The Future of Software & AI | Cognition’s Scott Wu

I've just always been this way. As a kid, I just hated losing. Like my first competitive memory ever is like when I was in second grade. I went to this seventh grade math competition. It was like a middle school competition that was held at the local university or whatever for middle schoolers.

Yeah. Yeah. I was like seven or eight years old. I was competing in the middle school math and like I did the math test and whatever and then they were calling out the names of the like here's who got third place, here's who got second and I was kind of like waiting for my name to get called and then I was none of them and I just remember being so pissed about that.

Yeah. I can't really give you a rational explanation for why it is.

I mean it's all I do honestly. I don't know. I think the like

Well, I think strategy game I don't know. It's the way even building a company it feels the same. It's just like you're calculating the moves. You're thinking about okay, if you do this and then this happens and then you do that and here are the different moves and you're like calculating out what comes out to success. You know, it's like a tree search, you know, where you're exploring the different options in the decision tree and you're trying to figure out how to lead to victory. Like that's like the only thing I do in my life.

I think that's right. Yeah. Yeah. I was a little brother growing up and so my older brother was four or five years older than me and naturally we'd play video games and similarly I would just always be super salty there as well. I don't know. It's just like yeah it's just always like that.

Yeah.

Yeah. Well basically everything. So obviously the main thing was math and programming competitions and so ever since I was really young that was like my life you know my whole goal was to become like world champion of competitive programming. I would do that all the time as a kid. I would do, you know, the really great thing about these competitions, too, is you know, you compete in your school competition and if you do well enough in that, then you make it qualify for the local or like the city competition and then if you do well in that, then you get to like the regional competition and the state competition and then you get to go to the national thing, the international thing, right? And so it's like a very nice setup where sooner or later you get to kind of meet people who are like you. Basically, when I was a kid, doing these competitions, going for that, it was like all I really cared about. Those people that I met through these national international competitions were honestly more like they were my childhood friends more than like the people around me in Baton Rouge, Louisiana were. And it was like we would hang out online, we would talk about math, we talk about problems, but all the other things too. I mean, I played basically all the different competitive games. So, like I played a lot of Super Smash Brothers. I used to go to tournaments for Super Smash Brothers. It was a lot of fun. Played Melee. And then I played like Tetris. I played a lot of poker. I played some chess. I was okay at chess. I was not good. I played some Go. My dad was a competitive Go player. My parents came to the US in some sense because of Go, which was kind of a funny coincidence because my dad was in grad school in China and he had a professor who like really liked him. The reason he liked him is because my dad was like a really good Go player. Like he was like a seven dan at Go, which if you were to call it in chess would be like I don't know 2,300 or 2,400 rating equivalent or something like that. He would play with this professor, you know, on the weekends and stuff and they like, you know, my dad would generally win and they would like talk about the games and stuff. And then that professor ended up moving to the US to come and teach. And at the time, you know, this was super early on in, you know, immigration from China to the US. And so it was not a very like it wasn't really a path that people knew that you could take. The professor wrote my dad and said, "Hey, like I came. It's great. Like there's so much more opportunity. It's so much better. Like you should obviously come as well. Like I'll help you with your visa application. I'll help you like apply to colleges here and everything." And so my dad applied to grad school in the US and that's kind of how we ended up here in the first place. I was born after we moved to the US obviously.

Yeah. No, she was always she was definitely the most salty I would say for sure. I mean she would

Salty just means that you take offense to the idea of losing.

Yeah, she would always be, "Oh, no, no, I'm better at this or I'm, you know, I can beat you at this, you know." And I don't think she I mean, she played ping pong a bunch growing up. Actually, she played on her school ping-pong team. Obviously, she studied some amount of math and so on. But it was just it was more her personality than any one thing that she really put all of her competitive energy into.

Yeah.

I think they would have been happy enough if I just got like a more traditional cushy job and did all of that. Like I don't know that they specifically steered me towards entrepreneurship and being a founder. But no, they were always

Yeah, I think she always told me that I was the best and she was always extremely proud of that, you know, it's the I had these like when we would go to these math competition.

Yeah, I think so. I think that's right. I think even when I was like tiny, she would tell me that I was extremely talented, extreme, you know, she was always a huge source of support for me and she always obviously believed in whatever I wanted to do, you know, because you would compete in math competition, you would get these trophies and stuff and like we didn't have like, you know, growing up like we didn't have like pictures of our parents like on the walls, you know, we just had old math competition trophies. It was like like my mom was very intentional about like no no the thing that we value in this household

Ours ones that me and my brother want of course.

You know, so like you know when I was pretty and my brother as well, you know, we both really like these competitions and so like you know like accumulate these trophies and she would always like every time we got one she would hang it up and we're like put it up on the mantel piece and everything and no it's like it was always I think what she really valued. I think education was really important to her and I think being the best was very important to her as well.

So, I think that's definitely true in terms of how it feels in order to get anywhere. You got to lose a lot. If anything, almost like a lot of the, you know, I mean, a lot of the guys you're talking about, if anything, like the way they got to where they are is by losing a lot and having their share of wins along with that, but like you just have to put yourself out there and do a lot. So, it's kind of an interesting thing to your point of like I definitely feel the same way that yeah, like losing feels way worse than winning feels good, but not by enough that it makes me want to stop trying if that makes sense.

So, again, you know, we're hyper competitive, but also, you know, the other thing about us is like we all had kind of started our own companies before this. So our founding team was a pretty big founding team is nine people and most of us had already founded our own companies before we had done different things. It was true for a lot of the early team and we've always thought about this as like this is the big one and so like we want to go for all you know we want to be a generational business like we want to build a hyperscaler and we want to go and do that and like maybe we'll succeed maybe we won't I don't know but like that's what we're going after and I think to me what that means in our field in building software is like you know people sometimes use the term like a coding agent or like you know like AI programming or something I always like you know I always hear that I think a little bit about like well we're not always I mean we're not going to be interacting with code for that much longer you know our programs might not be the right like level of abstraction but I think what is always true probably is that it will be the human computer interface and so what I mean by that is like the way that we think about Devin if we're successful is Devin is the way that humans can tell their computers what to do because that was the whole point of software engineering anyway right is just to be able to work with your computer and tell it what you want it to do I mean doing that for the world is like a massive opportunity and that's what we're really excited to go after.

Yeah. I mean, the simple answer is, you know, we're a bunch of nerds who we're all programmers and built software and so on before and now this is the idea of teaching AI how to go do that.

I think it is incredible. Yeah. I mean the Well, I would go further too and just say like at a zoomed out level, it's I honestly think that because you know everyone talks about AGI everyone talks about well what is the future going to be like what are human lives going to be you know once AI can do everything and I mean to some extent I think the thing that is most human obviously is self-expression creativity like having things that you want to make happen in the world and being able to go do those things right and so I mean a lot of how I think about this is basically we want to build the tool that gives everybody the power to go and make things that they want to make in the world. My co-founder has a saying which I've always loved. Which is, you know, we've been spending all this time living in survival mode as a species, you know, and now we're going to be living in creative mode. And I think that's right. Like I think you know, Minecraft survival mode is where you're like you know, you're growing food, you're like making sure you're safe from the monsters at night and whatever. And like creative mode is just like everything's up to you. You know, you have all the resources at your disposal. If you want something to happen, it'll happen. And the only question for you is like what you want to make happen. I think it's going to be amazing. And I mean, I think that is like the world that we're going towards and like we want to be the ones building that.

Sure. Yeah. So, Devin is today, you know, what folks know as an AI software engineer. And basically what that means is that Devin is a tool that anyone can use that will work with them end to end on building out software. And so we work with a lot of the biggest companies across the world, you know, we work with Goldman Sachs, we work with Mercedes, we work with, you know, a lot of areas of the US government at this point and so on. And we work with their software teams to just help them build more and do much more. And you know, the thing about it is in the last 20 30 years obviously I mean software is eating the world is the famous line. I think it's very much true. It's still like it's still got a couple order of magnitudes to go and in practice what it looks like is that teams use Devin to ship 10 times faster and to do 10 times more.

So now we're going to get into a philosophical discussion of what it means to be a programmer or a software engineer, right? And I think like you know if you go all the way back it's you know there was a time where programming was like using the vacuum tubes like plugging all those in and having it do the machine do the arithmetic right like the ENIAC was you know when the first in some sense that the first computer out there although it was obviously very different or it would have been like you know filling out the punch cards and like setting up the or writing you know the like putting down assembly you know or writing in BASIC or something right so we've gone through a lot of generations already is my point and what does it look like you know going forward when you talk about programming all it really comes down to is like how do you tell your computer what to do and like every single piece of software that you use if you're using you know Instagram or TikTok or YouTube or whatever like that's a piece of software that somebody or like some you know in these cases some pretty big teams of engineers came together and thought through all these details of okay here's what I want it to do here's how I want it to look here's what I want this button to do here's how I want to architect it every single little decision obviously was made by somebody but the computer itself is then executing it accordingly to what the wishes of its creators was right and I think what we'll start to see is that abstraction will continue to climb right and like you know we kind of see this already like at this point you don't need to know what programming languages like you don't need you don't need to know Python or Java or something like that in order to build your own software right and you can just say hey here's what I want I want to make a cool website that does this this and that or for example you know in my existing product you know here's what we have today and I want to change this thing or add this new plan or add this new feature and just have the agent go and do that for you, right? I think we're going to continue to go further down that axis. One important kind of like distinction I'd make, which is, you know, I think what we'll see a lot more of in the near future, is software today, the math only really works out to create software if it's going to be used at least like a million times or something, you know, and I'm giving a maybe it's 10,000 times or whatever. And my point is like if you want to go build a product today, you need a whole team of engineers. Engineers are expensive. You got to pay salaries. You got to go build all this out and you got to go and do that, right? You need that software to be used enough times or to create enough value for that to be worth it, right? And you know, something like YouTube passes that test because obviously so many hours have been spent on building YouTube, but way more hours have been spent on using YouTube and that's what's made that work out and made it feasible, right? But there's so many things out there which only, you know, very specific things that only need to be used a few times or even like only need to be used once, right? And so like all the white collar work that we talk about today even is very all right, you know, wake up in the morning. All right, I'm going to go look through these like 15 LinkedIn profiles. I'm going to look for this and that or whatever. Or I'm going to go fill out these forms or I'm going to do this data analysis and put this Excel sheet together with this research that I found. Right? All of these things are things that could be done with software. It's just it obviously doesn't make sense to hire a whole team of people to go make you that piece of software which you're going to use one time and never again versus just having the human go and do that themselves. Right? I think what we're going to get to is we're going to get to a point where you are just giving your instructions to that agent and the agent on the back end, you know, you don't even have to look at this, but on the back end the agent is going to figure out, okay, here's I'm going to write this code that's going to go do this. I'm going to put a script that automates this part. I'm going to do this. I'm going to do this. And that's what's going to allow it to actually go and do all these things. But what you start to get to, you know, as we're kind of saying, it's like this is really just how you control your computer and how you do what you want to go do, right? And you wake up in the morning and it's like here's what I want to go do. You talk to your agent about it. You figure out the task together once it has it. It can do the part of the literal like, all right, put the pen to the paper on writing code, but that task is like, you know, you're the one that's deciding what to do.

Yeah. I mean we've made a lot of steps toward it. I would say we still got a ways to go. You can use Devin today. You can use all the different kind of coding tools today. And you can do a lot more than you could have done, you know, 10 years ago. But certainly or even one year ago or 6 months ago. But certainly you know it's not at the level that we're talking about of you are neural linked into the AI and you can tell it exactly what you want it to do and what you want to see in the world and just have the AI go and do that. When do we get there? It's hard to say but I honestly I mean I think we'll have solved most of that over the next 5 years or so. In AI terms 5 years is like a century you know in the rest of the world terms. Obviously it's like it's kind of crazy to imagine that things can change that much in 5 years but I really think it will.

So true. I mean you see this in the progress itself. You see this in the scaling laws with the data. You see this in the revenue curves of the companies that are building an AI including your own. And it's a very, you know, it's like humans aren't really wired for this, right? Like all of our inherent fight or flight response are our kind of like our ability to kind of like measure things to vastly oversimplify if you're just, you know, fighting out there, you know, foraging for food or whatever it is like, you know, a good hunt will bring you, you know, a couple days worth of food or something. But obviously with the kind of exponential curves that we deal with, you know, the equivalent of a good hunt here could be a thousand years worth of food and we don't have that intuitive signal in our brains to really understand that, right, at a really deep like native level. People often underestimate, I think, how fast things can change and how fast the world can change. I mean, my parents even like drilled this into me because they grew up in, you know, in communist China. And they came to the US and you know, even that like we're talking about what are what were ultimately even like much slower scales of progress in some sense relative to I think what we're seeing today. But even that was like, as you can imagine, it was incredibly jarring to them. You know, the idea that like you come to the US and everybody has a car and all of these different, you know, it's like all everyone has all these household appliances and has all, you know, it was a very different life when they grew up in like, you know, the '60s in China and they were much much poorer and it was very very different, right? And it's like I mean, funnily enough, people got used to all these things pretty quickly and now we can't live without them. But I think we'll kind of undergo the same period with AI where 5 years from now it's going to be insane to think about all these things that we're going to have. 10 years from now we're going to have forgotten that we ever lived without them. Honestly,

No, you're way too generous. I don't know that there's I don't know that I have anything that interesting or that deep of an insight. I mean, I think it's

So basically an AI can just take a task that would have taken humans hours of work to go do. If you go and describe that task well enough to the AI, it will just go and do the whole thing and come back to you with the result and then you give it the next thing the next thing, right? And if you just ask from a first principles question, well, why can't that be days or why can't that be weeks or months of work? And then what does the world look like if everybody has an agent that can just do months of work for them at a time? Then you get to a pretty different conclusion from what we've all seen and what we've all lived for the last several years. And I think that that kind of first principles thinking is as different as it sounds and as crazy as it sounds. You know, this is one of those times where it's actually more correct than the simple pattern match if that.

Yeah. Yeah. And I think that's true and I think we will get there. If Devin could work for a year without any human assistance.

Destroy your competitor.

Oh, I agree. I just to be clear. So I and I think at some point it's kind of, you know, you see this, right, where it's like you know, when you're talking about seconds, you're literally talking about just like a specific command, right? When you're talking about hours, you're talking about giving it a task and having it do the task, right? And I agree with you. Obviously, bulk email is not, you know, for years, what you're talking about is like you're giving the agent a mission basically, you know, and it's like this.

Yeah. Exactly. And this is like what do I care about? And you know, the answer might be look, I'll give you a million different examples. You know, the answer might be like there's this like one, you know, societal problem which is really important to me and I think there's like I think it's a solvable problem. I think we can all be happy, but like you know, we really need to spread awareness about it. We really need to get folks to understand the points of it and we need to figure out how everybody should you know, work together and coordinate on. That's your point. That's a problem that an agent can think about, right? Or even you know some of the kind of like sillier things too like yeah you know there's this video game that I really like but I wish you know if I were making the game here's exactly how I would think about all these things and I feel like there's like this really cool idea if you know you could combine elements from this one game that I really like but then incorporate some of the elements from this other game and set the agent off on that mission of like look we're going to go and like make the coolest thing ever and the coolest game ever and like we're going to incorporate those elements we're going to think about how those like nicely intertwine and work together, right? Or if it's like here's this like piece of just like novel science which I'm just like really passionate about. You know, materials have been created this way for years and years, but like here's this like avenue of attack which I've been wondering about and thinking about of maybe there's a different like novel construction of materials this way. Send your agent to go work on that for years, you know, or months, right? And like have it study that and explore all these things and run its own experiments and try all these. I think all of these are I think soon going to be very possible. And to your point, it's very different.

Yeah. Yeah. Yeah. Then you'll have the AI which is like the manager AI of the missions. Yeah. And I think it's like I think we will continue, you know, it's my example of this is like I always joke about how, you know, if you think about our ancestors from, you know, hundreds of years ago or thousands of years ago, imagine them looking at us and what we do and it's like, you know, you're just pushing buttons, you know, and you're like sitting in a room and talking with other people and you call that a meeting and that's like those things that's work for you guys, you know, and it's like what do you mean that's work? You know, like I'm in the fields like I'm doing this every day, you know? I'm going in farming. I'm making sure you know making all of our clothes by hand or taking care of all these things and that's like work you know but like how can you guys call you know and my point is just I think what we will have going forward is going to look that different from what we have today. We will look at people who as we say just like have these really interesting curiosities that they want to pursue. They have like causes that they're passionate about. They have like fun ideas or like art that they want to create and like they're sending off their agents in pursuit of those missions and they will think of that as work and we will look at them and be like wow it's kind of crazy that that's what you get up and think about all day.

I'll say it was always two things. One, it was always related to code and software, which again probably has something to do with us all being programming nerds. And then two is it was always around the idea that these would be like real multi-step iterative processes. Which was I would say was a real hot take at the time like this was 2023.

Well, look, it wasn't even a product at the time to be honest. I mean, it was more just like a prototype or like a demo of what was possible and you know, we had been working with it and playing with it for a few months at that point. And the kind of just wanted to show people some of the examples of what it was capable of because it was a real like it was pretty insane for us to see as well. Like I remember the first time that it did like a real task like I could not sleep that night. And so it's just like showing that

The first task that Devin did was it like set up MongoDB for us and it's like you know it's a standard thing. It's issues that a lot of people run into all the time when they're going and getting their kind of like initial DB setup or whatever. But we would just like run into errors. You have this whole flow where you like you find an error, you Google error, find some message on Stack Overflow or whatever or you ask ChatGPT and it tells you, okay, here's one you should try. You try that thing, you run into a different error, you like paste that in

And then, you know, and like for this one, we kind of like at some point we're just like, 'Okay, Dev, just try to go fix it. Just go run the commands, like do whatever you need to go do it.' And then it worked.

And you couldn't sleep. It was truly like because again, it's you know, it's like just seeing the exponential curve ahead because this was a very, you know, very specific case. It was the one success that we had. It was very much like a, you know, like definitely a way better than average run. But no, there was this feeling of like why shouldn't all software and all products be built this way now? Like you can just tell it what you want it to do and have it go do it.

I mean it's kind of funny because yeah, to your point, like we did get hate in the beginning. We did not feel motivated at all from the hate to be like, 'Oh yeah, I know. Maybe you guys are right. Like maybe we should go and like, you know, focus on the more kind of like chatbot Q&A style product experiences.' Like maybe more than we should have. Maybe maybe we should have done something in the middle, you know? I I don't know. But but like I I think for us like when we had seen that and when we had done these different things like all of those like demos that we showed, you know, in in our launch announcement were like actual runs of Dev that we had done ourselves and like run into and been like, holy [__], this is insane for us.

It was always kind of like, look, you can, you know, we can debate when or or or like what level of effectiveness or whatever, but like it's just it's going to happen and that's how we always felt about it.

Yeah.

SWE-bench. Yeah. It was like 13%.

Yeah. At the time the best known was like 3 or 4% or something like that, but obviously, yeah, 13%. It still means you fail, you know, 87% of the time.

For sure. Yeah. I mean, so so so for us, by the way, you know, if you kind of think about this overall AI ecosystem, I mean, a simple way to put it, it's like, dude, we were late by a lot. You know, it's like OpenAI started.

Google DeepMind or Google Brain. I mean these are obviously, you know, more more than a decade old already. OpenAI started like end of 2015 or even, you know, like the Anthropic or or, you know, other that that folks talk about in the world like had been around already for years. Like we were getting started in early 2024. There's like a year plus out from the ChatGPT launch. A lot of the existing players were already there, you know. The same was true in code specifically. I mean there was, you know, GitHub Copilot, um, which had people had already, you know, engineers had already used for years and that was very much the like Q&A, like the autocomplete style experience of of you are working with AI.

You know, when you start a company, you kind of have nothing. Like you you have no right to exist is maybe one way to put it. It's like an interesting truth about the world which is obviously startups succeed. You know, startups succeed all the time and startup versus big company has been played out for, you know, for for forever. Um, but like there's no reason, you know, in terms of resources, in terms of people, in terms of brand awareness, like you have none of the things that, you know, the big guys have. And and so from that perspective, you have no right to win, you know, over over over what they're doing, you know. And then the reason that you're sometimes able to anyway is

if you really like, um, you know, plant your flag in the ground and and and put a stake into like what you think the future is and you run like hell towards that and and if you like turn out to be right on some of the core things and like I think we were we were wrong on a lot of things, uh, to be clear, a lot of things we, you know, so definitely we were early which is a very fair criticism. I think a lot of the details and the nuances like we learned and adjusted over time but the idea that like you would work with AI as a co-worker, you know, rather than as like a tool or a chatbot I think was I mean over the last couple years has obviously like really grown and I think it was like very important for us for our brand for recruiting for, you know, customer work everything for us to be, you know, the first ones that actually planted that flag in the ground and said

So, funnily enough, so that launch was in March of 2024 and to your point, it went very viral, but again, we didn't have any customers. We didn't have any revenue. We didn't really have like we had

So, we had that that was actually later on actually. That was that was end of 2024.

But, you know, initially what we started was just like, you know, a bunch of people came asked us for the product. We were like, 'Guys, I I don't know if this is a product that's ready for prime time, but if you really, you know, you can try it and like just let us know.' And so, like, people ran, you know, we had to go and scramble to build a system of like a okay, let's do like a pilot or like a POC, you know, and like we tried our best to like not overpromise and be like, 'Guys, like really, it's like very early, but if you want to try it, you know, you can try.' And we did all these and perhaps unsurprisingly, I mean, they were all just like failing

um, which is kind of natural. Like it's like you could do some pretty cool things with it. You could you could do some pretty interesting like toy demos or projects or whatever, but it was certainly not ready to work on like actual companies like real codebase and so on, right? And so from that point, this would have been like April and May of 2024. We were making agents work with like GPT-4, you know, this is like a very different era. So they they were much more primitive agents. We kind of talked about this and got to this question of like okay well what do we think it actually does look like when the agents start to get good enough for adoption and I think what we kind of came to was well

different tasks are different, right? And so so like there are some tasks out there that are just really mundane and really repetitive and you're just doing the same tedious thing over and over and over again and there are some tasks on the other end of the spectrum that are like really tough like architecture problems or like deep like, you know, deep issues that you have to like really understand all the context and have all the know-how how to to know how to fix and like people were trying to use Devin for all of those and were failing at it understandably and so then the question was like okay what are the the the natural tasks that are like if there's like a first task that is going to have PMF, you know, and real value from these kind of agent experiences where it can do the whole thing end to end what is that going to be we kind of said okay well it should be some of these like really repetitive tedious ones it's not so cut and dry that you can just like have an automated script that does the exact thing every time. So it does take obviously some intelligence and some amount of meandering, but it is like repetitive enough and scoped enough and like on a tight enough feedback loop that you can have an agent do it and it would be able to kind of go and diagnose and like fix that problem, right? And so that's kind of what brought us to some of these like naturally to to some of these like, you know, initial use cases which were things like migrations or version upgrades or, you know, helping people upgrade from like Java 7 to Java 8 or something like that which were kind of like, you know, as you can imagine enterprises had these massive code bases where they would go and do all that, you know, and it's like a 50,000 file codebase where you have to go, you know, it's like the same eight things that you need to change in each one you have to be a little bit thoughtful about how you make the trade-offs but like it's a very repetitive task, right?

Our first success ended up being with a company called Nubank, uh, biggest bank in Brazil, you know, by market cap at the time. And the use case was like one of these big migrations. And we had, uh, kind of a custom Devin that was like extremely extremely optimized, um, for doing that. And as we grew from there, you know, later on we had kind of as things got a little bit more mature, you know, we had both self-serve business and enterprise business and so on. But I think from the beginning we had always seen this value that building software in the real world and managing massive massive products that like millions of people use every day was pretty substantially different from, you know, just building a cute website or a cute demo from scratch, right? And I think we really really leaned into that and naturally it's, you know, all the Fortune 500 or all the biggest companies in the world are software companies in 2026, you know, even the ones that that folks don't necessarily think of that way, right? And like Walmart or CVS or JP Morgan or um or or Mercedes or what, you know, they're they're all software companies, right? They have massive massive teams of software engineers. They have tons of things that they're building and shipping and um and maintaining. And that kind of became like a natural thing for us to work on because like I think we had had learned very early on that we want to work on real problems and we want to work on things that that that matter and that people actually care about.

Today around 75, 80%.