Open Source NLU - Alan Nichol, Rasa

Automatic TRANSCRIPT

I mean what we're trying to build with Raza is the standard infrastructure for conversationally I. So the same way that you have these. have no brainer technology choices, right? If you're adding. Search to your stack, you're gonNA use elastic search. If you're working with containers used docker similarly, if you're building conversationally, I use Raza. That's where we're headed. That's what we're aiming for, and of course, if you want to be that level of standard, you'd better be open source, right? So that's in a nutshell, the way it started was that my co founder, Alex and I. Were, building a few assistance ourselves. We had the first couple on the slack platform. We had some people bang for them. And just realize that the developer tools were lacking in. So many different ways, right. So we were using the tools that everyone was the time with a I was big than dial. was called API. Time the review others. And? While we kind of came with his controversial statement, right? So two thousand, Sixteen Fiscal Messenger Platform was opening up. And there's all this high. You know such an Adela, saying the new APPS and I go all this kind of. Rhetoric. And the day before that announcement from facebook, we wrote a posting. We don't know how to build conversationally I yet. Right? Like nobody knows the tools aren't there and this isn't GonNa go very well. And I think that has sort of played out kind of as we predicted, which was that the push came from people wanting to users through messenger APPs right because APP downloads or difficult to get. Wasn't because all of a sudden. We'd stumbled into a great new piece of technology and we're like, wow, look at all the new things we can build. Right. It was very much coming from people wanting to talk to people in Messaging APPs. And so we said, well, look, let's start for principals and think what tools do. We actually need to build great compensation I and what developers need. And we were working with some of these cod products realize that they're great for getting started quickly, but you very quickly also pain yourself into a corner, right? And the other thing as a developer, right? If you're. If, you're building an ailing startup. You're doing something right and every single message that goes in to your application has to go through a third party to get interpreted. It's pretty precarious situation to be as a developer right, and so we built a few things and we decided to hack together own. System, and then we were running this meet up in Berlin chat bots, balls, Berlin meet up. And everyone was saying Oh we're going to do. We're GONNA. Do our own NLP. You know. We don't have time now, but maybe next month something like that. Right, and always you know what he did, and so we said, well, why don't we just open source hours? And then everybody can contribute rather than rebuilding. Thanks. Very. Innocent insight by this was sort of six or eight months after message from opened up and everyone who is still seriously in the space have the same thought which was. I'm really dependent on the service which might shutdown might so being freedom is charging for it. It's not a great position to be in as a developer, and so when we said look, here's an open source drop in replacement for the clouds will the using. Now, it was just kind of the right product at the right time and instantly got a big community farming around it. So that's was kind of where we started, but we'd been working already for. Six months on a better way to build dialogue, and that wasn't ready for Primetime yet, but we'd already been working on that problem which was okay. It's nice to have this system right and you bring in message and you get back and intended some entities. bought. What do you do with that information and you just layer more if statements on one another all the time that be we're going to get anywhere, and so we've been thinking about that problem very hard and. because. Of the success of Razzano you which was the name of the library, I punched it. We decided to go all in on open sourced look. Everything has to be open source. And that kind of set us down that path burn interesting. Very interesting. So came from chat bots that came from a real need in the developed community using these cloud services I, you had wall stability tools, yourself, you open them up, and the rest is history of the community call onset communities been a real real key to success. Right? Right. From the very beginning, the community has built this company with you

Coming up next