Uncapped with Jack Altman
· Jared Friedman, Garry Tan, Jack Altman
[00:00:00] S1: We recently worked on a redesign of the Y Combinator website. Actually, Garry and I did. We actually tried to write it down possibly for the first time. Do you want to explain this since it’s really your words on the website? [00:00:00] S3: It was a team effort. Honestly, we tried to return to what the founder’s experience is. We found all of these old photos. Actually, your brother is the first one on there, and it’s him with the double popped collar. [00:00:00] S0: We’ll pop that up on the screen. [00:00:00] S3: Then you start flipping through and it’s Patrick and John Collison. It’s Brian and Joe and Nate at Airbnb. You see them so young, and then you see them ring the bell. Basically what we’re trying to create is like Disneyland for transformation: from startup founders who are just starting out to literally the people who make the companies that really matter. It’s actually a social process. All reality is socially constructed. I remember when I found YC and came to my first Startup School. It was like being a fish out of water and then jumping into water. I was surrounded by people who were builders, who were earnest. When you get into YC, we take people who are earnest and technical and at the end of that process, hopefully they become formidable. When you go to the homepage, that’s what it’s about. YC is a transformative process. It’s like Hoffman or something. It’s not that New Age-y. It’s actually very chill. I don’t know how you would describe it. [00:00:00] S1: I’m a Harry Potter fan, so I prefer to think of it like Hogwarts.
· Jared Friedman, Jack Altman
[00:00:00] S1: I’m a Harry Potter fan, so I prefer to think of it like Hogwarts. [00:00:00] S0: When I did YC in Winter ‘16, starting a company felt like such a weird, odd experience. In some ways, having this group that normalizes what it’s about — where you’re surrounded by other people who talk about the new language and the new set of things you should be thinking about — it almost calibrates the strange dream that you’re in. That was a big part. There’s also always been the stamp of approval, which is hard to underweight. I’m curious how you guys think about that with outsiders in particular. One of the things that seems different to me at least is that now more than 10 years ago, and I’m sure more than 20 years ago, the ecosystem is an understood thing. Founders can read a lot of stuff online. People can read a lot about them online. There’s more known in general. But I still feel like there’s this thing YC can do, which is take people who maybe aren’t yet in that vortex, identify talent somehow, and bring them into that vortex. I’m curious if you guys spend active cycles thinking about that?
· Jack Altman, Garry Tan
[00:00:00] S0: When I did YC in Winter ‘16, starting a company felt like such a weird, odd experience. In some ways, having this group that normalizes what it’s about — where you’re surrounded by other people who talk about the new language and the new set of things you should be thinking about — it almost calibrates the strange dream that you’re in. That was a big part. There’s also always been the stamp of approval, which is hard to underweight. I’m curious how you guys think about that with outsiders in particular. One of the things that seems different to me at least is that now more than 10 years ago, and I’m sure more than 20 years ago, the ecosystem is an understood thing. Founders can read a lot of stuff online. People can read a lot about them online. There’s more known in general. But I still feel like there’s this thing YC can do, which is take people who maybe aren’t yet in that vortex, identify talent somehow, and bring them into that vortex. I’m curious if you guys spend active cycles thinking about that? [00:00:00] S3: All the time. I’m in the middle of my very intense addiction to Claude Code and Codex. Actually using this stuff is pretty wild. I basically recreated my 2008 startup, 70,000 lines of code, in about 90 hours over two weeks, while having a full-time job and trying to raise kids. It just really compressed a lot of my sleep. But at the end of it, I have a codebase that is better than what it took five engineers, and me taking anti-narcoleptics, to build for my YC startup. [00:00:00] S0: It’s crazy.
· Jack Altman, Garry Tan
[00:00:00] S0: It’s crazy. [00:00:00] S3: Something happened at the end of November when Opus 4.5 came out. I heard about it and I was like, “Hey, what’s going on?” The most interesting thing for us is that we’d been talking about it for years, been using it for years, and then it wasn’t really until December where it was like, “AGI is here guys, for code.” I feel like I could create in 80 hours something that I could not create with $5 million and five engineers in two years. [00:00:00] S0: Let’s stick with that, this coding topic. First, what are you trying to identify as greatness? If before this moment you needed to find great engineers, what is the thing you’re looking for now? How has that changed? Second, the startup advice has always been “write code, talk to users.” What is that now? [00:00:00] S3: Prompt and talk to users. [00:00:00] S0: But there’s a lot less time prompting.
· Jack Altman, Garry Tan
[00:00:00] S0: But there’s a lot less time prompting. [00:00:00] S3: This literally just happened and all of us have been collectively doing it. Harj did a project, Jared‘s coding all the time. Our whole team is using this stuff. We realized, “Hey, why don’t we actually put it in the application process?” For the first time, for the spring batch, you can upload a transcript of your Codex or Claude Code making a feature. We’re starting to figure out how to process it. We put it in a security sandbox because we figure it’ll be prompt-injected very quickly. You can tell a lot about whether someone can build just from how they prompt the agents. Do they know systems? For YC, you get a t-shirt that says “Make Something People Want” on day one. If you look at a résumé, you can sort of guess at whether they can make something, but you can’t really tell whether they can make something people want. You can look at their GitHub and maybe see if they can make something. The only way you can really tell if they can make something people want is if they did it already. [00:00:00] S0: And there’s not exactly a GitHub for prompting, if that’s what you’re asking people for.
· Jack Altman, Garry Tan, Harj Taggar
[00:00:00] S0: And there’s not exactly a GitHub for prompting, if that’s what you’re asking people for. [00:00:00] S3: But you can tell… Do they use plan mode? Do they think about systems? Are they prematurely optimizing? Are they over-engineering? Basically, what is a feature to them? What is the complete release? Do they think about the edge cases? How you do anything is how you do everything. My favorite Steve Jobs quote, he talks about how if you’re a carpenter, you can tell other really great carpenters. You don’t look at the front. Everyone looks at the front and they’re like, “This seems good.” But a great carpenter, an artisan, looks at the back of the cabinet. The back of the cabinet only other carpenters are going to look at. That was one of my favorite things. When I funded companies at Initialized, and when I was a partner at YC the first time around, that was the number one thing that I loved. Is there “game recognize game”? Would I go work for this person if I weren’t doing the thing I was doing now? [00:00:00] S2: This is how you spotted Apoorva with Instacart, right? [00:00:00] S3: Yeah. He came in and honestly it was craft again. He’d literally built an iPhone app that was a demo app. Claude Code didn’t exist, so he actually had to build it himself. The way I could tell was that — turning the cabinet to the back — he could scroll it and it scrolled really smoothly. That wasn’t true for that era of iPhone apps. The App Store and iOS apps were brand new. I feel like you guys do this all the time.