Speaker 2
It's almost like you have to lik to both demand side and supply side. Innovation sense, you needs, like, look of all the problems we can solve, which i'm point of se gesturing on a broadcast my left hand, and in all the capabilities we have, which is t the right hand, where is a good, firm sortof interconnection? So, like, if we take sa or product resolution bout, how would you artike that of the problem solution pare?
Speaker 1
So, i mean, when we started out, we started, we were aware that was like a move in the technology and a move in the product landscape, where, like, you know, bots had been like really poor, and like natty were starting to get like, not so poor. They were starting to, like, younow actually give compelling experiences in very limited circumstances. Wewre like, ok, there's something here. And then it was like, ok, el, can we take, like, our particular domain? Can we take like chath and conversations, and can we see if there's teres ters, that marriage, there's that like match between the problem and the technology that's actually going to to give like, great customer experiences there? And so, like, our outside resolution. But we, we started out, we didn't use newnericks or anything for aversion one, but we had conviction that it was possible to build something good here. And then we, like, we went and we iterated that as we went went and built, like, minimal technology investment, actually validate that like, a kind of a crappy, knocked together prototype would actually help customers pepe actually wanted. And then like, empty rista. And then like, iteratedeid or iterated. Now we're unlike versian tree or russian forever technology. It uses like, very modern, fancying or a networks. It gets like, best in class performance and accuracy. But, you know, the first feshin was was, was like elastic search, sort of like off the shelf, and like just validat that this will actually help people in so, ye, sir, you're searching like bret, first true product space. But you want to like guide that search. You want to be like, yet i know that there's there's something good in this general direction of the product space. I'm not going to end up with, like, i validate an amazing demand for a product that i is impossible to deliver. You don't want to be there. You also don't want to be like, i have an amazing algritm that will deftly move the needle for something nobody cares about. So la, you got a like, iterat on, like, on both sides of that equation. And then find some kind of like landing zone in the middle. Her
Speaker 3
contes me thar for is actualy a third leg of a stool. There's the probem. I'm just frot my hand to on of three hand sotat is a problem. At is a solution. And then there's the story. There's what you can say about it. And one of the things that i've struggled with wi aan machine learninghere, we've struggle with it too, is what you feel good about saying externally and what other people are saying externally. And so i think that the worst of this is a tragedy of commons, where, like, all companies come out and say, you
Speaker 1
know, they make huge claims,
Speaker 3
and then the people who actually know what the talk man say, like, their ludicrous claims are not real. But there's this comperitive dilemma where you're like, well, f our competitor says 80 % and and there's no way we think they can get tat theyre making some like, you know, this isthe some sub claim or some astrices there. But ours is like 50. And so how do you think about that? How do you think about the claims you can make and the balance between the problem solution and story? I
Speaker 1
mean, so it's very difficult, i would separate out the internal product development from the like, success in the market, like internal product development. You know, i guess this is true an income amy is not true everywhere. But like, you know, if i come when i say, ha, guysi i'm pretty sure we can like give a good enough prodit experience, m i'm at least like accountable if it turns out that's not the case at all. And so, like, internally, you know, i think you got to like, work with people and explain things well. But at least the incentes are alined. Externally, when people are like, competion in the market with like, mal products, like, i's realy hard, like, you know, you can totally come across, like, we mentioned this area. I come across like, products in the market, and like, i assess their claims, like tisclaiming to do something amazing, and like, doesnt actually, like, how do you evaluate that? And it's very difficult to do so without putting them together in ta head ahead. And i think, you know, thit's just a hard problem of certain machine learning products, or ai products, where people are trying to like make claims based on numbers andd it cuts through the entire industry. Like, even if i see a new research paper promising something amazing, and it's got like examples of, you know, this is what the m we said this to the a i, and this is what it said back, my first question is always like, well, was that a cherry picked example? Does it do not, like, nine times out of ten, or like one time out of ten? Because it's very different depending on each and so, like, it is always dis simpliced, like, well, what's the performance? Actually, you can't really tell unless you do some sort of head hawd you sit down and play with it. And i think, you know, this is just, this is something that's that's hard about the space. In terms of like, intercomr our customers are doing more like head to heads and like proof of concepts, an valuations. I loaf tot like, that's wonderful. Thats what we want to see a in terms of the space in general, you know, i think you've seen people make demos publicly available more and more. Like you're seen things like people go like dolly too, or whatever, like they get to get access to like, independent researchers, artheear and like, or they do stuff like in the papers, or saying, this is what it produces in one run on a standard promptin, no, that helps people to getther heter ando son s things like dolly too are like, extremely visual. Like, you can look at thate like, hey, this is like an image whyle. I can ad a glance at my cognison, visual higar that i've evolved through millions of years as a human, i can like, process a lot of information, visuall herand be like, yet, i tis this, this things doing something cool.