know what's important what the people were the insiders care about uh subscribed voice insider. You can just go to voice about that A._I.. Force last insider you'll find out more information there that is subscription newsletter in it's stuff that you just not gonna see elsewhere so check that out. Thank you once again David Pete and amy.

Alexa amy stapleton developer Dave Steve Kemp Dave pete Google Amazon Ron Krone Samsung Dave Yeah Trinity Audio T. R. Pete David Amir Hirsch Clinical Oh founder Dave Camp John Kelly bixby
Peter Durlach SVP of Healthcare Strategy at Nuance - Voicebot Podcast Ep 159

The Voicebot Podcast

00:00 sec | 4 months ago

Peter Durlach SVP of Healthcare Strategy at Nuance - Voicebot Podcast Ep 159

"This is episode one fifty nine of the voice podcast my guest. Today is Peter Lack the executive overseeing one billion in healthcare business at nuance. Welcome back voiced by nation Brechin. Selah here with another guest. It has a lot of in-depth voice. Industry experience to share nuance is a voice giant that has been around for more than twenty five years and Peter lax role in the industry predates even that timeline. One, quick shadow today for another five star review and apple podcasts, Pete Haas wrote the voice Bob. PODCAST is the gold standard in voice and conversational I there are a couple of reasons first. The quality of the guests is stellar. Get deep insights from CEO's in industry experts every week second well there. He just said something Nice about me as an interviewer, which Sperry of today, but I'm definitely thankful to have peace compliments. Thanks Pete for the five. Star, review. If you want to do something to. To support voice podcasts giving us a review is an easy step. Know each week for more than three years now we've brought you a guest that shared their experience, expertise, insights, and more just as Pete mentioned in his review. This podcast is abroad. Bring you the people that are moving the voice, my industries forward and shaping our collective future. Today's guests has a big role in that, but I suspect that if you don't know him, unless you've been in the industry for a long while Peter, durlacher. Durlacher started out audiology technology back in the nineteen eighties, and then quickly transitioned into speech technology in one, thousand, nine, hundred, eighty eight. Yes, before there was nuance, dragon systems are speech works. Peter was in early executive at articulate systems particular was acquired by the infamous learn outhouse, being one, thousand, nine, hundred, nine and Peter Spent a brief period before moving on to lead unveiled technologies a pioneer in the context center segment. Eventually, he wound up at nuance in the healthcare group twice with a brief interlude. Interlude at up EMC to nationally recognized hospital system out of Pittsburgh and western Pennsylvania for those review are familiar with healthcare, but today overseas nuances, one billion dollar healthcare business, which includes solutions used by over eighty five percent of radiologists and sixty five percent of physicians, incorporates speech recognition, dictation, full-on assistance into their daily practice activities in our conversation discuss peach, three decades in speech and voice technology, and we hone in on where voice is in. We'll be making a big difference in healthcare. Let's get started. Peter durlacher welcome to the voice about podcast. Thanks for joining me today. Thanks Brad very very happy on you. I really appreciate the time. Well, it's great like so really excited to talk to you. We've been talking about customer assistance for the last three months straight on this podcast, so you know, think twelve thirteen fourteen, maybe even fifteen hours maybe longer I don't know you know talking to people who have been deploying custom assistance on their own behalf. Some that have some tool sets to deploy for other people some who've got these new business models around it so all these different types of things. And make people think of nuances being there from the beginning and I know you actually predate nuance in some of the different companies that came together to form new odds but I thought we'd start out out so. You're oversee new healthcare I believe. It's over a billion dollar business. Why don't you give sort of short summary of? Has that context of what nuance healthcare does today? Yes, Oh, nuance healthcare is is obviously part of nuance over on. We're really made up of two rating divisions that really a pioneer conversation is for for B. Markets. We have really to operating groups. We have a digital intelligent engagement business that really looks at automating consumer engagement between a consumer and a large. Large brand so large telco financial services travel agents use us to create conversationally I whether it's on voice, text, email or chat, and we were very large business around that, and then the in the healthcare business I help. Ron is a billion dollar business about two thirds of nuance, the takes that same technology applies in the world of healthcare are really around the experience between the provider and the patient, so we build conversation late I that allows a both clinicians and patients. Patients to have much better experiences so on the provider side clinicians can basically talk to their electronic systems to document what they are doing for the patient and automatically turn that into tax structure data that makes it easier for the clinician to spend more time with the patient and also dries tremendous efficiencies around a revenue, performance and quality is driven from clinical documentation in healthcare on the patient side. We apply this digital engagement technology to make it easier for patients to interact. Interact with their health systems whether that's checking on an appointment checking on their clinical care. Maybe there's a follow up. Did they take their medications? Maybe there are some clinical finding that needs to be communicated. Follow up on, so really. We both attack, both sides of the healthcare spectrum that provider experience with the Electronic Systems EHR's that they use, and then the patient experience with the health systems are the physicians as they try to get high quality care from their providers. And when did nuance I start to point speech, technology and healthcare? Interesting you ask nuance. Really started as a company back around two thousand and around two thousand, two, thousand, five, the CEO at the time decided that it was important in addition to having core technology, a really focused on the contact center, which is the original legacy of of nuances kind of digital engagement business end. It was only on the voice channel. They were really not tax or or or email automation at the stage. That is today. And then they went around. Looking one other vertical markets are there where voice can really play in solving real problems and actually healthcare was the biggest market at the time for the technology, but nuance really hadn't made any play in that so late two thousand five. Two thousand six when nuance decided to get into the healthcare vertical actually when I joined nuance in early, two thousand six. Got It so. Fifteen sixteen years now you've been at it. Yes it's It's been interesting journey as I know. We're probably GONNA. Talk about. It even started the journey for applying conversational. Ai In healthcare started actually way before a nuanced got into it in two thousand six, but as you know, it's really rapidly accelerated over the last fifteen years an incredibly so in the last five seven years. I was looking it. You're you're linked in profile ahead realizes, but you started in the nineteen eighty s in speech technology. I'm very interested in. What that was like back then what the technology was I, guess had somebody ideology technology originally, but after that you moved into. Some speech recognition for healthcare in particular FAA. What's yes-or-no so? The journey when I got out a college in eighty six, one of my first job was I went to work for medical device manufacturer spun out of Mit that made what we're called tackled devices, which were profoundly deaf. It converted audio signals to tackle signals and creating vibrations on your arm, and you learn a here effectively through these vibration patterns as opposed to the auditory signal, so that's the first kind of inroad I got into the acoustics. If you will because I was not my specifically. My background and then in a around eighty eight, I got introduced to a startup called articulate systems. which was a pioneer in applying speech technology to remember back at the time, the Macintosh at just come out a couple years after the IBM PC at the beginning of the PC. Revolution if you will and one of the interesting. Issues with the Macintosh was ahead. This great graphical user interface at made it super easy for novices to to navigate the. Versus the DOS the famous dos prompt, the challenge was for sophisticated users, pointing and clicking your way through all these menus was really against the time because you knew where everything was you just? You remember that drop down to like Jesus. Can I just get to the function I? WanNa. Get to so. At the time articulate took this. Technology speech technology was out of the. Ib sort of out of the research labs at the time from Carnegie Mellon and MIT and others and work with an early partner called Dragon Systems which is a company, one of the pioneers in the space also Newton and we built versus voice I voice interface for graphical user interface with the product voice navigator, which was accommodation, hardware and software product. which was an amazing product? It allows you control the entire Macintosh by voice we want all sorts of award sold a couple hundred thousand units, but a few years. We realize that. Except for a small set of super power, users and people with disabilities. It didn't solve enough of a problem. Because at that time there was no dictation. You couldn't dictate power wrap. It was all command and control only discreet like find this do that. That was all the technology could do. At the time, and so we went around looking at. Aren't we probably had a couple? Hundred thousand users worldwide on that product, and we realized that in order to really make a business out of it. We had to go up the stack and solve real meaningful business problem versus being avalon to graphical user interface. 'cause you just couldn't make enough money at that time with the technology in where it was, and we realize about fifty percents of our users were physicians. And I was not a healthcare person. At the time, this was back in the late eighties and started talking a lot of these users that I found out. They were using it because even earl at that those early days. Using some of the clinical systems early HR's early radiology systems were very cumbersome. And they also wanted to try to use it to dictate, and so after about six or nine months of due-diligence, I realized that we we needed to pivot the healthcare 'cause. That's where the the pain point was. We started talking to clients and a within that process. We realize that even in healthcare where we should start. Was In radiology. Because radiologists dictated a lot, it was incredibly expensive, and it took four seven twelve days to get the transcribe report back. And back in the early nineties, we struck I actually struck up a partnership with. Medical Center the Radiology Group. Build what later became a product called power scribe, which is still by far the leading speech product in Radiology today that actually nuance acquired many many years later, and that's how we got into healthcare back in the late eighties early nineties. got. It so this is great, so if we talk about We talk about being in healthcare. nuance has been for fifteen years, but in fact you're working in this thirty years ago. Fifteen years thirty years ago and in Nuance, eventually bought the company did the the the product that you're working on back in the eighties yet? Basically, what happened was nuance. It's very interesting as you know. It's a very small university in the conversation away I wasn't calling conversational back then by the way it was just called the speech recognition business, but what happened was there were a couple of players in. In the space back then there were US articulate systems, which again was focused at very as horizontal play focus on healthcare. Then they were a couple of of horizontal plays. A company called Dragon Systems out noon mass a company called Kurzweil was also Massachusetts with Ray Kurzweil, and then IBM and Philips were doing some of their own work, and what happened was a company came along. LEARN OUTHOUSE BE! Back in the late nineties, and they started to buy up a lot of these. They didn't by IBM obviously Phillips. They bought Kurzweil. They bought eventually articulate and they bought Dragon. Connected that. I was acquired as part of the acquisition, but after about six months I I didn't really show a few of us. WHO found that? L H. H had some behaviors with the management team that we're on a famous in the industry, so we all quit. This was back in ninety nine. For the indictments way way before the indictments, there was immoral man. Yeah, it was. It wasn't obvious exactly what was going on, but there were some strange vibes it to myself and a few founders articulate did not feel comfortable, so we were only there for about six months. We left it in the indictment. Stuff was probably about a year later. Retake little bit, but interesting out of the bankruptcy. And I went off the found another company called unveil, which was a call Center Automation Company with the folks that I found a articulate with. And Out of the Bankruptcy Process Ellen Age Paul Ritchie, who took over scan soft, which was a spin off of Xerox. He turned. What was an image processing SPINOFF OF XEROX? It's old technology club vision ear. And end up buying two companies scan soft and the original nuance which was SRI. SPINOFF did call center automation. Rename the company nuance after the acquisition of the Stanford spinoff caught nuance, and then in two thousand, late, two, thousand five. A two thousand five neighbors a little bit earlier than that N- nuance bought a bunch of assets out of the Ellen H bankruptcy. which included a bunch of the Legacy Dragon Solutions Etcetera Etcetera and then in in two thousand five, when nuance bought dictaphone right before I joined they. They acquired the original assets of our system, which included power scrag show effectively. What happened was. All of the key assets of the space ended up ending up in nuance between two thousand and two, thousand and five and two thousand and five which. Provided the based framework for all the growth that's occurred some so all many of these early pioneers. All the technology ended up in nuance over that five year period. Right and so you know during that time. You're talking about speech recognition a and in particular in healthcare, and I think it's still a big deal today. You're talking about transcription, right? It's it's. It's it does transcription speech, but it's a lot more than that. Yes, initially. What happened was. We live power spray power sky was the first enterprise class conversational Abe I product in healthcare where you've got large scale adoption of the product for all the workflow within a given a given setting. So at the time, yes, people use our scribe instead of dictating into a recorder, and then basically having these tapes. Transcribed by a human a week to two weeks later, so it was all automated a basically from this this a synchronous recording the people typing. Speech Technology, and from there kind of took off to not just for radiology, but today all of healthcare, no matter what specialty you can use it for both real time, dictation or transcription, you can also use it for virtual agents might might might much like you do with Alexa or Google? You can now say Hey, dragon, or a epic, which is one of the largest EHR's and control system by voice, and we also have a now have integrated much more on intelligence into the product, so you're not converting what? What the user says from speech detector commanding the application. You're actually driving real time feedback while they're documenting so for example, a documenting an encounter about feature lack of just showed up. If their clinical information available in the record about meet that I haven't talked about the system does more than just process what you actually say. It also provides feedback to the clinician about things that they might need to consider because we know much more about the clinical situation than just converting exactly what they they say in the text. Right in that case, what you're thinking about is the you have access to the medical record itself information about Peter Lacking, in this example and the the physician might not might not have reviewed. All of the things in the might not have them top of mind, so if a certain topic comes up, you can prompt them. Yes. Exactly you can basically say example I won't get to clinical here, but if there's a patient in the hospital and they have a bunch of lab results in prior, clinical information. That says the patient likely has actor, x, plus Y, but the clinician only says it in certain less com complete way, the system will actually give them feedback saying hey, they're other clinical indicators. You Might WanNa. Consider Act why that's important. Not only for the care of the patient. 'cause you want to document what is appropriate, but in health care. There's two other things that are really important that if you're not in the industry, not is not obvious. One is for the for the provider, the hospital in the in the in the physicians. They don't actually get paid for what they do. They get paid for what they document. They do so. If you don't properly document, what's wrong with the patient? What you're GonNa do that affects the entire revenue side of. The equation and although no one wants to pay more money to the healthcare provider, which is obviously a good goal. The challenge is most of these providers operated to percent operating margin. This was pre cova by the way so for them to make sure they get paid from the pair. The right amount of money stay in business and deliver cares important thing the. The second important pieces, the quality of care really gets affected. If you don't accurately describe what's wrong with the patient in the record than all the downstream things about what should happen next, or if there's another clinician looks at the record. They don't have all the information. Do the right thing to take care of you, so the documentation isn't just this kind of. Thing. You have to do because you know. It's part of the process is actually a large part of the lifeblood of how care is delivered and paid for in the US. Right so when we think about where you were when you started out, it was largely just transcription, and they needed physical files. And then over time at what point did that turn into being? They were looking more in terms of creating digital files where it was actually going not just into a document, but going directly into any HR for example. Yeah, so I mean when we did power scribe back in the mid nineties. You know that that was purely digital. Right? I mean you know if you've seen radiologist dictate, but they're unbelievably fast. You could barely understand the fast ones yourself as a human I remember when I first saw that I was I was like Howard Hecker. We automate this I can barely understand what the radiologist saying was quite a cathartic moment, but the technology is really a ball, so yeah, it's in the mid nineties late to midnight actually early to mid nineties, it started all come digit. Another EHR automation happened a little bit later because. The complexity of handling like a primary care doctor and what they're talking about is more complex than radiologists or pathologists where they're what they talk about is more structured more repeatable than like your general family practice. It can talk to you about anything effectively. So what happened was in the late. I would say around mid. Two thousand five. Two, thousand nine, the general purpose speech recognition for the Product Dragon, which is today. It's a new product about sixty percent of physicians in the US uses product. It's all on. The cloud started become good enough. That, most physicians could use that without any transcriptions in the loop that has exploded since then so now all this technology's all cloud hosted no voice training. You could dictate in like twenty eight languages from basically any device including a thin client. That's not even really a PC or tricks or something, and the performance is just it's. It's not perfect, but it's pretty mind blowing. Yeah, so now in that process like for the Dragon Medical. It. Are there humans who are doing review? No The well, the physician always has a sign off on the chart because they are legally responsible, they are words that you've, but there's no human in the middle. It's they just rattle office fast as they can go in Texas shows up in or CERNA attacker. Athene or any of the HR. They just do a quick scan and they signed the note. Okay in that case, then the reviewer is the physician, yes, and that saying power scrawny most of the speech technology in healthcare today, The vast majority, not everybody done where the physician there's no other human other than the physician signed the no, there are some cases where some of our newest technology which I can talk about in a minute, which is kind of the next generation. That's still for a while. I may have a human in the loop for some period of time. Yeah so I'm interested in this to where we think about the transitions right, so we had this dictation to like documents whether they be physical or digital. We've got this transition to. It's essentially a voice input or voice Maybe not voice navigation as much as voice input It's a dictation of another form into a system as opposed to a document. And then I think we're headed with. That was really take idea of assistance. We mentioned before like it can. It can actually talk to them, so tell me about that. Because you've been in this industry a long time. They're always. There was always this idea of assistance that can help you out. I think you talked about early on right, but we have this idea like there's voice dictation. There's voice navigation I can get to somewhere, but I could normally do it with key strokes, and then there's an assistant which is a level up from that, so tell me about like your thoughts on that evolution of what the big break points event. Yeah, there's really I'd say three elements of that there's. Sort of this this assistant? That kind of the sort of is a hybrid of navigation, which again is what most people think when they take them insistent like Alexa Google or something where you can basically have this natural language dialogue if you will quote unquote with with a system to get something done in obviously in the consumer space, you, your audience are very familiar with. Of that in healthcare. You know we do that through. Hey, dragging her. Hey, epic her. Hey, whatever the name of the system is very natural via live where they can search the system, they do prescription by voice by say like Hey Hey Dragon five hundred milligrams. Twice a day by mouth. And refill every three months had that automatically parse triggering order? The HR and the physician ever touch the keyboard right so that's a classic. Virtual Systems. It's kind of a next generation of command. Example that you gave the second bucket is assistance where the systems giving intelligence back. My example of before. Telling, advising the clinician that there's some information on the record. That suggests something else than what they're documenting about. The patient that we also think of as assistant on. Taking command speed. You're turning into tech. It's actually giving some advice backer decision support back to the clinician in real time that sort of a category that we pioneered about three years ago. That's getting a lot of traction. Right now and then the third element which is kind of the holy. Grail is this thing we clam being clinical intelligence. We are instead of the user having to explicitly say what they want. You made the system into a listening system it from an ambient perspective, and it's just listening to the Congress in this in our world, the in this example, the conversation between a clinician or in the patients and automatically creating the notes and suggesting information without the physician, having to explicitly say everything now a clinician can still say things like A. What was when was peered? Relax lasts, you know. X. What was his last lab results, or when was his last physical etc, but they don't have to dictate anything. And, they don't have to do a lot of the manual task. It's an artifact of the system listening, so we launched a solution two months ago. The Dragon Ambien experience, which is the solution. Of that ambient listen to the conversation that can be a physical visit where you're in the room at the clinician. Through explosion of telehealth where the visit virtual. That's really the next generation. Tech in this space is completely transformative technology. That is basically. Create. Using neural nets turn colloquial conversations between clinicians and patients in language. The patient understands into highly summarize and structured note as if the clinician had dictated it after they left the exam room in in the old way of doing things with track. Right, so it's it's. It's like this idea in the past where you had the traditional dragon medical was a voice data input mechanism. And now you don't actually have to be explicit about what you're trying to input or something like that. You're going to pick up on clues in the conversation and some or maybe all of that conversation will be added to the medical record, and some of it will be passed into specific fields which are. which are relevant to the HR or the electric medical record in this case exactly right? Analogy here in our engagement business, too, which is something they were exploring. We haven't launched this yet is if you can imagine you know when you call up a large brand whether it's Delta. American Express Fedex all clients is by the way many of these clients record the conversation, so they can coach their agents right. You always hear that recording notice well. We're building technology ambient for that environment to. We can listen. You're already recording the conversation. The brands are not us, but the company's. What happens if you got automatic parse information in real time from that to do things like. Oh, this person's really talking about extradite automatically suggested that also Asian things that could help out in this environment with the next best thing to do so this idea. Of Ambient as passive intelligence as opposed to a human, all trigger is really what we see is the next generation conversational I? It's not just adding another skill to a virtual assistant. It's going from active to passive and then blending those two, and we see wide applicable of that, but the market that we've been most advanced on his our healthcare. And when you say active and passive, you're using that from the perspective of. In this case, the agent which could be the clinician could be anybody who's talking in. They would have to ask the system for something to get it. In this case you're saying it's passive in that because it here's a trigger or a concept. that it it recognizes that there could be some useful information or a useful action. It'll prompt the user here so that they don't have to ask for it exactly, and this is all enable from the amazing advances in deep learning neural nets, because basically you now can take these vast amounts of data sources whether that's call recordings in the context of space, or these conversations that that that we collect on the healthcare side, and you can use them to build very very sophisticated. Neural Nets at basically recognize these patterns and say oh I've seen this before, and this is what's likely to be relevant whether that's the create a note for the clinician because they just saw someone who's getting a knee replacement of this the seventeenth millionth time it's heard about an example conversation about a new replacement to contact in engagement. Call where this is. The five hundred thousand caller to some large bank wanted to check on mortgage rates in loan. And credit requirements days are just These are all example driven data model where with the with the new world where data is the software, you can start to build these much more sophisticated ambient solutions that. Do exactly what you said. They don't always wait for the humor to trigger them. So you mentioned deep. Definitely an area of interest to me and a lot of the listeners this podcast I'm interested in the architecture, so are using deep learning primarily at the level to to to Parse the concepts and understand where then you might be able to interject in provide added value or are using an nd deep. Our work, we're using all sorts. Whatever technique gives us the best results? But to answer the simplest level we do all of that, so for example we we in some cases we used. We use end to end models from like. What's been recorded a speech, the final results we also have we inter well. We've multiple models, so you might have models that are doing the speech recognition. You'll have differ models. That are doing the annal you and in the case of like. What you've got people doing actions whether that's agents like traversing. Faq or some internal sister imagine a physician traipsing through any, HR can also use it to track acts on other systems, and you then just tested measuring you see what combinations of the models produce the highest performance and we've seen. Different examples of where that different model combinations produce the best results because at the end of the day. Is You know it's all about the performance results on? On. Accuracy and air rage precision, recall et Cetera et cetera dependent metrics. You're using for specific case, but we've seen massive improvements again. I don't we couldn't do what we're doing today. Without the technology me just dragging alone, not even the ambient piece in a we'd use neural nets on that probably. I think for almost decade maybe not quite. Because, we you know we have new allies. We have three thousand terabytes of data and all our products from the cloud, so we have this massive learning system so every time for example in healthcare, anyone of the six hundred thousand positions will use dragon say something in any language or make any change that all gets aggregated. In these large nets that are used to approve. The the quality of the system in one of the reasons. We can do that on his. Which is critical as you mentioned, it's not just a model to use it. Sedate is out of a trusted partner of our clients. We only use the data. In a very secure way to make the system better we don't use. We don't sell the data. We don't monetize that. We don't rate derivative work on. A key focus for us because without this bass data source, we couldn't get the performance of the advanced models whether it's end to end or not, so that's kind of how we approach the client's world. Rights do you have a baseline model so anytime? You had a new physician to the system or let's say a new. Physician, practice or hospital? They would automatically benefit from your baseline model for their practice area. Assume yes, it every. Yes, we have a set of base models and both language and acoustic the system automatically. There's no training. Just walk up and start talking. Incredibly High and So, yeah, language and Acoustic I get, but also. Domain like sure. Yes, hundred everything all the models that affect. What language is the person speaking what accent they have? What topic are they talking about? That's automatically determined. From just listening to the user and automatically dynamically. Adjusted in the cloud and then adapts over time automatically in. Cascading models from general to specific. On and that's how we drive. The superior performance is squeeze out every little thing you can make. The system is close to percent accurate as you can. Right yeah, it makes sense so the general model that you have, and then you've got domain specific models for different practice areas presume, but then do you also allow for customization at the customer level or at the physician level? Yes, absolutely, they can add whatever they want. How do they do that in terms in in your system to? So for example So. You've got a lot of neurologist. Maybe who are using it drag him so new neurologist comes on. Maybe has a maybe has an accent that you don't have a big You don't have. A lot of data on, but they can then train it. So how do they train it? He will. The Acoustics are amazingly good. We have so much on acoustics. It's very I mean. There are some languages that we don't recognize. That's true generally will build. A base model before we deploy into that market, but for the the mark. Were you know we're in like I? said the twenty over twenty six countries. Right now so they just come up and starts working. The accuracy is really stunning and then. Any of the adjustments on the Acoustics happen automatically, but then they can also add their own words or macaroni. Those get stored in a user's specific profile, so we know this neurologist tend to say things in this way when they say ex wanted to appear y that sort of stored in income that outer layer the I know. If you start at the core research, the general models are in the middle and then as you. Wave out from the from the center out to the out rings that Outer Reagan's The user specific things that we. That we capture that are only relevant for that user at the most yeah yeah. Yeah, so like, for example I think a lot of people think that a healthcare is very common. It's taught very similarly around the world. People use the same nomenclature, but in reality it's. True I think like. A specific farmer pharmaceutical though Dron. It's pronounced Madrid. Mido Drain micro drain a couple other things I've heard his well. It's like five or six different pronunciations of the same now fairly common drug and so this would be the type of thing where they would basically train the system or some of those cases. You might know the variance, right what without mentioning other names. This is you know. Just put a little bit of a plug. This is one reason why. It's really hard for the consumer players in conversationally either get into healthcare in a serious way because. The language you know some point. They may be at do that, but right now. They performed Celta between our stuff and their stuff is Masol. We have this massive data, and so we've seen benchmarks that are the horizontal or three to five hundred percent, less accurate on the medical terminology and. For exactly the reason you said there's just so many variations are so many orchards. They're changing, and also it's one thing to say hey. Alexa set a timer or play you know Bruce springsteen. It's another thing to dictate a half page clinically rich content. From a user that's that's paid. Five hundred seven hundred thousand dollars. A year has absolutely no patients, the correct anything, so the performance requirements are just so much higher at if you're wrong instead of playing the wrong song or having the wrong time or you know you might put in record. That patient has something that they don't. which then can have really bad consequences downstream? Error is not caught, so it's just one of the reasons. We nuance has pick these verticals that. You know we don't want to compete at the horizontal. The Nila level we're focused on these highly highly demanding enterprise business models where the performance is required, that are just light years ahead of the consumer stuff, and you have to be deeply integrated to the workflow that the users use it all goes back to what you're talking about. which is it all starts with how good in the models on doing the core promise that you made to your buyer, which is, can you recognize highly accurately what the person's trying to do? Right right okay, so we've got this idea of. The speech recognition the Hugh. We've got a baseline models. You can customize them. Tell me about an example. I think one that might come to mind when you Vanderbilt Tree No. Different hospital systems where they've created a virtual assistant you talk about the feature set that they've employed. They're great example so again. It's a great group longtime client of ours. They are one of the. Has had a history of innovation. They used to build their own hr in the old days. And there's a a group led by pediatric endocrinologist down there. And they have the same problem. Every healthcare institution has which is that over the last fifteen years. The amount of administrative requirements placed on clinicians has exploded in these are things like related how they get paid in measuring so. Most physician most physicians today you may be shot. Maybe you this up. All your audience, spend twice as much time on paperwork and administrative as seeing patients, which is crazy right? I mean who will to medical school for ten years racks up all this data to try to take care of people. And then they spent twice as much time kind of clicking melting their way around the requirements out instead of taking care of people, so there's this. This massive physician burnout problem with the US were about Fifty Percent Law Collisions burnt out fifteen times higher than any other white collar professional. It's estimated that about four to six billion is wasted the maybe higher than that per year on Bernardo or seven thousand per physician. So you have all these physicians getting really frustrated. They don't have time with their family at night. They're catching up on their documentation there quitting. It's a real crisis so. Everyone's trying to do is how do I reduce? This administrative burden and much of that burden comes from clinical documentation. Placing orders and filling out these forms, quality measures, and that's kind of where we sit so virtual assistance like Hey, dragon, or in the case. Hey, VIVA VIVA is the name of the Vanderbilt custody agent is to try to make it easier for clinicians to do a bunch of that administrative stuff. Much easier so if you see thirty patients a day and you can, you can reduce ten or twenty clicks for patient. You can really start to cut down the amount of time. Dragging alone cuts about an hour two hours of unnecessary administrative time away from physician. That's that's using the keyboard or the mouse, and if you layer on top of that virtual system that allows you to navigate or trigger orders whatever by voice, it really makes a big difference. What Obama built is we have a conversational AI platform called mix. It is something that you would think of as competing with Amazon Lecture Google Dialogue Flow. It is a development environment that offer our clients can use to build virtual agents whether that's Fedex. Are Delta building virtual agent for flying? You're sending a package four. It's a it's a virtual agent to help. Make it easier to navigate any HR. So would vanderbilt does with our support. They built a custom virtual agent that integrates with their band built version of their epoch. HR To make it easier for clinicians to navigate lump information about the patient, a using conversational from. got. I WANNA go back to a data. Point that you gave you said it save about an hour. Per Day and just the documentation is that or excuse me? Did you mean an hour per day or hour per week? You didn't know our two hours a day a day. What happened right now? You know they they work. They try to do the documentation between visits. Some do it. After and then usually most of this has been tooth rows at night. Without going away from their families, it's called. It's called in the Industry Pajama Time. Documentation that literally so we can basically reduce all that after work time we actually in the they can track by physician. How much time is spent on every task and we have reams of data out. This is why you know. Our Dragon Cloud product went from zero users over four, hundred, thousand, four and a half year, which is again forty percent of the US population position because it's just a massive timesaver for them. Right it sounds like it's between twelve and twenty percent. Yes I when to our debate yet depend if you talk about workday or the whole hour the full day or the workday yet depending on the Mat and our to hour of time that they normally spend. On those tasks have been reduced, and we've also seen for physicians that want to spend the extra time they can see fifteen to twenty percent more patients. Yup absolutely okay, so you get the higher productivity, but you're talking about nuance mix. I'm pretty interested in this. Because that's that's relatively new offering. For for Duets, and either one of my questions was in our people using that in healthcare as well so it's mixed. It's is across use cases. It is a started out of our engagement business. Because that's we for years. We've been building these amazingly complex enterprise class. Omni, channel engagement platforms again for very big brands as I mentioned, and it became clear a few years ago that we needed a better environment in love. Our clients want diy tools to do it themselves, and so basically the team took all this knowledge and built it into this modern container based cloud infrastructure. We while Azure that that tool can now be used by us by partners. Our nation started in the context centers facing now. We're aggressively moving into the ELK. Got It so nuance mix that is, is that like one of the manifestations of your azure announcement you made last fall well. It runs on Azure. The Microsoft didn't contribute any you know. Technology per se to that, but it sits it it. Is A azure hosted service that we say that your cloud detrick cloud delivery cloud delivery service of Microsoft partnership was really around. They are working with us to jointly develop our DAX ambient product I mentioned earlier and we we. Our, Products in in our core products in Azure, but the but the the special relationship with Microsoft on the joint development side was around the product, but mix. It's Kuban Eddie's based container based framework runs on Azure. I got so for for mix. It seems to like run up against one of the one of the sort of a long standing critiques of Nuance. You guys have been around a long time, so you've had criticism for longer than most people. Is that people were totally reliant on nuance they'd have. They'd have to acquire the software. They'd have to hire professional services to build it. They have to have contracts to support it. Like how much autonomy do you have? Would you use mix? Too great question. You know it's interesting because it's much like the early healthcare days the reason that we did that as large bell answer question is because building a simple demo conversational I is. Building uprise class. One is really really are historic, so we put ps, not because we won the clients to be dependent on us. We put it because we needed to make sure the system. While not the given the outcomes, our clients were willing to pay for him, but over time more and more people want to do their own stuff, so we felt the time was red rose right, so would mix. Mix We don't have to be involved at all now. We have certain skills that many of our clients don't have or certain parts of skills, but technically they're. They don't have to rely. We have certain day. They do a hundred percent on their own. They don't use US at all. We have some that uses a lot in what they did. In Mix, which is really cool a couple things one is. They built sort of an agile scrum environment built into Mexico you can assign arts of project out to any team mother. It's nuanced. Employees are a client or a third party, so the customer can decide to form components to anybody. They want all working on a common project. The second thing with mixes a full end end life cycle, so the business analysts doing the armaments the. The You X. Ziona the speech scientists, the testing all occur in one environment, which is really cool, and the third piece is Omni Channel, so you can build on a single a example in healthcare so healthcare what clients want to do on the provider side is they want to reinvent what they call the digital front door, the connection between the patient and the health system. What they want is a system that you can go to their website. You might get an SMS. You might see a custom portal of the health system on your phone, or maybe you're trying maintain the drug regimented here into home, and all the person has as Alexis Martino at dot with Alexa running true mix. You have right wants deploy anywhere, so you can go one project. With all the logic all the dialogue that can handle all these different end points so for example a person could be at home. It could ask Alexa. Hey when's my next colonoscopy through? Mix Alexa will respond. Hey, it's next Wednesday at three o'clock. You could say hey, what am I prep? Instructions could give it to you three days later. You could be on your mobile phone with your health systems portal patient portal on your mobile phone, and you can type in our asking questions say I need to reschedule that ill. Ill know who you are. Continue the conversation. You could be back at your home three days later. Go onto the website. Look at prepped assumption, so it's sort of a a Omni Channel Way to control a seamless experience across these end points even with ones that are closed like Alexa Google, but through mix you have one way to manage that whole dialogue flow, and that if the client has, the skills can be completely, one are present on the client with nothing from new other than a licensed to the software. Got It so essentially what you do, you have a common l. you in the background you have a common library, responses or content you can deliver to people, and then it could be your as are in the front end. It could be a third party providing you intense that you can response exactly exactly. Okay, yeah, that makes sense like that's very much as you said. It's very similar to the way. Dialogue close set up or some of the other alternatives have any of your clients? Deployed production systems where there is a lecturer Google assistant on the front. On the enterprise side is I can do a follow up there with the folks. Run that division amount as far up to speed, but we have clients. We just launched us into healthcare like two months ago. We have a bunch. I can't out on forced to give names right now. We have a bunch of clients that are deploying systems that have one element that is. That is connected to thirty speakers in the home. Yes. Okay, yeah I think I think it would be. Just because a hip the you ass. I assume you're talking to you. Ask for your initial clients here Google. System would be difficult certainly Alexa. Amazon has a as a program for Alexa the biggest. Yeah, he's doing a hippo. Things appointed the other piece, though is that the the what these clients want is they? Don't WANNA use one environment for their clinicians virtual, which is all loss and then a different one for the patients to. It's better for them to use one environment, secondly, the core performance of our our you in our speech is just better than the other guys so they they do benchmarking all the time, and we generally come out on top on that because. In healthcare specifically, because we have optimize whether you're talking over voice, or you're texting something in languages, still healthcare centric, so it's it's the KPI's a care about completion rates, service rates consumer experience are just higher on the closed end points because you're right google an Amazon. Google have closed environment so you have to use there are so there is some degradation. Of the quality of the tax that we get out of that, depending on what the search string is, a nuance is not going to compete in the home in build smart speakers, Agassi, that would be crazy for us, and there are a lot of use cases that you want to deploy. In the home, and so if the patients in the home if they're using their patient portal, the web browser SMS, or whatever that's all nuance, one hundred percent tech, but if they use a smart speaker, you're right. They'll go through the Amazon Google voice piece that will get converted to tax all the while the dialogue management is handled by mix. What regardless of where the front speeches? WHO's speeches? God it so now did vanderbilt vanderbilt build using nuance mix they they use A. Version of that that we call a dragon medical virtual system, which is built on top of mix, but in the provider side we build a a whole set of intense that were prebuilt, and we integrated to dragon medical one so for Vanderbilt it's basically a a higher level than mic so that when they have this functionality, not only they get all the prebuilt tans, but you can easily switch back between dragon, medical and the For example you could say hey, viva, show me, Xyz how screen up then just start dictating at it, all just works so effectively a a mixed progeny that they're using in this case, okay and the banner bell. You're talking about a A clinician focused. Patient! And the patient piece that's the one you say is is new and people are just deploying. That's one hundred percent in mix native. Yes, okay, yeah, so you. You mentioned so this is big. Right in like one of the things seeing as obviously. A Alarm their large spikes and inbound patient conversations around Cova. I assume you saw that There are a lot of people who are trying to get through. They couldn't reach people. And trying to reach by. Some people had assistance where they be chat. BOTs voice assistance to flex and give more capacity in one way or the other. When you and you mentioned telehealth earlier as well. I'm very entrance what you've seen from your clients all your stuff in the cloud. You see what people are doing. What type of changes have you seen pre COVID middle cove. It now covid whatever we are I. Don't know if we're in the middle of it I guess we're in the middle of it, but word were were past the first phase. Hopefully. Yeah, some states. Maybe but yes. Oh, yes, Oh, what Cova did I mean? Obviously, it caused a lot disruptive just in healthcare in the first few months, people shutdown reelected. That has aggressively starting to come back in most states again again there some states that wobbling a little bit as A sunlight Florida parts of Texas etc, but the things that we saw were very clear and was universal number. One telehealth exploded obvious reasons, and for us the biggest driver for the occurred as our DAX are ambient product. We launched version for Telehealth, so all of those telehealth visits you can effectively DAX listen to the call or listen to the Tel virtual visit and create the note for the physician just like they would have done and physical and so that. Is the NBA assists? Yes, exactly so. That was a huge like flip a switch. Thing the second. That's that's. That's new with Kobe deaths. Do telehealth. The telehealth volume pre code was very small and only just went through the roof. Right, so we accelerate how how much bigger is telehealth now that it was oh months five months ago. Remember the thousands percents. I mean it's. It's it's insane. I can I can get you the number to follow up with. Three orders. More than that more than that. It's been I made people gone from like three thousand visits a year to like five thousand a day. I mean it's crazy. Now, some about selling down a little bit because physical stomach that most people think it's GonNa go down from the peak, but will never be in the low single digits was before, and that's mainly just because of how telehealth is is regulated paid for not because it's not usable. They were some structural issues that that in limited telehealth before, so we'll see what happens with the regulatory things post. Wasn't that one of the things for telehealth? Is that like? If a patient shows up at a hospital? It's a lot easier to have all sorts of documentation that something happened. When it comes to Telehealth, you don't have as many steps in the process where you're recording information. Well yes, and no. I mean the the very somewhere in the sense that in the office. You know you show up. They give you the clipboard. You have to repeat all the information again. It go see the doctor. Intel is based the same issues you have to do a pre visit. Check in, and then you see the doctor, so it's not identical. I mean you're obviously not doing a physical exam visit. You know bending your knees other certain visits. You can't do by for the standard kind of visit is actually very similar. It's just virtual. In terms of the steps now, most organizations weren't ready at didn't have these automated check in. So this is part of the problem. As as you mentioned before which was the second trend is all the contact centers would call patient access centers. Got Completely overwhelmed I mean the vine just went through the roof. Wait Times. People are frustrated, so we deployed lots of intelligent engagement voice spot, Ivr speech chat bots to as others did to try to alleviate the requirement for humans to be on the phone all the time that people just crushed so that happened and the other thing that we've seen which is unique to our side the houses. Because most hospitals operating at one to three percent operating margin when they lost fifty percents of their revenue, everyone went in the rent massively, so there's been a huge focus on revenue recovery because if they don't get the revenues up. They're going to go out of business I. Mean The feds GonNa to let Muslim God a business, but they're in a lot of pain, so all those solutions I mentioned earlier. That provide ai to help them properly document the encounter I mentioned earlier that part of the reason they need to do that is to get paid appropriately, so we have a set of solutions that use conversational and other parts of ai to help improperly document that than generate the appropriate reimbursement for a decent size health system. You're talking forty two hundred million dollars in incremental revenue a year with no change to how many patients they see or not them doing procedures that. That they shouldn't just because without the tech, most clinicians under document what actually happened in what's wrong with the patient and all the payment models on healthcare are tied to what you document, so they all leave all this money on the table and that money end. The patient's still pays. The problem is the money all ends up while not the problem planning your perspective of the provider's perspective. The money ends up with the insurer because the patients already paid the premium. The insurance, keeping it not reimbursing the provider, so it's sort of a zero sum game so depending on which side of the equation you're on, you think it's a good thing or not a good thing, but from the provider's point of view if they get underpaid for the appropriate was delivered margin problem is a real problem. Okay Yeah Fair enough all right, so we talked about telehealth is exploded in is never going to go back to the baseline. It's the the efficiency associated with it I think you're gonNA. See patient demand going up and we. We already saw insurers. It blew cost perceived was starting to push people into more telehealth for certain types of visits in those types of things so i. think that makes sense that's going to grow in Cova a big expansion. You mentioned the call center. What's going to stick in there? The the patient access center. What's GonNa, stick there. You think that's going to be long term. Are we going to see a lot more? Automated Assistant type of service for customer for a certain percentage of the interactions or we're. GonNa go back to a lot more humans. Where do you see that general healthcare specifically talking generally? Healthcare specific no question. I mean we. We have seen in explosion at. It's not going away before Ovid. Re reinvigorating what they call a digital front door game, which is kind of how patients connect with their health system or doctor was a top three priority for every health system. It's just jump alike. One be besides revenue. Recovery and so we literally in the last month I probably going on thirty. Mapping out. That may twenty exact calls where clients are asking us. You guys do this great technology outside of healthcare you do it for what can you do for patients? We show the mix. We show them over doing I literally. This is the HR vendors and provider I. Have Not had a conversation in the last thirty days where it hasn't been tremendous interest, and where this is not a priority that the biggest challenge guys have is how fast can they stand something up? It connects into their infrastructure and one of the big issues as you probably know, is just deploying a virtual agent on the chat side doesn't solve off because you really have to connect with a patient accent, so you have to have systems that connect between the voice channel and the channel be able to go back and forth, and that's kind of what we focus on because you can. You don't want route everything to the person nor. Nor you want to assume the the Chat Bot if you will can handle everything because nobody's jackpot can handle one hundred percent of the stuff you have to escalate for various reasons, people so you need these systems that integrate from the voice, traditional contacts in the world to this kind of new virtual Chat Bot world, so received massive acceleration. That I do not think that is going away and it's tightly connected to the telehealth. Explosion too, so I think that is definitely here to stay. Yeah, I presume like within nuance. You assumed that people would move towards more automation. The the patient access to digital front door. You expected it to go over. Maybe a more deliberate period than it is now we're going to see a big a big jump up. Totally yes, I mean. I think Cole would accelerate the digital transformation by by five years literally A-. Expense. So when you think about it got the digital front door. You've got the telehealth. What else do you think changes with? Voice Interaction Assistance in the hospital or physician practice so one of the cool things we've seen as an extension of our ambient our Jack Bronx is where opening that up to an ecosystem, so that would explain it in the physical room. We have a device that we custom-built goes on the wall. For sort of the. Hardware device. Yeah, so it's an optional that you can. You can do the Basic Dax with just a mobile phone, but if you want the highest performance, some of these added functions, we had this physical like it's ambient sensing device that we custom-built. It's got mold a microphone array, and it's got a machine vision option. In end we use the microphone array to do all this dire boltholes, speakers get high off the source `isolation for the Audio. And the machine vision we use, it's Connect heat mapping. We don't do full RGB's so you don't recognize the person. It's just like where the person is movement. So they're very interesting. Applications at our clients are super interested, so I'll give you. An example of they were opening up the streams of that audio and visual to third parties through standard eight cloudy Pi's so couple of examples, translation services translation healthcare is a huge deal. If you've got things that can handle multiple speakers with a display, why can't I do language pairs automatically two translation when I talked to a Spanish speaking patient. That's a huge arm. We don't do translation language. Translation opened that up to third party. We're working with one partner on that now. Second one is is a bunch of startups. Do Depression screening. From Acoustic signals proven that from your voice patterns you can predict the likelihood of status of some depression. Cute right, and so we're opening this up to these people thought like a mental health clinic and again patient consents all this right, the pace of wheat. It's all about for the patient's consent to be recorded. We don't record but assume that they can send Arkansas real over ninety percent. Because we don't do anything else with the audience, just help the physician. Take care of the patient, right? We're not selling ads or doing something crazy with the data. You can actually do depression scoring off of that screening. That's a huge value at. In the in the hospital room at the bedside. Tons of application, so people really fair for older people full risks. Patient gets out of bed and falls agenda. Be Older patients. Break their hip. You have a camera in there. You see the patient moving in a certain way towards the end you can alert the nurse. The patient does not the ring for anybody who just avenue alert. Another big thing is is is is San. Cleanliness like handwashing like is the number. Lack of handwashing is number one caused for. Infections in the hospital which killed a lot of people. Unfortunately, you can actually see you. People wash their hands when they come in. There's a whole suite of kind of these ambient. Application that's why the next generation of voice is ambient. It's this passive sensing thing that brings intelligence to bear without the human. Ask for something all the time, so those are just a couple of really simple examples. Where we think there's going to be explosion of these new applications and some of those are a lot easier to solve in our core, Kodak's product or trying to convert a conversation to a full of full note, so some of these are even easier, technically problem is. These people these people that bill this have no access into the environment that they need to be in so for us. Ninety percent of hospitals and sixty percent clinicians already use our stuff so for nuance to stick a device into the exam room is not that hard with clients because they trust us, and so once we get that point. It's like having a smart speaker in the home, but it's a much more narrow constrained environment where we could easily have millions of exam rooms. Rooms connected to the cloud to then enable these other services. This is also a big reason. Microsoft got excited beyond. Just generally wind up. Be More relevant in healthcare and work with us. On this Dach- from is they have a lot of interesting acknowledges that Aylwin plus their third parties. Where if you have this ecosystem play, you can start to bring a lot more value to the kind of in in workflow at the point of care less That's really exciting I think. That's right now this device does it have speakers as well as a microphone? End Display. Display as well now. I got it so patients address it y'all they can use it. They go running for the patient, and you can imagine where if patients put in the room and instead of the nurse and clipboard. Why can't you have a virtual check in process using the device using a virtual agent totally can do that. We have a bunch of clients that are looking to do that. There's no reason at a nurse at the do that for the first intake. Look at the data asked a follow up question but the Ashram. Watch how you're feeling. Are you feel date you know? Anything changed since the last visit. You feel threatened at home. You know all the questions you get asked over. There's no reason of ritual as you can't ask a question. Right. Makes makes complete sense to do it that way, so do you expect or action? Say Do you have anybody who's deployed in the field today? We just rolled out our dash solution. That covert hit, so we got rid of the hiccup with that but I would expect in the next six months. You'll start to see not just yard already deployments at the core DAX, but they're using. Mobile devices as the audio input, an end, the device, the physical vice, starting to get rolled out as we speak, and we're working with a number early partners on this ecosystem place I'd say the next six to nine months. We'll have some real I don't early indicators of some of these other use cases. What's the name of this physical device? It's called the I think it's GonNa be. The DAX note the latest names I think they keep changing. It's it's it's just the it's the DAX. Ambient sensing device okay. And that'll essentially you'll be rolling that out into. Pilots fairly soon. Yeah, we started to do that already. Yes, okay, so you do have some in the field we do have sauvages, Kobe visiting people tougher for the last couple of months, so we have a small number and then. where? Where we started to ramp that up as we speak. I I understand so just like a final question. We'll sort of get away from covid since you're working both chat and invoice. What's your expectation about? We just do in terms of conversation, so we can say clinician side patient side. She can break it down. What percentage of conversations do you expect in the future for? Interactions are gonna be voice versus chat. So on the physician side vast vast majority is going to be voice i. don't know what the percentage is but vast. That's that's the way these. These. Over Ninety over ninety. Yeah, when they're dealing with patients, they are looking up something in the HR. There's no reason they won't while they could speak to. They use cases of that, but it could type in the patient physician encounter yacht, not ninety plus percent voice I on the patient side. It's GonNa. Be Much different I. Think you know it's going. I think it's going to evolve I. Am this is you know I? Don't know quasi educated guess probably like seventy thirty digital non voice over time. It's not there today because most of the. Applications like chat email web browsing. You can just do so easily through tax, but there are certain applications certain user groups like older folks. Who feel very comfortable invoice voice I think for certain use. Cases will be very different like I could imagine. A. There's a bunch of senior health coach applications being deployed onto like. DOTS and Google homes. Where where the the senior treats, the the the assistant like. A social thing and the social dynamics are much different when it's by voiced. Tax, I think the high. Those are going to be much more voice related because there's already studies that show they're not as lonely. When talking to their assistant, and physically they're not as facile so I think the average, maybe seventy thirty on the tech side, but I think you'll have these pockets that are very heavy voiced where we say seventy thirty. You're saying seventy thirty. What more chat than? Mork more more chat over time. Yes, now it's the you know. It's eighty to ninety percent voice today in this is a journey, but some of these use cases where voices so much more match voices so much quicker so much more natural. The voice thing I think is gonNA. Really be the dominant thing, so I think in the patients. Some of these at home nursing home post acute by voice. I think simple applications. Where they want to interact with like appointment checking that's going to be bygones of you're doing like more of a symptom checker type, thing or more complex transaction. My guess is that's going to be done over the texture. CHAP! Got Okay. Peter Durlacher, that was great. We've been talking for a long time. I appreciate you. Spending so much time today to share a little bit more about what nuances doing today, how can the listeners learn more about what you have gone out in us, healthcare and keep track of all these developments you have coming up. Yeah, thanks. Brad thanks for the offer in the time is really a great conversation. I mean clearly we have our website in dot com. We have A. WE RUN A ton of educational series webinars where clients speak about what we're doing. Those are all on the website. That was very useful. We also do a lot of. Events through the healthcare or Health Management Academy Chime Scottsdale Institute if the event. Ever happened again for those two though they'll be virtual, also lots of ways to learn about us I probably started the website. All right perfect. I voiced by listeners. Thank you for listening once again all the way to the end I'm your host Brechin Selah were here each week with people who are making a difference in the voice of the conversational ai world, and was really happy to be able to bring nuance not this week. You can find me on the twitter African. Selah read voiced by day I. AM WE HAVE A new report coming up pretty soon around voice? Voice in health care based on some recent survey data. We've done post Kovin, so we'll have an update on a report that we did last fall. We were looking at for use in healthcare as well so definitely. Look for that. You'll find that researched at foist dot I researched voiced by that Ai Peter. Dir lack of new ATS. Thank you once again. Thanks Brad. Appreciate the offer to join you today. Appreciate the time, thank you.

nuance HR US Alexa Google Alexa Dragon Systems Google partner Peter durlacher virtual assistant apple Peter lax CEO Pete Haas Amazon Vanderbilt Electronic Systems EHR
Bixby from a Developer Perspective with Murphy, Kibbe and Haas - Voicebot Podcast Ep 70

The Voicebot Podcast

58:47 min | 2 years ago

Bixby from a Developer Perspective with Murphy, Kibbe and Haas - Voicebot Podcast Ep 70

"This is episode number seventy of the voice podcast, today's guests. Joe Murphy, Roger Kibi and Pete Haas. We talk about Samsung Bixby, and what it means for developers. Okay. So I wanna take a moment to thank our sponsor this week whittling. Go they have new product out called cast lingo so cast like short for podcast, and you can check it out at WW dot cast, lingo dot com. Okay. So let me tell you what it is. If you're an author a podcast or a coach teacher, an expert in something where you have an audience, it'd be interested in what you had to say could really benefit from that thing cast lingo is the easiest way to publish a micro cast on Amazon Alexa, and what micro cast will think of Twitter is a micro blog. It's a short form of podcast. And it's really simple to do you fill out a form online at castle and go dot com in three to five days. You'll have a published. Amazon. Alexa, skill you can name it. What you want? And then you in the interim download mobile app in it. You just press record. And you have seventy seven seconds to leave some sort of message a joke, an inspirational quote, some type of information you wanna convey you. Just submit it, and that becomes live in your Alexa, skill instantly. So I checked it out. So you can check it out yourself voice Bhatt says, so it's called voice bought says, you can say Alexa, talked voice about says. And then you hear some sort of comment from voice by something. We're working on our first couple are going to be about data and the industry not surprising to many of you. So I definitely think you should go. Check this out. It's a really interesting innovation. And I think this is the type of thing that we look at with voice. And we say, hey, how is different what can be different about? How we work with these things. Kathleen goes a good example of that. So check it out at WWW dot, Kathleen, go dot com. WWW dot Kathleen, go dot com and many thanks to whitland go for sponsoring independent media voice about podcast now to the show. Welcome back voiced by listeners. I know that many of you loved my interview last week with Adam char about the Bixby launch. It was an exclusive. He went deep on the background with Siri, and with vivid to Bixby and actually revealed something most people didn't know that this whole journey of his started twenty five years ago. So if you didn't listen to the interview, I definitely recommend it because it's it's important around the Bixby story, but you know, what it's important about the evolution of voices systems. And so for me, it was a rare treat to get so much time with him right after they'd launched this product on the floor of the Samsung developer conference in San Francisco while I was there. I also got the chance to speak with a number of developers who've already worked on the Bixby platform and some of those voters have also worked on Alexa, and Google assistant. So they're able to provide contrast as well as some insight around Bixby itself, which was really unique opportunity this early in the development process, and I think something that a lot of people wanna know. Okay. So Bixby is different. It's interesting. How's it different? Because most people already have a frame of reference. So that's what we saw it to do. So our three guests are all working around Bixby, one of them, even the eventual winner of the ten thousand dollar prize for the Bixby capsule challenge that was Roger kitty. But I I we're gonna have Joe Murphy. He's the founder and CEO vocalized dot AI. They conduct performance testing of voice, assistance, and smart speakers. You've probably seen some of his data and the voice about that. A articles were shows what's better Amazon echo, Google assistant in terms of general performance characteristics. How they understand accented speech. So he gives his perspective on what's coming out. From Samsung from a performance standpoint around Bixby, Roger Kibi is founder of voice craft and Pete Haas is founder of conversation curve. Both our developers that were finalists in the Bixby capsule competition, they also have extensive experience building for Amazon election, Google assistant. They offer their perspectives and developing for Bixby and how differs from those other platforms. So before we get to our inter. Views. I want to thank further chief who gave us a five star review on apple podcasts further chief said, quote, my favorite recent interviews are Emerson Sklar Jeff Adams, past present and future voice assistance for this newbie. It's fascinating to hear how much history there is with voice technology, and yet how credibly early we are in this movement to smart voice assistance. I think that actually echoes exactly what dot keet Laos, Adam chyers colleague at vivid. Embiid Spey it Samsung it is early. But there is a lot of history. So that's what we're trying to do. We're trying to tell you the history of voice assistance, voice technology, and what's going on today and eventually where that evolves. So thank you very much further chief for the five star review. If you are a listener, please take a moment to share your thoughts about voice about podcast and the apple podcasts or tune store in the Amazon, Alexis skill store or your favorite podcast app. It helps other listeners discover the show, and if you do so you just might get a shout out the voice about podcast. So thanks a lot further. I'd now do interviews. You will I hear Joe Murphy for vocalized day. I then followed by Roger Kibi who is the winner of the Samsung developer challenge, which we didn't know at the time when I interviewed him happen the next day, but congrats to Roger for that. And then filed by Pete Haas. Who was also a finalist and has really interesting things around product search and basically a store that he set up on. I'm Bixby creating new capsule. So enjoy the interviews and let me know thought. Okay voice about listeners. I'm here today with Joe Murphy, founder CEO vocalized at a and we are at the Samsung developer conference two thousand eighteen San Francisco at the Mosconi center, and we had an interesting morning today in which oh, yeah. Really was. Thank you for the interview. And we attended the keynote session, and let's just get the elephant out of the room. Okay. There was a foldable phone. I know amazing has nothing to do with voice. But while it was impressive. And now we can jump back to sixty. Well, so yes, there was a photo phone and we saw an action if you it downstairs yet. No, I haven't okay, I've got to go. I've got to go check that out. I don't know if they have it down tires out it because it was in that guy's pocket, and I don't think it's leaving his body. He took off. So that that was one of the things that I I pointed out to people on Twitter because I was just like an aunt hopeful phone, and I've been talking about Bixby and all these other things, and I'm like, oh Bixby big launch pretty significant stuff. But all the headlines are going to be about the phone. Yeah. But it is notable. They opened the keynote with Bixby an AI and they're glad you said, yes. And kind of contrast that I think to earlier in the year at the apple developer conference. How voice was such a miniscule part of the show and here Bixby invoice really is layered on top of almost everything they're doing except the foldable phone. Yeah. The foot will foam was sort of like that one more thing the end which used to be a big thing at apple developer conferences, now is like dark mode is one more thing. I mean, it's sort of leaves you wanting more. So did you go to the apple developer WDC this year? I was there earlier this year, and I was kind of disappointed on the lack of Siri horsepower. At the presentation. It was really just shortcuts. Yeah. And that really didn't have a voice component. And we've seen it. I mean, it's not that impressive in. Traditionally I always so originally I came from the mobile phone world. So traditionally Samsung apple are the two competitors that you see going to head with the iphone versus the galaxy. Now, I see starting today shift where who is Samsung's main competitor might not be apple in the future because as they're going more and more towards the expe- smart things, and what really hit me was the multi modal nature of Bixby. It's more than Alexa. It's more than Siri, and it's heading more towards Google assistant territory. Right. And I thought that could make who do you think the number one competitor is now Gogel, but their number one partner for mobile. Who was on the stage. Yeah. Google is up there with only third party on the stage. I believe right. Absolutely for the keynote. So did someone you think sort of think through this strategically and say, you know, we have to reaffirm part of the Android community. So we need to have a what what was he up there? Three minutes. Yeah. Three minute interlude by someone from Google. Do you think that was really the motivation behind that? I hate to think what people's motivations are. But it would seem like they needed to give them a nod and some assurances that we're in this together. But I think at one of the sessions after that, which was solely focused on Bixby. Right. One of the questions from the audience is what motivates me to use Bixby versus Google assistant when they're both on my phone, right? And I didn't really feel like there was a good answer. It was not clear to me why somebody would be motivated to use Google assistant. Versus Bixby especially because they came out with one of their core principles was one assistant. That's right. And they kept reiterating that and I wanted to raise my hand and say, well, there's about a couple hundred million people out there that have already chosen their one assistant. So how do you win them over and kind of what came out of that was Samsung, and I wrote down the number was shipping. Five hundred million devices a year, death, right? So they do have access to this hardware ecosystem. That's right that their competitors. Don't have and I could see that as the way for the Bixby layered across the TV. The refrigerator the mobile phone. So they have every touch point they need, including the smart speaker now at the galaxy home that that could be their way in the house, but other than that, it's it's an uphill battle for though. I think with with Samsung, it's always you have to look at a couple of different ways. One. There's always Korea. And so having a full suite and Korea and dominating the use cases and experiences. There is a part of Sam strategy always global company, but Crean company as well. And obviously even bigger in Korea than than they are here in the US. The I think it's really interesting that you bring this up about who they're competing against and what the rationale is. Now what kit Laos did say a couple of times that. You know, it's early he seen this before he's been through a couple of these rodeos, and that people haven't really selected. He an example is he said there were seventeen browsers when Google or seventeen when Google first launched right? And so there's not even that many voice assistance out there. So he didn't say that. But that's that's it was the location. So you know, if it's better it'll win. Yeah. If it's better it'll win, but you do have that embedded base. It's I agree. One hundred percent it's early but it's not that early. And some people have already made their decisions. This is the kind of conversational assistant that I'm using on a day-to-day basis. So you'd better give me a compelling reason. If you're gonna ask me to switch. That's right. And I actually believe that the the one assistant would probably work against them. But I don't believe it'll be one assistant, and they kind of hammered that point one assistant a couple of times and the other. Question came to my mind with Amazon, Alexa, and Microsoft in court Tana, and the assistance talking to each other and wanted to hear kind of labs impressions on how they think that's going to work because it goes counter to the one assistant unless it's done invisible to the user in the background. But I don't know if those kind of partnerships would be realistic. Yeah. I mean, I had interview with childcare from Samson vid labs magician the magician. Yes, he did. Magic trick onstage with Bixby is really interesting actually did to the breakout. Sessions one as well. This is fun. You know, they really believe that they will be the one assistant that they can win that. And it wouldn't surprise me at all. They can't be for some people. Samsung has a lot of assets. Google also has a lot of assets that are also very valuable and Amazon certain countries ecosystems has a lot of assets as well. And so, you know, my expectation is that this is what they've rolled out today is significant. And that frankly, if they did open it up to Bixby being a service to Google assistant as well. So that there is interoperability it could actually be a wonderful top. I don't know the Google would necessarily wanna do that. But they might do it to sort of maintain position with good partner on the Android side. And I could see then people would have access to Bixby even if it wasn't their primary assistant, right? And I think that could be really helpful because of the number of domains you have to cover and how you do that. It's going to be difficult for everybody to do at themselves. It's definitely going to be interesting. I think the galaxy home coming to the US will be another touch point that will help set the tone for how he's going to do within the market. I have high hopes and high expectations because Samsung always makes beautiful hardware. Right. And the more hardware they make and the more touch points. They have around there around your home. The more likelihood that you're going to interact. With Bixby, and because Bixby is now open and they announced a whole suite of tools today for developers to develop their capsules they call it. Right. It's great to see the energy and the investment that's hands lungs putting into it. And getting back to the point you mentioned earlier in the early days of search engines, there were probably a dozen or so and now there's one or two main ones that everybody has one. There's. I could see a similar thing. So I can get back to that point and say Samsung's making the right moves earlier would have been better. Right. But what they're doing and with their hardware ecosystem and with smart things. And even the plays in automotive with Harmon, they're setting the table to succeed. It's just now they have to give users and consumers a reason to really make the jump. That's right. That's right. And so what about the the the launch demonstrations today? Stuck with you the most probably the the tools that they're now making vailable? So I was really impressed with the openness and the approach taking with. I think it's Bixby developers dot com. You can log on and start developing. Now what they're claiming the quote, that's resonated. With me was that third party developers are now treated as first class citizens. Right. And I think developers will really gravitate towards that. Now. I haven't spent enough time on the tools to say. Is that actually accurate or not? But if it is accurate, I think it will attract developers, and that's something obviously they need to do right now. How many skills does Alexa, have forty thousand fifty thousand seven hundred thousand whatever the the forty five in the US. Yeah. Slow sixty. But I think getting people to now right skills for or capsules not skills for Bixby is the next challenge that they have. It's probably the biggest challenge that they have the tools look attractive investments there. We'll just see time. We'll tell I think for developers a lot of a lot of home. Listen to this podcast are going to be sitting there thinking, okay, I'm not making money on two platforms. Do I really want to not make money on three? Now. I think it's fair to say that they talk about this marketplace and to roll out early next year that's close as they would say to a date and marketplace will also include tools for monetization. They weren't clear. If when marketplace comes if monetizations day one or not it's not in the today. One thing they did say is they're very flexible on the business models on monetization. They didn't come out and commit to. Here's our cut. Here's your cut. I think they realized they need to be flexible, which was really good to hear. This was top people at the lab saying it. So it's good to hear that. They're thinking at the top bef- lexical monetization, and how they can make it work for big companies and small companies, and that was another point that DJ co actual. Closed with. Don't know the exact quote, but he he came out and said, whether you're big company or your small company Samsung is committed to work with you. And I think that's another thing that will resonate really well with a lot of the players in voice right now who are brand new startups getting into the space and Samsung is basically inviting them in with open arms, right? And this wasn't like some division executive saying we love developers. This is the president of the company. Yeah. Up saying that we're committed developers, and he in particular has really been out front. So he's taking the lead on rolling these things out in the February van last year's event who you can see that the commitment is at the top. And I should point out that just to elaborate for those who've didn't see the keynote that when they talk about this idea of developers being first class citizens with with their very explicitly saying as the other popular voice assistant platforms. The tools that the first parties have native, Alexa, native Google are different. From what the developers have the developers? Don't have access to all of the features. Correct. And so when the platforms decide to roll out their own first party capabilities, they often make the third party capabilities that are somewhere look wanting and its place, you know, Amtrak out there and said to me we use the exact same tools as the developers. There's no difference. I wanted to touch on one point that. I think they've labs brought up several times. And Adam brought it up was how do we take voice from being a handy thing to have to an essential thing to have. And I do see Bixby with their hardware ecosystem being able to bring that multi-modal Bixby voice and displays was it tap touch talk. Yes. That's right. So they do bring that across their hardware. So it'd be really interesting to see if they can bring this to it's now essential in my life versus a nice to have. I'm glad you brought it up. So they talk about multi-modal, which is true. But it's really the the interesting thing about Samsung, I think is multi surface and the surfaces being all the variety of devices having a common experience across those. It's not clear to me right now that that's necessary. So do I need to have personalization for my. Microwave, for example. Maybe I don't know. I'd have to I'd have to see it in practice to really know. Do. I need personalization on my thermostat. Maybe people are doing that today. Some people like it, some don't like the personalization watch. We see backlash of that. We certainly see like personalization around travel, those types of things that makes a lot of sense because the Shembe taste. So we'll see you know, that personalization. But this whole idea of multi surface in a way that it's going to take a little bit longer for Amazon and Google to get to in part because the they're doing so little personalization today, and as part because of the way that the hooks that they have into those other surfaces very light and on top of multi surface, which I love that term, by the way. That's what I was thinking. But I just wasn't coming up. But on top of multi surface. They also introduced the one you. I right. Oh, right. And it's nice to see this new clean. You I across those surfaces. And now, I'm really getting into a place where I'm interacting with devices very intuitively because I know how to use my phone, and now a similar you is appearing on my TV and starts giving me that feel of in the system, and it's very easy to move from one surface to that's right. And and it is worth pointing out that among those five hundred million. Devices. They roll out every year, they are number one in smart TV's shark. In fact, I think they're like number one by a wide margin. Last time, I looked at it. Pretty sure like the next three don't even add up to them or something. So so definitely a big thing. Why don't you tell the voice by listeners who you are? And what you do just a brief overview. Sure great. My name is Joe Murphy, the CEO and founder of vocalized AI. We are a research company working on benchmarking services. We developed software tools for benchmarking ASR performance. We're looking at really getting the industry together to start discussing how these devices work, not only for the big players in the industry, but for the third party partners and quantifying how well these devices work benchmarking, performance one device versus another and informing consumers. This always listening device you're bringing into your home. There's a few things you should. About it. Some of those things are how it handles accented speech. How it handles waking up accidentally what happens to your formation. How it handles different types of talkers such as youth versus elderly. These are all things that we've been working to develop a suite of tools. So you can run the device through the test. We and then get a feel for how it performs compared to other devices out there, and there's a couple at least a couple articles maybe three or four that we've collaborated on you've written. Some I've written some of those about some of the data that comes out of your testing, which I think anybody in the industry should look at that. And what I really like about what you're doing is you're creating this database that we're gonna see change over time from. Yeah. Absolutely one of the things that drives me nuts in Silicon Valley is ABI tending these meetings and somebody will get up on stage and say the problem of speech recognition has been solved. It's a solved problem. And then in my lab, I see if I use accented speech or youth speech or elderly speech or lower volume or entered his background noise. The real world environment of how these devices is being used has not been solved. So one thing I would like to drive home people, please stop saying that speech recognition solved problem real world, speech recognition, how people interact with devices is not solved problem is. So well said in fact, I just did a national survey of US adults. The number one thing that people said is keeping them from using voice assistance, more is it doesn't understand them well enough, and they said if understood me better, quote, unquote, they would use their voices more. Absolutely. And I think that's probably a good way to wrap. This would be perfect. And you know, what we'll see? I mean Bixby the one thing that really didn't get confirmation of. But it is a new Bixby. So this is. Not the Bixby that rolled out in 2017 seventeen. Correct. But it's not clear to me when the new Bixby actually hits the phones and those types of things so those are some of the things we'll be looking at. But just for the listeners if they want to follow vocalized, I follow you and social or something. WWW dot localized. Oughta is a is the website and were on linked in. We also have a YouTube channel perfect. Okay. Great. Thank you very much. All right. Cheers. Thank you. Okay. So we're on site here at Samsung developer conference in San Francisco. The big news was well foldable foam. But we'll put that aside. The big news was Bixby as we know is Bixby to Dotto a don't ever used that term. I had the opportunity to sit down with a couple of developers who've already worked on the platform with some of the early they've done some demonstrations and of the call sort of the beta release, and my first guest is Roger Kibi. Roger, why don't you tell the voice about audience your name? What you do? Just give them a little bit of your background. Sure. Roger cabbie and about twenty years of technology. Experience a lot of that working for big companies and helping drive their technology strategy for the past year and a half. I've been pretty deep. Immersed in the voice ecosystem originally building out a company to help small medium businesses established voice presence. Now, I've I've cut a pivoted over to provide voice consulting services for companies to. How do I stash voice presence? What's that gonna mean from a business and technology perspective? Okay. Great. And you have been doing that largely on Amazon and Google you didn't start I'm Bixby, but you were enticed at Bixby. So let's start with this idea of what stood out to you today. In terms of the announcements, the keynotes that you participated in particularly given the fact that you've seen a lot of this already because you've been working with. You know, it's hard to because if you've already seen it, right? Some of that stuff was was not new news, right? And things. I think one one fact really stood out to me Samsung ships five hundred million devices a year right now, all of those are going to be Bixby abled. But some large portion of them are going to be which means that all of a sudden Samsung could release an enormous amount of BIC speed enabled products in a short period of time. And then that added with already been pretty excited and interested in how Bixby and really underlying technology their take on on how to think about and build voice, applications or voice capsules the Samsung term that quivalent of skill or or at action how tell me about their philosophy the protest, they take. And how it's different. I I would say there's a very kind of NL you kind of a. AI focus there, which at first when you go in there and having developed several Alexa, skills and a couple of things Google actions. And it's a different take different way of looking at things, and and I had kind of rearrange shift, my brain, so to speak to understand how that would that would work. But there's some pretty interesting things, for instance, when you trait do training, right, which you do all those this. You can see there's almost this. There's an explanation when you run, and you can see how the NFL you is actually processing the data has analogy us today for the old database people explain plant, but it's an explain plan for NFL you and I've never seen that before. And it's actually pretty interesting and informative to have that there because then you can understand if it didn't understand me or didn't respond the way I wanted to I can see underlying mechanics and how the engine understood that the phrase there. I think I think that's really fascinating. So they expose how the assistant is interpreting what you're saying they do in a much deeper and richer way than anything I've seen on the Alexa. So for troubleshooting. This has to be very useful. It is it is it's it's really nice. And do you aside from that? Like what you can see do you really feel the Bixby handles intense differently than Alexa, Google assistant. I'm going to say. Pro not really and things like that. I mean, the kind of the way it handles things in the way, you set things up the way, you do the training to me is really reminiscent of Google's action flow, actually. Okay. And they are and that reminds me somewhat of it. And so that isn't unique. It's this exposure of of when you go, and okay, I train it. And then I can see this kind of example model I could actually write a little app speak to it. And then I can see how it interpreted that. That's that's the unique part. It's not so much the training part. The development is it's a little bit of different paradigms. Whereas I'd say to my mind, Amazon is mostly co driven Google is there's a lo- more configuration driven. Right. You write the code. But but dialogue flow is there's a lot of configuration here. Samsung is more on the configuration side of matter of fact, some in some ways even more than dialogue flow. Although the tool at this point, probably isn't. Has has all the bells and whistles that dialogue flow. Doesn't look it doesn't have a look and feel like a dialogue flow configuration. No different. It doesn't. Because a lot of that configuration is in these text files. Right. So for instance, with dialogue flow, you're doing a lot of configuration. It's a it's a UI base configuration pulled down a lot of settings. Very you're doing something similar with a with a capsule, but you're actually just Iraq. She writing in their their these dot B X files, which the configuration files there. So they're they're you I requires you to kind of understand their configuration file format and how to do that. And that's definitely there's nothing wrong with that. But at I it's a challenge to kind of understand how that works and how to make that all fit together. Okay. So why did you decide to take the plunge into Bixby because you're making so much money in your monetization for your skills? Oh, yeah. If yeah. Exactly, my monetization. On my Alexa, skills is just amazing. Same challenges. Everyone house. You know? I think it's. I looked at it. I'd I'd read about and seen some videos online about the underlying technology and does really interesting approach not got me excited. And then when I thought about this Samsung, the reach kind of quantify those numbers that I heard today like, you know, this could be pretty interesting platform. Let me go try some things. Let me write the, you know, the the Hello world apt that everyone writes, initially and looked at it. And then there was the developer contest to go. And you know, sure there's a little bit of financial ward. But everybody likes early did lexicon. Finalist and I'm fine. Thank you. Thank you. And you know, we're all going to. How you're buying dinner tonight. Exactly. We're all going to the final presentation is tomorrow. So yeah, it was a little bit. Try something new and because I got some hints and ideas that they might be doing things a little bit differently. Go try it, and certainly the resources and the size of Samsung means they could they could put a pretty big dent in the market if they have the willpower energy to do that. Okay. So tell me what your capsule does. So my capsule fairly simple. It's called what been and the whole idea is we've all gone to dispose something there's a recycle bin a compost bin trash bin. And you're like, well, what do I put wear a so I'm not the only one exactly half the time. We we put the wrong things in the trash bin or and and too much in the trash bin the wrong things, recycle and combos. And I was like, you know, that data's available about where I should put things although a little bit messy side to get that data together. So it's relatively simple. You simply ask can I compost? This can I recycle. This can. I trash this. What been this go into? There's about a thousand different items that understand. Okay. So that's yeah. And understands and only you tricks around the you have to build all the intents, and yeah, yeah, it's a little bit. Like I said, though, if I were to build this as Alexa skill to compare. It would be writing a lot more code because a lot of is configuration driven. Okay. So a lot of the with Alexa, the intense, and I'm writing coat a lot of that is I had to configure these configuration files understands these things that actual amount of code is. I dunno fifty percent of what I might have had right with Alexa, which is so much of it is done by Bixby itself. Does. Yeah. Bixby does a fair amount of itself. Yeah. It's like having a giant library. Yeah. Yeah. You could say that. It's like having stack overflow right there. Exactly. Yeah. Yeah. So the other thing I wouldn't mention which I really didn't get exercise. It looks pretty interesting about the platform is this idea of learning and what they mean by that is really personalization. And it's smartly understanding if I asked this if I keep on answering the same question the same way and number of days or -cations arou- stop asking that because I'm on soom. Right. So this is the personalization personalization feature. Did you turn that on? I didn't. Because for what I needed it just wasn't needed. And I'll be honest, their documentation isn't a notice where fully needs to be to fully leverage these nor the amount of sample code. So nothing I'm saying that they don't know. But that's pretty it's contextual e aware time during the week. I asked different questions during the weekends the morning ASA different one than at night that looks really interesting to me I could do that with with both Amazon and Google, but I'd have to write a bunch of code to do it having the framework do that. Looks really intriguing to me definitely want to understand that better. But I do think with conversational interfaces the smarter. They are and the more. They understand me. And who I am. But the better they can be doesn't worse than asking the same question. Every time I invoke rice skill action capsule. So that looks super promising, but I really didn't exercise it fair enough. So you're mentioning to me that you got to meet a lot of the core Bixby platform developers. I did now were they were they were they give labs or were. They samsung. Well, you know, you gotta ask how long they've worked for worked in this. Right. So yes. So one thing that Samsung is a lot of their well, all of these conferences law. There implies are here. But a law the core Bixby people I met one guy who'd been there for five from labs, and he is writing a lot of underlying code, and he was just super excited to. Actually, see people get it out there. Get it out there. The end, you know, the end developers so to speak use it, and we had a long really interesting conversation, and he's super energetic and excited because like any developer you love to see people using your product you love to hear what they like. And don't like, and and lots of energy there's a product manager I talked to a project manager. And so a lot of the Samsung employs were quite interested in us as kind of earliest earlier earlier among the earliest developers. What we thought of the platform? What worked what didn't work? So they're pretty open right now about oh, you need to talk to this person. They'd love to hear about that. Right. So it's indicative almost and you're like one of fifty. Yeah. I mean, I think someone throughout the number seventy today, but we're not talking about fouls, and we're not talking about one hundred. We're talking about something lesson that in fact, I think just a few weeks ago, they were they were like. Forty and they were they wanted to get more people in, and you know, frankly, I don't know that they really advertise. Very heavily. Yeah. They're trying to figure I mean to my mind developer vandalism. One of the kind of livered I things of any platform right there. You know, you can definitely tell there's been a lot of underlying engineering effort. But as far as the next step, which is right in a lot of great documentation. A lot of great sample code developer evangelists, whether it be events like this contest or media beat up and that they're still developing the front end of that Prosser front end of that process. They still have a lot of work to do. Which makes it more challenging some ways to build. There's not a lot of people to answer the questions. Right. And Aaron is not a lot of sample code there. Also makes it I'd say more fun in some ways. Right. It's a wild wild weapons. Yeah. Sounds like what you're saying. Is that you really do need to take the time to learn it. And so there is a learning curve associated. Whereas, but once you understand it. There's a lot of things that make it go really fast. So if you make the investment, there's an accelerated on the back end, I think that's fair. I think that's fair. I think it's I think the investment I wouldn't underestimate if you're coming from. Let's say Alexa, background that the amount of kind of relearning and rethinking you to do because it's quite a different platform and the way it handles and thinks of the one thing that you're like, oh, that's totally different. I've got the really reorient. It's gotta be the it's all the configuration files and understanding. It's a metal language that really is there, and it's like, wow, this is like a simplified programming language to to learn. Well, it's funny because was dock said it in the this afternoon session. Maybe I don't think you were there. But he said or maybe it was Adam they said, oh, you know, objective c. Right. They were talking about the app store needed objective c in order for to do a lot of things you want to do with apps, and they weren't really saying that they had built something, but they're sort of intimating. There was kind of like that for for voice voice assistance. I think there's been a lot. I think this kind of meta language they've written to configure. It is got some thinking around how to think about not only voice. But hey, I is there. It's definitely you can see that. There is a a methodology in a way of thinking behind it. I don't know if I can put my exact finger in what describe it that. Well, but there's definitely a philosophical belief that this is how you should do things that pervades that product. So when will the voice spot podcast listeners people to try what been so my understanding is in early twenty nineteen that's the best date. I would be one of the first hundred whatever. So right now, they're trying to sign up. To build their capsules rights. We have Aisher wrote a couple of them and trying to get you know, when they released their their capsule store, whatever their place market. Okay. Thank you. I'll be an early early developing early. It's just pretty exciting. Right. You know, one of the challenges in the Google and Amazon ecosystem is pretty. Yeah. And you know, I mean early mover advantage, right? And get get some traction air. We'll say that's another attractive thing about the platform. Is is hey, you know, we all discover Bility's easy when you're among fifty or one hundred when you're among forty thousand is a lot harder charred. And so how can listeners check out. What you've done on. Alexa. Follow follow you on social. Yes. The best way would be on Twitter. So at Roger Kibi R G E R K I B B perfect. Thank you so much. Roger, I appreciate your perspective. Thanks a lot. Brett, my pleasure. Okay. Back at the Samsung developer conference, and I got the opportunity to sit down with another developer Pete Haas who has worked with Bixby Bixby Dotto. That is so well, why don't you just tell the listeners your name? What you do what your background is. So my name is Pete Haas. And I've been developing for voice since two thousand five which is nice thirteen years worked in an IB, our company, of course, it was the only option for voice really just follow that journey and probably wants to re came out. It was a really big deal to have that experience. Right. Did that blow your mind? It really did. Because I. It was just something. We never experienced before this open ended thing where it was just kind of democratized to an iphone, right? You are doing all these rules basis. Domes Harz terrible. Harz terrible build something that would do that will cost you a million dollars. If you build something quivalent to what the airline systems were doing at the time, right? So fast forward. I was on a quest define. Well, how can I find a speech recognition that would be as good as let's say Google voice? And then Google released an API. Oh, great. I have transcription at a level. That's pretty close. Right. But I was missing the one piece which was the piece right? So even though I could get the transcription, I know ano- you head. No. So I went to the signal conference in two thousand sixteen and they introduced a million API's that we're doing the same thing. Many of them got bought since. And that was the key. Right. So ever since that I've been really on that path and twenty seventeen I opened my own voice agency. It's just voice wanna do a mobile app or website. You can call somebody else for that fair notice dissenter DNA, and I've just been kind of following that path and that led me to the Bixby developer conference here Samsung developer conference, and okay. So we'll we've talked about this with with other gas. You went to the keynote this morning. What jumped out at you? Like Roger said, I mean five hundred million devices a year, right? And their plans to roll it out. I think they said by twenty twenty right? Every. Yeah. This year. I mean, that's amazing. Yeah. Yeah. Quite a footprint. I mean, that's nothing. Like any other companies really have? We'll tell you. This is an aside. I saw the they've got these different TV's. They call it the wall, and there's the other one that's like the frame. Yeah. The frame disappears. It's like, it's if people haven't seen it, I recommend you check it out because it's basically just Ashida glass that becomes a television. But when you turn it off, it's like transparent, you could see the wall behind it or you can float. Now, they've got things where you can float pictures on it and stuff like that. That is so amazing. And I'm sitting there thinking, okay. So this is visual most people listen to this podcast about voice. And that's what I'm focused on mostly. But we start thinking about multi-modal, which are think about the screens around us. And sometimes we ask for voice, but we might get a visual complement to that. And they've got some devices that are going to be perfect for that. Where you could just put them in many places around the home where you wouldn't normally have a television. Yeah. But most of the time it's. Art. But just that one time that you need it when you say something to shows up there. Yeah. Yeah. It's amazing. And then all the other appliances. I mean think about refrigerators or talking to your coffee maker for that matter. I mean, it's gonna be interesting how they roll this out. And all the additional use cases that are going to take place that don't really exist now and the screen component that I think that is really key. It's kind of baked I think the screen component is pretty important. Okay. So let's talk about your perspective. I'm bixby. So when did you first start using a couple months ago September? Okay, I found out from user group and entered the competition and opened up the ID and said what is this? It it looked as Alexa, Google action. It looked nothing like that. Okay. So. Yeah. So you got the ID. So explain to listeners what's in the ID. Well, it's pretty full featured. But I mean, you really have to kind of build everything there's not really you. I where you're dragging out boxes to say. This is my intent and do. No. So you basically have to Drake com concepts. Right. So you write these concepts. What the concept it's a noun. So if you have a product a product is a noun and an action is right. So you take actions on that, and you basically defined that layout. But as I code goes, you're not really writing code, which is fhu to me. It's a winter. So let's talk about this. So this is the I don't remember because like dynamic language programming. Then Emma, Jenn. It's cogeneration. It literally builds it for you. And when it shows it to you, would you can actually see it after builds it. It's actually a flow chart, it looks like a flow chart, and it's it's easy to Rico behind it. Yeah. You can see what's going on. But I guess their argument is that it's Gaels because one of the problems is you can build for somebody say thousand things in time. They say thousand one things your. Not ready for that. Right. And Bixby kind of builds that in and to me, it's not even a voice, it's really an AI platform. You know, it's really baked in where you think of it this way if as a skill developer a lot of the intelligence is on let's say Amazon side. Well, they're doing all the processing. And then they're giving you a request. So you have to fulfill that order, right? There's no intelligence on on your end unless you build it. Right. It's really just programmatic response. Right. Right. And so. Bixby's kind of bring you in a warm hug, it just kind of brings you into this AI platform, and once you get over a fairly steep learning curve because it's a different paradigm. But once you get it. It's kind of exciting to see the use cases, you can really create. So what is your capsule? See? So you from is is the big people say Alexa, gives you a string. What does Bixby give you? What do you mean? By string give you a string requests saying this is this is the this is the request that we want you to fulfill. Oh. I don't know. I mean, it's. I mean, you're basically building this capsule that will fulfil these intense, but you're not even really calling him intense. And you're just taking actions upon them. It's not clear exactly how that's gonna look in production. I guess you call it. So when they release it. I guess I think we'll know a little more how that's gonna look right? But right now, what's interesting is you don't even really have the name them occasion. So you can just make requests for dinner. Raise reservations at five o'clock week from next Tuesday, and that's perfectly valid thing to ask. And if you're able to fill fill it, and it's part of your soul. The other thing is they'll do this thing where. Every assistant is unique to you. So you're going to have your own set of capsules that are unique for bread, and I'm gonna have some they're unique for me. And so when I request something for reservation, look at your. Restaurant that I may be enabled this capsule. Do you didn't enable it did you lay to enable it to like select or does it does the system just select select based on your past behavior. I think the way it's going to work is if there's something not enabled yet. They'll have maybe the top capsules can fulfil that request. And somehow you'll be a blend able it and put it into your kind of personal. Yes. Set of. Okay, personals. All right. So you built a capsule your part of the contest, right and a finalist. Congratulations. So what was it? What did you build? So it's called beautiful, and it's a fictitious beauty supply chain, very similar like Alta. Imagine a brick and mortar stores across the United States with thousands of products. And each of these stores has a beauty salon in the back. We can book. Can appointment for hairdresser. What? So I really wanted to demonstrate the search and discovery features of Bixby so simply asking something like fine hair conditioner. Well, there's thousands of probably hair conditioners. So what will happen is the first time you go through the experience, it will try to collect some certain traits for you like straight here. Do you have wavy here? Trying to fight for all these kind of traits that are specific to you. And then it will give you recommendations based off of that the next time you go in and ask for hair conditioner or some product. It will use that previous exchange. You had to give you a better. And so you've enabled that feature where it learns it remembers who you are, and and that's all done without account linking and all these other things. Right. It just automatically knows who you are as user. Yeah. Right. All right. It'll be interesting to see how they do that when the Samsung galaxy home shows up the smart speaker, which is a multi user device. It's one thing to do it on the note nine, which is a single user device. Yeah. And they made a point to say there is certain experiences are personal to you. And there's things that are kind of shared across a family. A house got it. All right. So this is like a demonstration type of thing. You didn't you don't own. Yes. Beautiful. So you created a number of products in you've been able to demonstrate how it works. If you were to try to do something comparable. Google or Amazon. How would you do it? Would it be different? Yeah. Well, yeah. I mean, you wouldn't have some of the features or at least I'd have to build some of them on on my side. Like the personalization features that are kind of baked into the Lincoln. Yeah. And then the screen is just native to it. So you, you know, that's the thing. There's a special component on the phone where you just press the button and say your rents, and, you know, get the screen response where you can start the conversation with the voice, but then finish it through taps typing. So what something that you really like about Bixby that you think is missing from the elected skills kit or actions on Google. I mean, certainly some of the preference learning where they're building that in. That's great. I love that. You don't have to do the named indication that is just that to me so much friction that to me that's not natural language if you have to speak in that way. Right. So it's not like the operator model. The others work off where you say I wanna speak to. Yelp to get restaurant information. Right. You just ask for the restaurant information. And it may give you yelp. But make you something else. Right. And you struggle if you don't have a brand name like yell. You have some small skill at your developer trying to bring up it's really hard for them to remember your skill name. That's that's right. So you've built a number of Alexa, skills. We were talking about one earlier. Why don't you tell the listeners? What that was. Okay. It's called team time. And what it does is it. If you're a team snap user, which I have two boys and they're in soccer in many sports. And they're constantly asking me, dad. What jersey I'm wearing today. My red or my blue what time is practice. Where's it look all these things? So so all that information. Unlike the team snap website or mobile app. So they could look it up. They could look it up, but they don't because their kids. They don't earn. My wife doesn't even really download apps. So she's constantly asking me because I'm the only one teams so. Now, you know. Okay. Just yes, collects problem. This is a real world problem must be solved. How did you solve it? I just linked it up to team snap has an API, and you can just link into it. And honestly yesterday, I was packing to go out to San Francisco. And I heard my wife using it and she came up steers. I said did you use the app? She goes. Yeah. You want to know what time practice was? That's perfect. Did you get alerts on your phone like someone's using the though I kinda hear the yeah? Yeah. There's a little beep at the that's good. So we're talking about this a little bit earlier that there's a lot of people use this team snap. Apper software, but you still have this discovery and promotion issue. So you know, how how you tried to address that? And do you do you see something breaking through to help make it easier for people to discover and start using the skills? I mean, what I've found is if you are big brand you certainly can shine through the skill store. Or if you know, you're one of the select few that they pick as one other feature skills, do very well, or if you were super early, that's that's worked out great. But then there's this other group of the wrestling ninety nine percent that it's hard to break through the noise run. There's forty thousand skills, and in there, how do you think that will be different with XP? It's hard to say exactly I mean, one of my strategies is because I'm so early. I'm really considering really going all in on Bixby or going a little harder on it. Because you know, you just be kind of the first guy to buy, you know, Malibu beach property, you know, fifty years ago, you know, it's kind of a land grab so strategically. I'm thinking that way that that's maybe place. So they'll have the S nine or ten now for the the flagship phone. I mean, they've got the no nine, and I don't remember how the galaxy S nine is coming up. Right. So so they'll have galaxy S nine coming up March April. I think that's usually when they launch. And so I'm guessing that's when marketplace will come out as well. They seem about right heels. Right. Yeah. So you could be one of one hundred capsules and people have they had there's going be a way to like browse. I don't even know if. There is you just have to act as Bixby right now. Right. I don't know if there is a way to browse and cancer, something's a cancer others. Yeah. But that would be certainly an opportunity particularly for something that's going to be high volume. Yeah. Like Ambien annoy skills. Something like that. You know, certain people have done very well with it is a thing. Yeah. It's very popular. I've noticed that. They do get a lot of reviews to like unlike most things, so so you're thinking about maybe going all in anything else that you say, hey, if you are a Google or Alexa developer. Here's one thing. That would be helpful for you to know when you start out. I'm Bixby super steep learning curve. Maybe it's not so much that it's just it's just a different paradigm. So you really got over that. I it's a different way. It's a different approach in some ways. I think it might be better. Because again, it's more scalable in the responses that you could handle over time. Right. That's great. Okay. So how can listeners learn more about you? And what you're doing? Look me up. Pete hoste. Oh, look at my agency conversation curve dot com, and on Twitter, and all the usual places. That sounds great. Thank you very much for spending some time with us today. Thank you voice about listeners for once again listening all the way to the end, I'm Brechin seller. Your host each week. You can find me on the Twitter at Brechin Sela, our guest this week. I really wanna give them a shout out. And thank them for taking time. Joe Murphy, Roger Kibi, Pete Haas, really great that they can share that type of perspective. So that you voiced by listeners can not only learn a little bit more about Bixby, but you can understand how it's different than the other players in the market how that might evolve over time, and how the bigs befuddling -ality may influence what you start see steam coming up on Alexa, and Google assistant. So watch out for that. I also want to recommend you head over to voice about that for slash research voiced by die Ford slice research for those of you the missed it. We have a new report out called the voice assistant consumer adoption report. We did this with research partners, postering and rain. It is the first research that actually looks at voice assistant us across surfaces. So we did a big report earlier in the year around adoption of voice assistance on smart speakers. And I think that is become sort of the touchstone that a lot of people look at how are people looking at like thirty pages twenty twenty charts. And how people are using smart speakers. Well, we looked at doing the same thing today. We've got twenty seven charts. It's thirty pages of analysis in we'd go really deep into voice assistant use smartphones. And surprise surprise. It's different than how people use them. And smart speakers. So is developer is an industry founder is a marketer. You're looking at this. These are things you want to know. But we didn't just stop there. We we look deeply into this for voices systems on smartphones. We looked at it for smart speakers as well, which which many of you have seen, and then we also compared it to other. Surfaces like here. Ables smart TV's computers, and in the car, so go check that out voice about that Ford slash research. And you can download that report. I think you really going to like it. And that's what we do voice about our research. We what we really focus on is answering the questions that were interested in and the things that we think you're also going to be interested in, but nobody else's looking at it. So because we're so focused on the voice industry, invoice assistance. We can go deep. So definitely check that out. Thanks again to my guest this week. I look forward to having you back next week with another amazing guest.

Bixby Samsung Alexa developer Google Bixby Bixby Amazon Samsung Bixby founder and CEO United States Amazon Twitter Roger Kibi Bixby Bixby Dotto Roger apple Joe Murphy AI Siri