Exa: Organizing the World’s Knowledge

1. 0:00 – 1:15

   Intro & Exa’s $85M Series B raise

   1. NDNicolas Dessaigne0:00

      [upbeat music] So today, I'm very lucky to be joined by Will Bryk, the co-founder and CEO of Exa.ai. Exa just announced an incredible round, Series B of $85M at $700M valuation. That's crazy. Congrats.

   2. WBWill Bryk0:19

      Thank you.

   3. NDNicolas Dessaigne0:19

      So to get started, can you tell us what problem you're solving?

   4. WBWill Bryk0:22

      I mean, the problem is that there's this new kid on the block, and it's called AI, and it needs to search the web. And so traditional search engines like Google and Bing, they weren't built for this world, they were built for humans. And so Exa is actually a search engine built for AIs.

   5. NDNicolas Dessaigne0:34

      What's different from a Google or normal search engine when you are targeting AI instead of humans?

   6. WBWill Bryk0:40

      Yeah. If you think about a human versus AI, they are not the same. So, like, a human is like this lazy creature that types in a few keywords, gets a list of links, then clicks on the links, like, likes UI, all these things. An AI is like a super crazy, like, can at-- like, you know, can type out a whole paragraph explaining what it needs, like, can scan, like, 1,000 results and only wants the highest quality knowledge. And so if you're optimizing a search engine for this creature as opposed to this one, you're gonna end up making all sorts of design choices. So for example, you need to be able to support more complex queries, and AI is not lazy and it c- it can make those queries. And you need to be able to support, like, a ton of results returned to the search engine. You also need to, uh, support, like, all sorts of customizations that

2. 1:15 – 3:10

   What Exa is building: a search engine for AI

   1. WBWill Bryk1:15

      customers that own the AI wanna make. So it's just a whole different world, really.

   2. NDNicolas Dessaigne1:18

      So at the end of the day, like, these AI, they work for a human.

   3. WBWill Bryk1:22

      Yeah.

   4. NDNicolas Dessaigne1:22

      So how does human interact with the end result of that actions?

   5. WBWill Bryk1:26

      Yeah. 'Cause I, I mean, ultimately, everything is for the service-

   6. NDNicolas Dessaigne1:28

      Yeah

   7. WBWill Bryk1:28

      ... of humanity. We're not just trying to make AIs happy. So I mean, basically we're, we serve companies. There's another, there's another difference. Like, we're not a consumer search engine like Google or Bing.

   8. NDNicolas Dessaigne1:35

      Yeah.

   9. WBWill Bryk1:35

      We serve, you know, companies, enterprises, startups. Uh, and they, you know, have some sort of product or application that it has AI inside it, and whenever that, uh, application or product needs to search the web, it'll use Exa. So, like, Exa is powering undernea- under the hood a lot of these systems, these AI products that a lot of us use e-every day.

   10. NDNicolas Dessaigne1:52

       So if I understand correctly, I mean, when I look at the market, I've not seen anyone else trying to do what you do today.

   11. WBWill Bryk1:57

       Yeah.

   12. NDNicolas Dessaigne1:57

       And people are relying on, uh, Google or maybe Perplexity, but there is no other product. Like, how do you explain that?

   13. WBWill Bryk2:03

       Okay. So there are a couple reasons why you don't see other companies doing exactly what we're doing, and it's because, one, it's extremely hard. So, like, building a search engine from scratch, like, people have been scared about doing that since Google existed. Uh, and, like, it takes years of research and development. So you kind of need a, a team that's crazy enough to do it. But then also a very unique thing is our business model, which is, like, we're not trying to be a platform that humans go to. We're trying to be search infrastructure that works under the hood in all these applications. And that's just, like, a whole different, like, idea that no one really thought about, and we just, like, we're fine doing that. Like, we are very happy serving the world and being search infrastructure. Like, no one needs to know there's Exa underneath the hood. And that just allows us to build all these things that, like, companies like Google and Bing just won't.

   14. NDNicolas Dessaigne2:40

       And I think you have two products, right?

   15. WBWill Bryk2:41

       Mm-hmm.

   16. NDNicolas Dessaigne2:41

       An API-

   17. WBWill Bryk2:42

       Yeah

   18. NDNicolas Dessaigne2:42

       ... and Website?

   19. WBWill Bryk2:43

       Yeah.

   20. NDNicolas Dessaigne2:43

       What's the difference?

   21. WBWill Bryk2:44

       This is very akin to, uh, OpenAI, where they have, you know, their fa-- well, they used to have their GPT-4o. [chuckles] And then they would have their o3. Uh, they have their fast GPT-4o, and their, their, like, slow-thinking o3. And so we, we also, um, we have, you know, our fast search, uh, that is synchronous and returns very fast. Uh, and then we have our, like, slow one called Websites, and it could take a minute or even 10 minutes or some cases a day, uh, to get you the highest quality results. And I think the point here is that, uh, when AI systems are searching

3. 3:10 – 5:05

   Why Google never finished its mission

   1. WBWill Bryk3:10

      the web, they need a whole diverse spectrum of, of latency profiles, and so Exa's building that.

   2. NDNicolas Dessaigne3:15

      So you started Exa, uh, before ChatGPT was even released, right?

   3. WBWill Bryk3:18

      Yeah. Yeah.

   4. NDNicolas Dessaigne3:19

      Uh, at that time, if I remember correctly, you were even trying to compete with Google-

   5. WBWill Bryk3:22

      Yeah

   6. NDNicolas Dessaigne3:23

      ... to do a web search.

   7. WBWill Bryk3:23

      Yeah.

   8. NDNicolas Dessaigne3:24

      And, like, first, maybe why did you try to do that? And then second, like, what did change when ChatGPT was released?

   9. WBWill Bryk3:31

      Yeah. Yeah. Sure. So we started Exa four years ago. This is way before ChatGPT came out. So this whole idea of search for AIs, it wasn't really-- uh, we, it wasn't on our radar.

   10. NDNicolas Dessaigne3:39

       Yeah. There was no AI-

   11. WBWill Bryk3:39

       Yeah. There was no AI to build-

   12. NDNicolas Dessaigne3:40

       Needs to search yet

   13. WBWill Bryk3:40

       ... a search for.

   14. NDNicolas Dessaigne3:41

       Yeah.

   15. WBWill Bryk3:41

       But what we wanted to do was build a, a better search engine than Google because at the time, GPT-3 had recently come out. This is summer 2021, and it was magical. You know, it could, like, understand a whole paragraph of text at, like, a deep level. And at the same time, there was Google, which felt like it hadn't changed in a decade. So the idea for Exa was like, what if we could build a search engine that could understand you on a deep level, uh, and give you exactly what you wanted? And we kind of built it for ourselves, like we wa-- for our s- humans.

   16. NDNicolas Dessaigne4:03

       Yeah.

   17. WBWill Bryk4:03

       We built it for humans.

   18. NDNicolas Dessaigne4:04

       [chuckles]

   19. WBWill Bryk4:04

       Uh, but not-- We were like, you know, a nerdy type of human, and the thing about nerds is, like, they want the highest quality knowledge. And so we were trying to build a search engine that, like, didn't give you SEO, it didn't give you ads. It just gave you super high-quality knowledge, exactly what you asked for. And it turned out, uh, when ChatGPT came out and we started getting all these requests for, like, API access for these AI systems to use search, the thing we had built was perfect for these AIs because, you know, nerds and AIs are actually very similar. They both want the highest quality knowledge. So we were-- The whole time, we were actually building the ideal search engine for AI systems, uh, we just didn't know it.

   20. NDNicolas Dessaigne4:33

       [laughs] And, uh, how fast after, uh, ChatGPT was released-

   21. WBWill Bryk4:38

       Yeah

   22. NDNicolas Dessaigne4:38

       ... did you change direction?

   23. WBWill Bryk4:39

       Pretty fast. So we, we had this search engine that we had-- we launched on Twitter. It got a lot of excitement. Like, Andrej Karpathy retweeted it. It was pretty cool. Then ChatGPT came out two weeks later, uh, and that stole a lot of the thunder-

   24. NDNicolas Dessaigne4:48

       Yeah. [chuckles]

   25. WBWill Bryk4:48

       ... in, on the Twitter world. And then we pretty quickly started getting requests for API access 'cause people started-- I mean, the, like, the actual acc- uh, access to, like, uh, ChatGPT-level tech, like, came a little bit later, like a few months later. But then we started getting, uh, API requests, and at first, we kind of just, like, dismissed them. We were like, "No, no, no. Like,

4. 5:05 – 8:00

   Early pivot moments and lessons from YC

   1. WBWill Bryk5:05

      why would you even need an API access to a search engine?" Like, "No, we're building this, like, n-new type of technology." We didn't really think about it. And then, you know, I got requests from some-- One of the first ones was some, uh, VC, uh, in Germany, and that was interesting. And then, uh, my, my roommate was, like, building, like, all sorts of projects. And then at some point, like, we were like, "Oh, wait a second, th-there's something here." And, and there was actually this, this big moment when me and Jeff were chilling in his room, and, like, we were like, "Wait a second, what if we're a search engine for AIs?" And it was like-- And the second we said the, the, the expression, it all, like, clicked into place, and we were like, "Oh my God, that is the world that is coming," and it became very obvious. And this was, like, many years ago, so it-- still there weren't many AIs to serve, but it was very obvious where the world was going.

   2. NDNicolas Dessaigne5:43

      And so did you change, like, direction, like, the next day?

   3. WBWill Bryk5:46

      Yeah. Right-- Pretty much right away, we started, like, building an API, and, like, we were like, "Okay, let's give it to these customers." So we built an API. It, it was just a wrapper over our current search engine and then, like, with some, uh, pricing.

   4. NDNicolas Dessaigne5:56

      Was that contrarian at that time?

   5. WBWill Bryk5:57

      Yeah. I mean, no one-- Like, we were the first ones to say search engine for AI, and it sounded crazy at the time. Uh, it doesn't sound that crazy anymore.

   6. NDNicolas Dessaigne6:04

      Was there any other, uh, contrarian bets that you made in the early days?

   7. WBWill Bryk6:07

      Yeah. I mean, there were a lot. Uh, it's a very contrarian company, I would say, which I love. Like, I, I almost always do the thing that other people aren't doing. Like, I, I-- Once people start to do a lot of things, I wanna do something different. A, a big one was just when we set out-- So again, like four years ago, we set out to build a new type of search engine.We were like, "Okay, how are we gonna do that? Well, we're gonna-- You know, the bitter lesson applied to search. We're just gonna figure out a way to, like, pour a ton of compute into a search engine. How do we do that? Well, we need to get a big GPU cluster." Uh, so we were like, "Okay, let's raise a seed after YC and spend half of it on a GPU cluster," which was, like, insane.

   8. NDNicolas Dessaigne6:39

      I can, I can remember that. [laughs]

   9. WBWill Bryk6:39

      Yeah. Like, we, we had that conversation. It was, like, an insane thing to do, but, but from first principles, it made sense 'cause we were like, "We need to develop new research in order to make a new type of search engine, so it makes sense to spend most of our capital on that." And yeah, I mean, in some ways, we didn't follow a lot of YC advice, but we didn't talk to users. You know, we, uh, we just, like, you know, focused for, like, a year and a half, uh, building, uh, a new type of search engine, which, which actually, like, turned out pretty well. Yeah.

   10. NDNicolas Dessaigne7:04

       YC's next batch is now taking applications. Got a startup in you? Apply at ycombinator.com/apply. It's never too early, and filling out the app will level up your idea. Okay, back to the video. Let's speak maybe a bit about the product and some of your choices.

   11. WBWill Bryk7:20

       Yeah.

   12. NDNicolas Dessaigne7:20

       I mean, you decided to crawl the web yourself, right?

   13. WBWill Bryk7:23

       Yeah.

   14. NDNicolas Dessaigne7:23

       Why not rely on an existing search engine?

   15. WBWill Bryk7:25

       Yeah. I mean, there's definitely partly an i- the identity thing here where it's like, "No, we're building a search engine from scratch." But also it, it turned-- it, it was the correct, uh, business decision because by owning the full stack, we have full control over the technology and can, like, customize it for customers in all sorts of ways, and customers really care about you having your own search engine for various reasons. One is customization. Like, "Hey, I only wanna search over a thousand domains." Like, there are a lot of enterprise customers who want that, and, like, if you're wrapping Google, you can't filter to a thousand domains. It doesn't let you. Or, um, you know, if you want to get a thousand results instead of a hundred, like, you, you only can do that if you own your own search technology.

5. 8:00 – 11:20

   The shift from developer tool to AI infrastructure

   1. WBWill Bryk8:00

      Uh, then there are other things like zero data retention. Like, if you want true zero data retention for, like, a financial customer, which basically all of them need, you need to have your own, uh, independent search engine. So, and that's just, uh, three. Like, there are, like, tons of reasons.

   2. NDNicolas Dessaigne8:10

      Mm-hmm.

   3. WBWill Bryk8:10

      And it, it turned out to be totally the right call, uh, to build our own search engine, just control our own destiny.

   4. NDNicolas Dessaigne8:15

      S-s-speaking of, like, crazy bet here, you also decided, like, very early on to build your own model.

   5. WBWill Bryk8:20

      Yeah.

   6. NDNicolas Dessaigne8:20

      Like, like, there were, like, already some open source models. Why not just pick an open source model and fine-tune it?

   7. WBWill Bryk8:25

      At the time, the open source models were pretty bad. They're still bad. I think, um, the open source commun- like, uh, models for, uh, for like, like language, like, like language modeling, like generating text are pretty good, but in terms of, like, open source search engine models, they're, they're, they're very bad because, like, you take the best off-the-shelf embedding model, and you, uh, you try to make a search engine out of that, it'll be really bad 'cause these models are meant to search over, like, hundreds of thousands of documents, like, millions of documents, not, like, hundred billion web documents that are very chaotic a-and, uh, there, there are all sorts of reasons why, like, off-the-shelf models don't really work well for search.

   8. NDNicolas Dessaigne8:55

      And so I guess you are, like, constantly training this model. Like, what does your infra looks like?

   9. WBWill Bryk9:00

      We're constantly trying to think of ways we could pour more compute into the model to make it better. Uh, and so that's why we have, like, like, you know, at first we have to-

   10. NDNicolas Dessaigne9:06

       Do you still have your own hardware? Uh [laughs]

   11. WBWill Bryk9:07

       We do have our own hard- So we, we've, we've leveled up to, like, uh, from our original cluster, now we have, like, a $5 million GPU cluster, so 144 H200s. We call it the Exa cluster 'cause Exa means 18-- uh, 10 to the 18th, and, uh, and it's 18 nodes. But now we're gonna get, y-y-you know, even way more compute with this Series B. We're basically trying to, uh, teach this model about the entire world's knowledge.

   12. NDNicolas Dessaigne9:28

       How is that different from, like, an LLM, a traditional LLM? Like-

   13. WBWill Bryk9:31

       Oh, uh, well, an LLM, uh, is trained on the world's data, but then it, it compresses it into, like, a, a set of weights, uh, and, like, the problem is, like, it can't memorize the whole web.

   14. NDNicolas Dessaigne9:39

       Okay.

   15. WBWill Bryk9:39

       So, like, even if you just think about it from an information theory argument, it's like the web is, like, you know, a huge amount of petabytes and then, like, the GB4 size model is, like, a few terabytes. Like, you literally can't fit the whole web in-into the weights of this model. So, a-and that's why, like, an LLM will know when Albert Einstein was born, but it doesn't know when I was born, uh, because it just hasn't memorized everything. And, like, the knowledge on the edge is, is the most knowl-kn-uh, valuable knowledge. Uh, so, so both, like, LLMs can't memorize the whole web, but also the web is constantly changing. So your-- LLMs are always gonna have to rely on search.

   16. NDNicolas Dessaigne10:08

       So how do you make sure the results are actually good? Like, how do we evaluate-

   17. WBWill Bryk10:12

       Yeah

   18. NDNicolas Dessaigne10:12

       ... uh, the results?

   19. WBWill Bryk10:13

       Like with just normal AI research, like, uh, evals are extremely hard.

   20. NDNicolas Dessaigne10:17

       Mm-hmm.

   21. WBWill Bryk10:18

       And, uh, it's hard to get, like, one eval that captures everything you want.

   22. NDNicolas Dessaigne10:22

       Is there a benchmark out there that works for what you're doing?

   23. WBWill Bryk10:25

       No. So, uh, there's really none. So, like, uh, that, that even makes it harder. So not, like, not only is there not-- you can't make one eval, but y-and you, you always have to have, like, a diversity of evals, but then also, like, there, there's no standardized evals around search engines, and so we're basically making those all in-house. And we wanna be thought leaders here, like we wanna publish. We, like, we wanna publish papers about, like, our evals, how we run them, so that other people can contribute and, like, really start that movement to have evals around search engine 'cause it's so important, uh, but it just doesn't exist. So it's been hard to evaluate the model, and, uh, there's a lot of secret sauce that goes into the evals themselves.

   24. NDNicolas Dessaigne10:55

       But you're still going to, uh, publish some benchmark others can-

   25. WBWill Bryk10:58

       Yeah

   26. NDNicolas Dessaigne10:58

       ... compare themselves to?

   27. WBWill Bryk10:59

       Yeah, yeah. That's our plan.

   28. NDNicolas Dessaigne10:59

       Okay, looking forward to that. Okay, awesome. Looks like your customers, your users are AI agents now, right?

   29. WBWill Bryk11:05

       Yeah.

   30. NDNicolas Dessaigne11:05

       Mostly. Like, at least it's changing. How do you see that evolving? Do you think that, uh, a few years from now it's going to be mostly agents?
6. 11:20 – 14:05

   How AI agents use Exa behind the scenes

   1. WBWill Bryk11:20

      building full, full-out agents. And I think, like, o-over time, everyone will move towards agents 'cause they're just, like, smarter, better. Uh, they have different properties, where, like, an agent can, um, take a longer time. Agents make more sense in, uh, products or workloads that are asynchronous. So I think, like, over time, yes, everyone will move towards agentic systems. But I mean, an agent is really just, like, heav-heavy calls of LLMs, so it's not that different.

   2. NDNicolas Dessaigne11:40

      Does that, uh, change the product in any way? Like, how you're going to build the product?

   3. WBWill Bryk11:45

      Yeah. I mean, you could imagine what the world looks like with agents. Like, I, uh-- One thing we always do at Exa is, like, imagine what the world looks like and then build for that world. Uh, because if you're building for the current world, like, the, like, AI changes so fast that, like, you're gonna be outdated within a year. So, like, what will the world look like? Well, yes, there will be agents doing all sorts of-- taking all sorts of actions, and for example, one thing they might do is do, like, 100 searches, uh, within one request. So you talk to your agent, and then it does 100, uh, searches in the background. Well, then what do you need? Well, you need those searches to be really fast because, like, if it goes from, like, a second to 100 milliseconds, suddenly you've just 10X reduced the, the latency of that agent.

   4. NDNicolas Dessaigne12:17

      The latency is kind of really important. Going to be a high priority.

   5. WBWill Bryk12:19

      Yeah.

   6. NDNicolas Dessaigne12:19

      Let's now talk maybe about the news you're announcing.

   7. WBWill Bryk12:21

      Yeah.

   8. NDNicolas Dessaigne12:22

      So that crazy Series B-

   9. WBWill Bryk12:23

      Yeah

   10. NDNicolas Dessaigne12:23

       ... raising $85M at $700M valuation.Why now? Why is that much?

   11. WBWill Bryk12:28

       Yeah. I mean, some would argue that that's not enough to build a, a search engine from scratch.

   12. NDNicolas Dessaigne12:32

       So you're doomed?

   13. WBWill Bryk12:33

       [laughs] Right. Uh, no, no, no. There, there's a lot you could do with $85 million. Uh, and we've been very, like, frugal and clever about how we spend the money. Uh, for example, you don't have to index the entire web. But yeah, I mean, we're going to basically scale up everything, so scale up, uh, our GPU cluster, uh, so we can do, like, way more research way faster. Um, we're gonna scale up, like, our crawling and processing of the web, uh, which... And the web is pretty big, so, like, uh, uh... It's not too big, actually. It's actually-- It, it's, it's manageable. We're getting to the point where it's, like, manageable. Um, and then of course hire the best people in the world.

   14. NDNicolas Dessaigne13:07

       Looking ahead, uh, what does search looks like, like, five years from now? Like, both for agents and for humans.

   15. WBWill Bryk13:14

       Search five years from now is kind of-- I think of it as, like, the search that we should have had when I was born.

   16. NDNicolas Dessaigne13:18

       Okay.

   17. WBWill Bryk13:18

       Like, you know, 'cause, like, we're a civilization that made it to the moon. Like, we should have full access to the information in our world. Like, there's so much information, there's so much knowledge out there, and it's actually so hard to f- to find it. And, you know, like, you know, and whenever you, like, stumble upon knowledge, you're like, "Oh, I wish I knew that before." Like, that experience should go away. And so, like, one way you can think about the world in five years is, like, any sort of request for information that you need should just immediately happen, and you have no, no blockers. So for example, like, recruiting is an information blocker right now. Like, you are a company that needs people to hire. You should know all the people you need to hire. Convincing them is another thing, but at least you should know about them. Uh, like, why, why would you ever not n- Like, you should know who are all the engineers that perfectly match your company. And so that's, like, a search problem that we wanna solve. Or for example, if you're trying to do sales, uh, like you wanna sell, you know, you're, you're trying to sell to a certain type of company, you should know about all the companies in the world.

7. 14:05 – 17:30

   Organizing the world’s knowledge “for real”

   1. WBWill Bryk14:05

      And if any new one comes out, you should immediately know about them. Like, that-- There's, like, this coordination problem in the world that we just accept as normal. It's not gonna be normal in five years. We're gonna look back and be like, "How on earth did they function without perfect information?"

   2. NDNicolas Dessaigne14:15

      Sounds like it's, uh, use cases you already are serving today, right?

   3. WBWill Bryk14:17

      Yeah. We're already serving those use cases today, and, like, they're very good and, you know, far better than a lot of stuff that exists out there, but, um, it could be perfect. I want it to be perfect.

   4. NDNicolas Dessaigne14:25

      And speaking of hiring-

   5. WBWill Bryk14:26

      Yeah

   6. NDNicolas Dessaigne14:27

      ... I, I guess that you'll hire even more now.

   7. WBWill Bryk14:29

      Oh, yeah, for sure. We're definitely hiring.

   8. NDNicolas Dessaigne14:30

      What, what are you looking for?

   9. WBWill Bryk14:31

      Yeah, we're looking across the board. Like, uh, really the fundamental thing is, like, people crazy enough to build the next generation of search. We typically hire not for experience, but for, like, extreme, like, intelligence and hunger, extremely good engineers who are excited to, like, scale up, uh, a search engine to, like, a trillion pages, researchers, or people who wanna do AI research. You don't even need that much experience. But if you, like, you wanna do experiments, you wanna create training data, you wanna create evals, like, X is a great place to, like, learn those things rapidly and then just, like, become an expert. I mean, also, like, just go-to-market people who wanna sell. Like, there's, like, a huge market for this.

   10. NDNicolas Dessaigne15:01

       [laughs]

   11. WBWill Bryk15:01

       Like, uh, there's, like, a huge opportunity, and so, like, also, like, change a lot of lives. Like, like, uh, there are a lot of people, there are millions of people using us now.

   12. NDNicolas Dessaigne15:08

       You had that, uh, fun hack, like, uh, these maths puzzle posters.

   13. WBWill Bryk15:11

       Oh, yeah, yeah.

   14. NDNicolas Dessaigne15:12

       Can you tell us what you did there?

   15. WBWill Bryk15:13

       We try to do creative things. So, like, uh, in this case, it was, uh... We were like, "Okay, how can we, like, hire people better?" I had, uh... Maybe we could do something fun, like, in the physical world, 'cause no one does that. And we were like, "Oh, wait, what if we just put, like, posters around the city? Okay, what would be on the posters? Why don't we make it, like, a puzzle?" Because if an engineer sees a puzzle, they're gonna-- Like a moth to a flame, they're gonna wanna solve it. And then the puzzle will link to Exa, and that'll be all this cute thing that'll make them like Exa.

   16. NDNicolas Dessaigne15:34

       Did you hire anyone?

   17. WBWill Bryk15:35

       Yeah, yeah. We hired one person from it. So that was like, in terms of expected value, it was pretty high. Like, it, it took us, like, an hour to think of the idea, you know, uh, an hour and a half to think of the puzzle.

   18. NDNicolas Dessaigne15:43

       How many people applied?

   19. WBWill Bryk15:45

       Oh, like, 100 probably. Yeah.

   20. NDNicolas Dessaigne15:46

       Okay, 100.

   21. WBWill Bryk15:47

       Yeah, yeah.

   22. NDNicolas Dessaigne15:47

       And got one good hire-

   23. WBWill Bryk15:48

       Yeah, yeah

   24. NDNicolas Dessaigne15:48

       ... out of that. Oh, that's, that's worth it. Let me conclude maybe with a couple of questions about, like, more your entrepreneur, like-

   25. WBWill Bryk15:53

       Yeah

   26. NDNicolas Dessaigne15:53

       ... journey.

   27. WBWill Bryk15:54

       Yeah.

   28. NDNicolas Dessaigne15:54

       What would be one thing, uh, you'd love to have known when you started?

   29. WBWill Bryk15:58

       If, if I had known how constant the challenges are. Okay, so every day I come to the office, and there are four really hard problems, like, four fires and four amazing things happening, and that just never changes.

   30. NDNicolas Dessaigne16:09

       Right.
8. 17:30 – 21:15

   Competing in a post-Google search world

   1. WBWill Bryk17:30

      you know, 57. And then I don't know what the final boss is, but it's probably crazy.

   2. NDNicolas Dessaigne17:33

      So was there any, uh, big challenge you faced along the way?

   3. WBWill Bryk17:36

      I mean, yeah. There have been tons. I think one interesting one is, uh, you know, right around the time ChatGPT came out and we were building our search engine, there were rumors that Bing was, like, combining LLMs with search. And we were like, "Oh no. Is that gonna, like, solve search?" And, you know, we were worried about Bing Chat. And then when it came out, like, n- n- it didn't matter at all. And I think the, the learning there was, you know, don't worry too much about competitors, uh, because, like, you know, they're building whatever they're building, and often it fails, and they have all their problems. And, like, if you focus on what you're doing and just, like... If, if it's a big enough market, you're gonna dominate in whatever you're doing. So, like, you know, o- over the years, like, there have been so many, like, "Oh, no. Is, like, Google gonna do this? Is, is Bing gonna do this? Is OpenAI gonna do this?" And it's like, it doesn't matter.

   4. NDNicolas Dessaigne18:13

      So you don't really care about competitors anymore?

   5. WBWill Bryk18:15

      I mean, I think about them a- and it's important for, like, marketing and strategy, but what matters more is, uh, team execution and, like, speed and velocity. Yeah.

   6. NDNicolas Dessaigne18:23

      Thanks, uh, thanks so much, Will, for joining us today.

   7. WBWill Bryk18:25

      Thank you.

   8. NDNicolas Dessaigne18:25

      It was great to learn about your journey with Exa.

   9. WBWill Bryk18:27

      Yeah.

   10. NDNicolas Dessaigne18:27

       And, uh, can't wait to see, uh, what's next for you. [outro music]

Episode duration: 18:39