Start by giving your background a little bit because you've got a particularly interesting background and what you did. You're the founder of Instagram and you went to Anthropic, so talk to us about all that.
Yeah, maybe the through line. I've always been interested in the combination of what's possible with software and computing and then how do you actually make that real for people. And so when I joined Stanford, I studied symbolic systems, which is a very odd degree, although it's gotten more popular in the years since, which was this combination between cognitive science, design, computer science, philosophy, psychology, and AI. And the AI at the time was still mostly finding paths through mazes, not what we call AI today. But I love that kind of intersection of, you know, you can have the world's most powerful software, but if it's not comprehensible to people, you haven't actually succeeded in doing anything real. And kind of carrying forward in that, I first worked at a company called Meebo, which was actually a great training because the company culture and the recruiting and the hiring was so strong there. And from that, co-founded Instagram. And Instagram, a lot of what we were focused on was like, rewind to like 2009. Mobile apps were still emerging, cameras were starting to become more ubiquitous, but most people still felt like their photos were kind of stuck on their devices. A lot of what we were trying to do is like on every screen is very clear what is the one thing that we want you to do. You know, is it stupidly easy to figure out how to create a graph to become good at Instagram and still be able to progress in that as well? Then I did another startup with the same co-founder as Instagram called Artifact at the beginning of AI. So AI-powered news recommendations. I still get probably every event is one person that comes up to me and is like, I loved Artifact. I just shut it down. We did it for about three years and realized it was not on the trajectory to be the kind of place we wanted to spend the next three years. So we ended up selling it to Yahoo. It lives on as Yahoo News. And then I had this intersection moment where, you know, do I go start another company? I love founding things. I love 0 to 1, or do I go join somewhere? And my previous three years at Artifact showed me two things. One is coding was about to change. So I had started using like early coding models to accelerate my own work. At Artifact, it was so early. Still though, I would draw these models and have vision yet, so I would draw ASCII art of the screens I wanted to implement. And say, hey, ASCII art, like, you know, dots and dashes, and say, great, please implement this in Swift, and hit enter. And it was, you know, two tokens for a second. I'd go make a coffee, maybe take a walk. And when I got back, it was like a start of what was actually going to be good. But you can see, I mean, the trajectory was already there. So number one, coding was going to change. And number two, the models were going to provide this intelligence layer that were going to take months of development time of a feature down to, you know, potentially hours. So for us, some of the intelligence that we baked into Artifact were things like take an article and summarize it, take a headline that's written to be super clickbaity and then rewrite it to be factual. These are features that, if we had to spin up a research team, would have taken us months to do well. But in this it was already, this is 2020, 2021, an API call away even for this initial version. So it was really clear to me that the next generation of software was going to be built and powered by these labs. I, you know, talked to the folks at Anthropic. They shared that vision. Also, just love the culture there. We can talk more about that as well. That felt like that was going to be the real springboard, and I wanted to be a part of that.
I think what you're doing is actually not only amazing, but also, it was crazy unlikely when you folks started and what you've accomplished is no small feat. But before we go there, any thoughts on like a widely held belief in tech that you think is fundamentally wrong?
There's this belief that design is going to go away right now, and it's, you know, that you can prompt the models. Okay. By the way, what's that? Dylan was here. Yeah. So Dylan, as we both share this contrarian belief, to me, there is still the difference between software you use and software you love. And I think that that gap of I actually come back to this, I'm choosing this. I have like changed my behavior because of this. I've like gone off and told a bunch of other people about it. I've yet to see that with a lot of sort of, you know, off the cuff, you know, vibe coded stuff. I think there's still that need for what's the craft, what's the reason? Like, why does this exist? What's the brand? Right. Why does it all mean like to a person? So that's one where I feel like the demise of thoughtful design software is like overstated. Now, the actual role and how you got there will absolutely have to evolve. But I still think that there's a need. I see it internally. We can talk about my transition to working on labs now. I spend so much time talking to designers now, and their role has definitely shifted. But it's, I think, as essential as ever.
And so if you were to think of the big mistake the product folks are making, because right now what's happening in the labs is there's been this huge investment that's happened in research. Research is such a foundational element of, you know, building out the models. But the companies that are doing well are the ones that also have exhibited very strong product sensibilities. Any kind of thoughts on what mistakes are product folks making in this next wave? Because they've actually tried to pour too much of the previous wave to the next wave, and what are the things that we should be reframing for ourselves?
Yeah, I mean, I think one really significant one is staying too much in sort of mock up or sort of, you know, faked prototype land before you actually get something real that you can actually then get a sense of the model capabilities. A lot of what we do is try to build products that don't work yet, and then wait until the next model release and see what does the product work now. Like we've built computer use. So, you know, Claude driving replicator and like as a prototype in my third month at Anthropic, this was 2024, still out. And it was terrible. Like it was, you know, it was clicking all over the place like it was, you know, it just didn't work. But even that prototype taught us, all right, where are the model capabilities? And every time the research team had another research model, we throw it in there to be like, well, it got past the file menu and I got to the open me, you know, it got a little bit further. And then there was this moment, I think it was Sonnet 3.7 where we threw it in there. And actually we didn't expect it to have improved that much. So we kind of did it in kind of as a good practice. And I suddenly got a ping from one of the researchers there, like, look, it actually got good. We can actually productize this now, but we wouldn't have known it if we didn't have that harness throughout. So I think you just need to reframe your sort of purpose as a product person from being I deliver a beautiful product requirement doc. I deliver beautiful wireframes. My designer and it's headed off to engineering to software is now a living breathing system with this nondeterministic wonderful but also, you know, infuriating sometimes engine at its core. You need to get to that sort of moment where you're experiencing that as quickly as even in the enterprise, even especially in the enterprise.
And so all right. So let's talk about because you and Dario and everyone has been pretty vocal at Anthropic about this notion of safety and security and how you need to actually not just have agents run loose in the wild west. But do you feel like more autonomy is better or you will get better adoption rates if you can control the autonomy a little bit of agents?
I think that one of the things that's magical about what these models can do now is, given the objective and fairly high level instructions, find their way there. And, you know, sometimes it's kind of endearing. Somebody to sit on one of our engineers sent me here, asked Claude to prepare a sort of a vacation preparation doc for his family, but Claude had hit a block loading the website of the vacation rental. So instead of showing photos, it like drew its own imagination of what this house look like. You're like, oh, I'm not sure that was exactly what I wanted, but it's kind of like it's creatively trying to solve the problem. Like, if you've never seen a model try to like solve a problem in one of these because it's fascinating. It's like, well, I can't do that. Should I try this or I still can't do that. I'm going to pretend like it. It really wants to solve that problem. I don't think you want to sort of prevent it from doing that. Like hold it like entirely, because that is part of the magic of being more attentive. But you do want to sandbox it in the right way. So what we find a lot in the magic systems that we built inside enterprises is all right. The box that Claude is running in, what are its permissions, what's the minimum amount of permissions, and what's the sort of conduit between that and the outside world? So you want autonomy, but you want autonomy in a sandbox with the right sort of abstractions around when order is off. Exactly. And when it goes off, and then when information flows in and out. That's also a really good point around adding observability. So it's yeah, autonomy within sandboxes with the right connectivity elsewhere.
And so talk to us about coding. You folks have actually really focused on that one and crushed it. Where does coding go from here? What does coding look like in the next year and the next two years? Next three years? Yeah. And by the way, talk about not just the fact that it's going to get to 100% code writing, but how does the software development life cycle change? How do the roles change? What should the CIOs and companies be thinking about to adopt to this new model of coding?
Yeah, I love this topic. I feel like I could talk about it for an hour or so. I'll try to be brief and pointed. I'll maybe give a vignette from what it's like right now inside Anthropic on our labs team, which is, we are regularly producing 2 to 3,000 line pull requests with each other, especially on the live stream where we're moving extremely quickly. It's being written by Claude. Claude products and Claude Code are being entirely written by Claude. If you know, when Dario, I think was on stage a year ish ago, I was like, oh, by the end of the year, 90% of code is going to be written by Claude. People thought it was crazy, but we were already seeing that trajectory. Right now, for most products at Anthropic, it's effectively 100%. It's just Claude writing. And then what we've done is created all the right scaffolds around it to like let us trust it. So the few things that the how software development has changed, one is having Claude trained and sort of prompted to be a really good adversarial coder of year has been fantastic. I'll put up pull requests, and Claude will come back and say, here are the security vulnerabilities. But here's also how it could be refactored. Here's how it could be different. And sometimes it, you know, the sounds adversarial coder view. I was talking to the person that had built it internally, like, what did you do? And like you showed me the prompt. It's actually not rocket science. It's just really prompting to be like a super tough grader. Like what are all the different problems that you can do? What are the different automated checks? And enforcing some softer sort of requirements as well. So that's one of them, that I think really matters. The second one is like being willing to let Claude re-architect things. I was working on an iPhone prototype, and I had gotten to the point where the sort of initial architecture hit a wall, where I was just having trouble parsing it, asking Claude to step back first, write some verification, and then go off and refactor it. You know, we've talked about there's components at Anthropic. We're not an old company, we're only a few years old, but you can accumulate tech debt at a very fast pace, especially in the age of AI. Paying down that tech debt, that calculus has really changed for us, too. So that's been another interesting aspect of it too. And then the third one is I felt this a little bit a year ago, and I really feel it now is how much all the other processes become bottlenecks. So, we had an outage of our sort of continuous integration system yesterday and, you know, that would have been a mild annoyance because you maybe were producing one change or pull requests a day, and you have to wait an hour. When you're producing, like, no joke, a dozen, at least per engineer on the team. It was just like I felt physically pain because it was like all this stuff is just like coming up the works. It also makes recovery more challenging. And so that focus on developer tooling on developer efficiency, on the developer infrastructure that was already mildly important is now sort of, the any blocks or outages or friction becomes a sort of penalty on your teams.
Is the scarcity now getting to be reviewing and auditing code rather than actually writing code?
It's reviewing, auditing, integrating. Right. Because now you've also got the thing where you've got multiple engineers contributing to the same place as well, and then driving alignment first on like what the product should be, which is sometimes always the hard part. Now it's really clearly seen as the hard part, but also what is the sort of principles and architectures that you're going to want to build within your application. This is even more important in the enterprise. But aligning on that and what are those standards so that when you're reviewing or when Claude is reviewing the code, you're not just kind of duct taping things together, but you're evolving towards some consensus Northstar around architecture code principles like problems that you're solving that ends up becoming the bottleneck.
And if you know, the internal philosophy of Anthropic is extreme use of experimentation, right, of AI, and that you folks are very actively talked about that when you think about advising the enterprise, does that kind of extreme use of experimentation something that you recommend to customers?
