Thank you so much for coming and joining us. It's a pleasure to have you here.
Thank you for having me. I'm really excited to be here.
Could you just, for the people who don't know you, just introduce yourself and tell us a little bit about Cognition?
Yeah, so I'm Scott. I'm the founder and CEO of a company called Cognition AI. We build Devin, the AI software engineer. And so we work with teams all over the world, you know, big companies, startups, everything in between, to figure out how to build software much faster.
It's amazing. I think you got super viral in the early days with the first demo, which was this vision of AI really driving software and coding in a way that felt impossible today. Now, this is part of everyone's day-to-day, right? Everyone would just like have five different agents running right now, and it felt like a dream two years ago. It felt impossible.
Yeah, I know what you mean. It's kind of funny. I mean, we definitely had a lot of periods in the middle. I mean, because when we started, it was kind of outrageous, you know? And so there were definitely a lot of times where it's like, 'Oh, you could feel like it was almost there, but it wasn't quite working.' And then there were all these questions, 'Well, does this make sense or does it not?' But like you said, I mean, I think this year especially, with all the capabilities in 2026, I think we've really seen the background agent take off. Yeah, and now there's like consumer agents that people use that are kind of this form factor. There's agents for all kinds of different use cases, right? So it's cool.
Yeah, that's incredible. And like looking back, clearly you had this vision that it eventually would be possible, but you went through this amazing journey of the growth of Cognition and the technology is growing on top of it. What were the things that went as expected and what were the big surprises that you had the past two years?
Yeah, no, thank you. Well, first of all, thank you for the kind words. I mean, it's kind of funny because I think that the technology itself getting better — our whole bet as a company was that it would get much better than people thought. I think it still was surprising, you know, how stark it was and how much better it's gotten obviously over the last two years. One of the things that I think that we had to learn — like a lot of stuff. I mean, I think, you know, working with you in the early days actually, I think we learned a lot of these things, which was, you know, when we started, it was much more of like a demo or like a prototype, you know, and it was cool. You could do some fun kind of hobby tasks, but working in the real world on an actual code base, I mean, it's very different work, you know, and so learning all these details about what is it actually like to deploy at a real company, what does it actually look like to, you know, work on a big code base and actual real use cases, you know, working with all the integrations. I think a lot of things there. And then definitely a lot of things where over the last two years or so, we've had to kind of like speed run everything. And so like I guess some funny examples — honestly they're a bit embarrassing, but I was like, you know, we didn't have any designers in-house until like eight or nine months ago.
That was definitely — yeah, we should have done that way earlier, you know. Now the product — like, the product was there were a lot of cool things about it, but it definitely did not look good until much more recently. Same thing, you know, marketing — even like starting to ramp up much more with marketing. Like I think all these things, I mean, it's funny because it's only been like two or three years, but basically we've kind of had to like speed run building all these functions.
Yeah, it's fascinating when you talk about being able to deploy at scale where you're deploying in the biggest institutions in the world. So, I imagine that also requires you getting to a level of sophistication very fast, and maybe you have a lot of time for the design side, but you have a lot of time for the diligence, the rigor, how robust the deployment is, which is pretty cool.
Yeah. Yeah, I know. It's like — I mean, I think it is true, honestly, and we talk about this all the time, that, you know, Silicon Valley — I think people think so much about the raw intelligence and the capabilities, but of course, I mean, there's just so much more than just that in terms of how do you get it out into the real world and get to a point where everybody's using it for all their different day-to-day activities. And so, yeah.
That's super cool. And one of the things that really struck me, you know, when we met, you know, the early days, is that you were creating an AI agent for coding, but every time you talked about the future, it was never about replacing software engineers. It was always about this additive view of reshaping and designing the work of the future and the union of the tech plus the human being the superpower. And you see that in the products, how the messaging, the products, but I also see that in the forward deployment engineers in the enterprise with Cognition. At Databricks we also see the similar aspect of like the union of human plus AI being that kind of holy grail.
How do you think about this today and how do you think that evolves in the future?
Cool. Yeah, so I think, you know, it's funny because I'm very bullish on the AI capabilities. You know, when people ask me about how smart — I mean, it's like, and now, you know, everyone's talking about all these math problems that it's solving. I've always thought it's, you know, all of these logical things, actual decision-making things — like AI will get there and it will do it. I think that's actually different, to your point. I think that's different from saying that AI should rule the world and we should all have AI overlords, which I think hopefully we agree that that's not really what we want. And if you think about like what does that end up meaning, and then I think, you know, a lot of what it looks like is kind of like, you know, it's up to the humans to decide, okay, what do we want to build? What do we want to do? What are the high-level kind of decisions that we want to make? What is it that we want to create in the world? And the AI can help you with all of the execution and so on, but it should still be up to you what to do, right? I mean, otherwise, you're not even — you know, like the humans aren't in control anymore, right? And that's how we've always thought about it. And I think, yeah, like you said, I mean, I think in practice, you know, you can on the one hand say that the capabilities are at a certain place or whatever, but what does it mean for humans to actually like have the capabilities at their fingertips? The reality is, of course, you know, you still have to build really thoughtful product experiences, you have to do the right education, you have to, you know, onboard people onto using these agents and so on. And I'm sure it's a lot of the stuff that you're thinking about and, you know, and building, too. And so, yeah, I think it's — I mean, I think we'll always have that, you know.
Yeah, and I think that there's an element around the human is responsible, is accountable in a degree that no software system can ever be, right? So which I think adds a layer of trust that I think is super important.
For sure. I think it's kind of like the AI will get better and better, but, you know, there should always be some human-AI communication layer almost, right? And so then all of the questions in between of like, what is the information that you show to the human? How does the human kind of work and give their thoughts to the AI? How do you iterate together? Like there's all these products that you have to build in that space. And I think that's, yeah.
That's awesome. That's excellent. We first met two years ago. I think it was your first international trip after founding Cognition. Came here in Brazil. Then I think last year you had your first developer conference also in Brazil. And you're just announcing that you're opening offices in Brazil. Why Brazil? What keeps bringing you back to this country and how do you think about it?
Yeah, no, I mean, we're growing internationally in a few different places, but I think Brazil has always been a meaningful hub for us. You know, it's an interesting question. I mean, even back in the early days, obviously when we were working together, you know, I think there was a lot of that. And I think for sure, you know, a lot of these folks like Nubank or Itaú or others, you know, have been big early partners of us. I think one of the big reasons too is that I think Brazil, you know, is a country where there is so much demand for software, there's so much to build and so on, but obviously there's not enough software engineers, right? And, you know, how do you solve that problem? Of course you do that with having each individual engineer become much more effective with AI. And so I think that's one of the reasons that we've seen folks really adopt and kind of grow that quickly.
