Claude Code creator Boris Cherny on the end of the software engineer

Some people think the idea that we're all going to lose our jobs to AI is just hype. This week, I'm talking to one person who might make it a reality. This podcast is brought to you by Atlassian Rovo, the AI that takes your team from AI novice to AI native. Welcome to Platformer. I'm Casey Newton, and my guest this week is Boris Churny of Anthropic. He's the creator and head of Cloud Code, which is the fastest growing AI coding tool in the world, and may also be a working preview of fully automated software development. So if our first two guests on this series, which is about AI and jobs, try to calm us down about the risks, I suspect that Boris is going to rile us up. He has said that software engineering jobs will start to go away as soon as the end of this year, and I'm really looking forward to pressing him on that one. But before that, as always, we're going to begin with the numbers. Each week, we kick off the show with fresh data trying to help us make sense of what is actually happening on the ground. And to do that, once again, we're bringing in Platformer Fellow and Gen Z AI correspondent, Ella Marchianos. Ella, how are you this week? um i'm wonderful i've been reading lord of the rings return of the king and it's kind of changing my life uh that has that has nothing to do with ai um it's just truly excellent human generated writing i mean there is like a parallel between the ring and ai um that i always think about it's like you want the power and then like maybe there's in fact something like insidious that comes with the power of AI, we see again and again. It's true. Well, you know, I will say, you know, I'm always glad to see you reading human-generated writing, as you call it. But I have one caution for you, which you may have already considered, which is that people in Silicon Valley who read The Lord of the Rings do then go on to start some of the most terrifying companies ever with names selected from Lord of the Rings. So, of course, Palantir, Anderil. so have you had any inkling yet of like a sort of a name that you've seen and that's made you think I could probably do something really evil with a company named after this you know like minus Morgul it's in stealth right now so like maybe we should cut it out of the podcast but I have some hope that we're going to do some we're going to flip the narrative on Sauron related tech company names We're going to do really great things with our wonderful team of undead horsemen. Fantastic. We look forward to learning more about this company in the months ahead. In the meantime, though, I wonder if you have seen anything interesting this week related to AI and jobs. Yeah, so there's this study from Microsoft. they've surveyed 20,000 AI users. So worth keeping in mind, we're like narrowing down to the population of people who are like actually using AI at their jobs. And there's this thing they're calling the transformation paradox, which is basically 65% of people who are using AI at their jobs, they're worried that they'll fall behind if they don't adopt AI quickly. And also people are finding it useful, like 58% say they're producing work that they couldn't have if they didn't have the AI they have. Like a year ago, they wouldn't be producing this kind of work. 66% say it lets them spend more time on high value tasks. So like, one, people are using it and they're liking it. Two, they're worried that bad stuff will, like they won't be able to keep up if they don't use it more. However, the like really big contrasting statistic is only 13% say they're rewarded for experimenting with AI at work. So basically, there's this gap between like appetite for AI use and like the institutions themselves that people are working at. This is so interesting to me. And I think it gets at something very real based on my own conversations with people at their jobs lately, which is that AI is like all stick and no carrot. You know, the like the prototypical CEO is conducting an all hands saying, we're going to AI all of the things immediately. You must be AI-ing at all times. And yet when workers go through with this, it doesn't seem like there's actually much reward waiting for them on the other side. Yeah. And I think another thing that a second Microsoft study found like a little earlier this year was when managers actively model AI use, AI use goes up like 17 percent and people also trust agents 30 percent more, which like should you be trusting the agents? That's another question. But I think that kind of there are a few different ways you can like try to encourage your employees to use AI. One is like AI is the future. Guys, please use AI. And another is like, hello, direct reports of mine. Here are like the specific ways like I use AI as they're surveying here at my job. And like, here's how you could try to do a similar thing. And it turns out like that kind of strategy does seem to get results. That makes sense. I mean, my question is like, what would this look like if managers were actually financially rewarding their workers for using AI? It's like, you know, right now, so many people understandably are reluctant to use AI because they do not want to train their own replacement. They do not want to like hasten the end of their own job. But I think if they had reason to believe, hey, if you become more productive because of this tool and you raise the output of this company, like you are going to share in the spoils of this. I don't know if this idea just comes across as like pure communism to Silicon Valley, but like I have not heard basically anyone who seems to be trying this lately? Yeah, I mean, I guess like to some extent there's this thing going on where at least workplaces are like subsidizing tokens, which sometimes I would like to bring up doesn't end super well. So let's hear about this. So at the same time as we're seeing this like gap between, in fact, like worker demand and like how much managers are rewarding their workers for using AI, We're also seeing, quote unquote, token maxing at major tech companies. Now, do you identify as a token maxer? I'm not token maxing. I'm not running agents at night. When I get an AI to do code for me, I'm kind of sitting there. I know what it's doing. I just there's just like isn't any task for me as a journalist where I'm like, you know, I need to make have Claude make like an enormous code repo where it's like constantly thinking to itself or something like I don't I don't even know what like non-destructive activity I would do. Well, I told Claude to rebuild Palantir from first principles, and that started to burn so many tokens. We actually have to pull the plug on this. So thank you for your moderator. I'm in favor of token moderating, okay? Token minimizing, we don't need to do that. But token maxing is a bridge too far. But you're saying that there are companies in Silicon Valley where they are truly token maxing. Do you have a good example for us? Yeah. So Amazon. So previously we got some stuff meta this year or this week. This year, a lot of stuff has happened this week. Some Amazon employees reported to the Financial Times that basically now that Amazon has adopted this new tool MeshClaw, which is an internal Amazon tool inspired by OpenClaw and is encouraging workers to use it and also has these leaderboards within teams like various leaderboards. of token usage. Some people, according to employees, are just like running agents, not even that do productive stuff, just like as we saw at Meta, like maybe just sometimes like randomly go in a loop so that their like token usage goes up. And for example, at Meta previously, we saw a bit more numbers of like the biggest token usage in the biggest token leaderboard. It was like hundreds of billions of tokens. And it was like an amount that clearly would have cost Meta literally millions of dollars. We're like some of that truly was going down the drain. Another thing that I found really interesting about the dynamic described in this Financial Times article is the official word from on high is that it's in fact your token usage is not supposed to be a metric your managers take into account. but employees still think that managers look at it and so in fact they're still just increasing their raw token usage and then also amazon has this uh high up corporate target for 80 percent of devs to use ai every week um and so it's like the signal from on high is we want you to be using ai to an extent that people are like in fact sometimes at least doing absurd stuff where like To me, I'm not there. I'm not one of these managers at Amazon. I don't know as much about how you manage a team of software engineers. But like, I feel like as a dev, I would want less one doublespeak about like how my AI is being tracked and like two more productive metrics that like are clearly communicated and relate to how I'm adding value with AI. Yeah, and I think that this story speaks so well to that Microsoft Work Trend Index that you brought to us at the top of the show, because if workers do not feel like they're going to be rewarded for using AI in a very specific way, they may use it in a very silly way, right? They're going to try to honor like the letter of the intent, which is use AI, but like miss the spirit of it, which is get better at your job. And maybe they would get better at their job if they had a strong financial incentive to be using it in their work. Maybe they wouldn't even burn so many tokens. But you know, Ella, I know one person who probably is not that stressed out about the major burning of tokens that we're seeing all around the industry lately. And that is Boris Cherney, the creator of Claude Code and somebody without whom I think the token maxing trend might not be possible. And we're going to bring him in and talk to him right after the break. Thanks for joining us, Ella. Not sure how to actually use AI at work? Most AI tools promise to save time, but they set outside your workflow. They're disconnected from how things actually get done, and they don't make your to-do list any shorter. Alassian Rovo works where you work, across Jira, Confluence, and the rest of your stack, securely connected to your company's knowledge, context, and permissions. Turn meeting notes into Jira tickets. Draft campaign briefs in Confluence. Instantly find the right docs without digging. This isn't generic AI. This is AI that understands how your team actually works. Less searching, less busy work, more progress. That's what an AI-native team looks like. Learn more about Rovo at rovo.com. My guest today is Boris Cherney. Boris is the creator and head of CloudCode, the agentic coding tool that Anthropic put out in May of last year. In short, it's a hit. a box that you type words into and spits out code. Within eight months of launch, Cloud Code was responsible for about 4% of all code pushed to GitHub. And by February of this year, it had hit an annual revenue run rate of $2.5 billion, the fastest enterprise product ever to hit that mark. Part of what's fascinating about this is who built it. Boris doesn't have a computer science degree. He studied economics, dropped out of college to run a startup at 18, did a stint at a hedge fund, and then spent five years as a principal engineer at Meta before joining Anthropic in late 2024. He didn't even arrive there with a mandate to build a coding tool. He showed up to learn the API, started hacking on a side project that initially just hooked Claude up to AppleScript so it could see what song he was listening to. And within two months, he had a version of Cloud Code that 20% of Anthropics engineering team was using on the first day. So our first two guests on this series were somewhat bearish on the idea of AI automation. Aaron Levy from Box and James Manjica from Google both made the case that jobs are just harder to automate than they look, and that while the future is going to bring serious disruptions, that doesn't necessarily mean massive job loss. I think Boris just has a very different perspective. As a software engineer, he is actively working to automate away his own job. He says that 100% of his code is now written by Claude. He ships between 20 and 30 pull requests per day by running five Claude agents in parallel across five terminal tabs. And he said publicly that within a year, the title software engineer is going to start disappearing, replaced by something that is more like Builder. So as always, when I talk about Anthropic, it's important to say up front that my fiance works there. You should take that into consideration as you listen. But this was one where I just couldn't imagine talking about the future of AI and jobs without talking to the inventor of Cloud Code. So with that, here's my conversation with Boris Cherney. Boris Cherney, welcome to Platformer. Thanks for having me. So you joined Anthropic in September 2024. And my understanding is that no one told you to go build a coding product. You were just trying to learn the API. So can you tell us the origin story of CloudCode? Because I've read that it controlled your music. Yeah, all of these things are true. So I joined this team called the Labs team, which built a bunch of cool stuff. So we built CloudCode. So I built that. There was a different person that built MCP. There was something that built skills. And then two other people built the desktop app. And that was essentially the size of the team. It was like a tiny team. We just sort of like built the feature in the course of like a few months. And we didn't know because a lot of these were kind of weird ideas. We had no idea if they were going to work or not. So, you know, for, I guess like 75, for Anthropic for a while, the focus has been on the same kind of stuff. You know, it's always been about enterprise. It's always been about coding. It's always been about safety. And we kind of knew that somewhere in this journey, we should probably build some kind of product. kind of early on in Anthropic, we didn't actually know if we wanted to build products. But, you know, if we're building, if we're building products, then you know we need to build something coding related because it helps us build better coding models so that everyone can use those models And it also just like helps us study safety There sort of a bunch of reasons to do this We didn know what it should be though So at the time, if you look at like coding products at the time, they're all like IDEs, IDE extensions, like the capability of the model back then, this was like Sonnet 3.5. It was not very good yet. So if, you know, the best it could do was like fancy autocomplete, you wrote a little bit of code and it would complete the line of code. That's where the model was at. And we had this feeling that there's this like model overhang or product overhang. And it's this idea that you could build a product that does something that the model is actually totally capable of doing, but just no one has built a product that lets the model do that. And I got to tell you, it's still the same feeling today. There's still this like intense feeling. The model can just do all these things. There's no product that lets it do that. And so, you know, we wanted to build a coding product, didn't know what it was going to be. And so I just wanted to learn how to use the Anthropic API because I was like, all right, we're going to build a product that should learn how to use the API so I can build the product. And I just built the cheapest possible thing. It was a little thing that ran in the terminal. It was the thing that I could build so I didn't have to build a user interface or an app. It was just fast. I built this thing in a couple of days and I just started giving it to people to see if they would use it, how they would use it, just out of curiosity. and I remember like over the next few weeks more and more people at Anthropic started using it like first it was just the people that literally sat around me then it was kind of like the next layer of the onion kind of outside of that and then like a few weeks in just like a lot of Anthropic was using this every day and it was weird because it was a little prototype and the terminal it's like the most engineering possible product a lot of engineers don't want to touch a terminal but they did it and they used it I've read that within five days of the initial release half of the engineering team was already using it. And I wonder, as that was happening, did you have a moment of thinking, okay, like software engineering just changed forever? Or are you still sort of iterating on the product and pushing pull requests? Dude, I was so focused on just shipping this thing. Like for me, as soon as I got this idea, I just spent like every night, every weekend, this is the only thing that I thought about, the only thing that I worked on. Like I started having dreams about Quadco back then. And like, that's still kind of all I dream about every night. It's just like, what should we do next? Like, what do we build for the product next? So, you know, I think now there's a chance to kind of zoom out a little bit because a lot of people are using it and we should kind of like, there's a lot to learn about the way people are using it. But for a long time, we were just so focused on building. I just didn't even have a chance to think about like, what is this? Yeah. Was there a moment when you did sort of do that zooming back? Because I have to imagine part of the reason that you're dreaming about it is like, you realize that, you know, it might be too, like, minimizing to say that you stumbled across it, but it does seem like there was a sense of, like, somewhat accidental discovery here. Was there that moment of, like, oh, gosh, like, yeah, this is different than some of the other things I've hacked on? Yeah, I mean, there's a lot of, there's a lot of surprise. Like, like I said, broadly, we knew we wanted to build a coding product, but, like, no one thought this coding product would be in a terminal. There was a lot of moments of surprise, just so many. The first one was when Quad, you, when Quad told me what music I'm listening to, And there was like a couple of versions of this and we actually have like a, there's a video demo I recorded of this and we actually just like donated it to a computer museum. Just it's this like very weird historical artifact. Like, and it was this video. I remember posting it on my Slack and there was like two people that liked it. Two reactions because no one understood what, you know, that this would be it. But yeah, the first moment was like, I asked Claude, what music am I listening to? and it wrote like a little bit of code to like open my music player and it wrote the code in Apple script which I don't know and like I wouldn't have thought to like write code to answer that that's crazy and it just sort of did it and I was like wow this is surprising it solved the problem in the way I wouldn't have as an engineer and you know over the last year and a half there's been so many moments like that I just actually had one of these like with co-work every time that we were releasing a model, I kind of experiment with it and, you know, kind of like see what was the frontier of what this thing can do. Cause that's like one of the hardest things about building on a model is it's, it's advancing so fast. You just have to kind of like recalibrate every month as I'm sure, you know. Yeah. And, um, I used to work for the first time to book a bunch of flights and usually it works. Okay. This time it was the first time it worked perfectly. And you know, anytime I travel, I use cowork to book it and yeah, it booked, um, it booked eight flights five hotels the only mistake was one of the hotels was just like way over budget yeah i just thought okay this is this might be like a little too pricey i think it was like five thousand a night or something and i was like co-work wants you to have a great time when you went at your stay you know and i was like please please rebook this one but then you know otherwise like it just like worked for like a couple hours and did all this it was just it was so cool like i i just feel the surprise like every every week every month so i'm gonna get to co-work in a bit But this feels like a moment to zoom out a bit, you know, from the story of initial discovery spreads rapidly through Amphropic and now has become a default tool for a very quickly growing number of engineers. And it is one of the products, I think, that is sort of making this question of jobs automation feel really salient, at least for the software engineers, but maybe more folks than that. So during our first episode, Aaron Levy was talking to me about the same subject and he said he didn't see jobs going anywhere, that there's always going to be a kind of last mile of human work that the software can't do. You have publicly predicted that the title software engineer could start to go away as soon as this year. So is Aaron wrong about this? I think there's a bunch of stuff that's true and a bunch of stuff that we don't know. So, I mean, like the trends are just, it's just exponentials. Exponentials are very hard to think in. So honestly, everyone that's saying that they know, no one actually knows. We're all guessing. And some of these are like educated guesses based on what we're saying and based on history. I think what's going to happen is a few things. One is there's going to be a lot of companies that need less engineers because engineers are more productive. So you just don't need as many engineers to do the same work. I think at the same time, there's going to be a lot of companies that need a lot more engineers because every engineer is more productive. The company can do more things. It can start more products. It can create more businesses. And like you see this with our team, like we are constantly bottlenecked on good engineers. We are hiring as quickly as we can. And there's a lot of companies and a lot of our customers are exactly the same. So I think both things are going to happen. And it sort of depends on the company and it depends on the business. And I think there's this other thing happening where all the roles are kind of blending together in this kind of interesting way that I don't think anyone would have predicted. Like, you know, our manager, Fiona, she has not coded in like 15 years and she joined QuadCode and now she's coding. And, you know, Kat, our product manager, codes and, you know, like Megan, our designer, codes, like everyone on the team codes, like you don't have to be an engineer anymore. And so this is what makes me think that over time, if you kind of project this trend a little bit, what's going to happen is everyone that's not an engineer is going to code a little bit more. Engineers, you know, like me, I haven't coded in six months. I'm like building stuff all day, but I haven't written a line of code in six months, like over six months now. And so I just see it all kind of blending into one thing. Like we can call it a builder. We can keep calling it an engineer. We can call it a product manager. I don't know what the title is, but the role is changing. So the way that we conceive of these roles is definitely going to change. But what that means for like how many jobs are available at which companies is still very unclear. year. Yeah. And I think history, you know, like have, has a lot of examples like in, you know, different ways, like, you know, like the tractor was invented. I was actually just like reading about this the other day. Tractors were invented in the 1890s. It was this guy, uh, John Froehlich invented it in like Iowa or something. That sounds right. And at the time, if you look at like the way that farm work worked, it was all horse powered. You needed horses to do farm work. And even though tractors were invented in the 1890s, it wasn't until the 60s in the U.S. that there were more tractors than horses. It took like 70 years. And it's sort of like, if you look at the trend, like the number of tractors went up, the number of horses went down. Both things happened and the intersection was in the 60s. And there was like, there was a bunch of reasons for this, right? Like the technology of tractors was magical. It could make it so you can harvest a lot more crops. Your productivity was a lot higher. But at the same time, if you're a farmer and you want to learn how to use a tractor, you need to learn. Like it takes training. And at the beginning, the tractors were expensive. So, you know, actually in a lot of cases, you still wanted like horses because it was still cheaper. And they were like not very good at first. You know, like maybe you could use like you could use it for like wheat, but maybe not for corn. And so it actually took a long time for someone to make a tractor that would work for corn and that would work for like okra and, you know, like all the stuff that you're, you know, using this machinery for. And that just took like a while to figure out. And I think the thing that we're seeing right now is like this on a speed run. and you know it's but it's sort of the same thing we're hitting like very similar issues totally i mean this is the kind of like ai as normal technology argument that like even as labs come up with just incredibly capable models people tend to be slow to change or organizations are slow to change and so it can just sort of take time for these technologies to filter through companies at the same time i think people look at what has been reported about anthropics revenue and they say it doesn't seem like it's taking that long this time around right so i think we're still trying to like hone in on what is the actual rate of change here yeah okay so okay so here's a question for you um yeah do computers make you more productive yes yes they make me more productive but does making me do they make me more productive feels like a different question than do i work less because of computers if that makes sense so so like because you can do more stuff you like do more stuff. You can like fit more things in the same eight hours or whatever. Absolutely. Like, I mean, like to be candid, you know, I used to record one podcast episode a week in addition to writing a couple of newsletters. I'm now experimenting during this mini series with doing a couple podcasts a week in addition to writing multiple newsletters. And AI is a reason I can do that. It is an incredible research assistant and podcast producer. And so my, I'm able to produce more, but I don't feel like I'm working less. And that's not a complaint, by the way that's just sort of like how i'm you know navigating this moment yeah yeah that's right that's right i i feel the same way like i feel like i can do so much more and all this stuff like i i didn't i just didn't get to before because i didn't have enough hours in the day now i can do but there's this like little um there's this other like weird historical thing where like in the 90s when computers were being adopted by companies like the personal computer like you know like after the mainframe after these like big industrial computers that cost like millions of dollars at some point they got miniaturized so like the average startup the average company could just like get computers there was this question of like are computers making you more productive and actually what people were complaining about is that they're not in the 90s like this was like an open question like do computers actually make you more productive and like now we look back on it it's like duh like of course they do like i can't imagine like going back to pen and paper but there's this really awesome uh harvard business review article i think it was like 1992 or 1996 or something. And essentially the case it was making is like they studied a bunch of companies that were adopting computers and they were like, okay, these ones are getting more productive. These ones are not. What's the difference? And what they found is the ones that are getting more productive are the ones that threw away all their paper, you know, filing cabinets. They threw away all their like paper and pens and all their like desk drawers and stuff. And they just replace it with like a computer at the center of everything. And then there's all these other companies that were still like, you know, they have like teams of people writing everything on hand by pen and paper. And then there's like a computer in the corner that's like used for something. And so like the first category has big productivity gains, which increase. The second category does not. And I think it's kind of similar right now because like at Anthropic, we really organize everything around Quad in every way. Like when people join the company, if they have like questions about like, you know, how do I write code? Or, you know, like how do I contribute to this code base? The answer is you ask Quad. If your question is like, how do I file like, you know, like an expense receipt? It's like, you ask Claude. If the question is like, what's the next company holiday? You ask Claude. So it's sort of like all this stuff that you used to have to do manually, you used to have to like go to someone, you used to ask Claude. It's like, it's just at the center of everything. And I think this is what we're starting to see with a lot of companies that are really getting it is they just put Claude exactly at the center of it. It's not like a thing on the outskirts somewhere. It's like you have to change all the business processes and that takes time. And it's a lot of change to figure it out. Absolutely. And I've been reading recently about what they call a Solow's paradox, which I think you basically just referred to, but it's this observation by an economist in the 80s that, you know, he said the computer age is everywhere except for in the productivity statistics. And the reason was, despite, you know, what at the time felt like a very large build out of computers, you were not seeing people get much more productive. As you just noted, Boris, like eventually those gains did materialize and it was because the companies had just reinvented their workflows around the new technology. And so the question now is, you know, how quickly might it take the economy to do that? I wanted to ask you a couple more questions on sort of like fine-grained software engineering because I've heard you say that coding is effectively solved. you haven't written any code in six months. Sometimes I see engineers pushing back on this idea and they say, you know, look, coding is not only about typing. It's also about judgment and taste and critical thinking and agents can still be quite bad at those. So what do you make of that critique? Like, are there parts of coding that remain unsolved or is that in your view, that's just has become something else? Yeah, I mean, the critique is totally right. I think this is one of those things that just gets like kind of taken out of context. So, you know, I, the, the full quote is coding is solved for the kinds of coding that I do. And like, for me, like I work on pretty simple code bases, you know, like what CLI, like the CLI is like a pretty new code base, like the desktop app and the mobile app. Like these are pretty small, simple code bases. We have so many enterprise customers, you know, like a lot of our customers now are like the biggest enterprises. It's not just like startups and indie devs anymore. It's like NASA is one of our customers. So it's like, you know, they have a really big, really complicated code bases. And so for them, like, it's not solved yet. Like, the model is still not perfect at it. It still makes mistakes. Its code isn't always perfect. And when you think about the kinds of stuff that engineers do, coding is a small percent of it. It used to be that it's like, you know, if you look at my day, maybe like 50% of my day used to be like actually like typing code. But then the other 50% was like talking to users. It was brainstorming and coming up with ideas. It was like debugging and kind of just thinking through how something works. It was this planning. There's just all these other things that engineers do that are not coding. So when I say coding is solved, it's like it's all for the kinds of coding that I do. And coding is just a small subset of what engineers do. And you actually see this you know for all the engineers at Anthropic and I think more and more engineers in the industry When the model does the coding they freed up to do all this other stuff that they actually enjoy doing a lot more Like talking to users and figuring out what's next. And QuadCode has been just 100% written by QuadCode for over six months. That's true for a lot of things at Anthropic. Beyond QuadCode, that's true for Cowork. It's true for a lot of other products. And we're starting to hear more and more customers like this. like um i was doing the talk for uh the latest y combinator batch this was like earlier this week we did a we did a fireside and i i used to ask at the beginning of every talk i do raise your hand if you use quad code now everyone uses quad code so i stopped asking that question and so instead of the question that i asked is um raise your hand if a hundred percent of your code is written by quad code and you know this is like the latest you know the most cutting edge startups but they're small startups usually it's like a few people um half the hands went up and then i was like okay raise your hand if uh none of your code is written by you know the the model and there's one hand that went up this is that ever like a couple hundred people and then everyone else is somewhere between they they were like between 50 and 100 so i think like is coding is starting to get solved for like a bigger and bigger percent of the code that we write you know like our team is our early indicator of what's happening in engineering engineering is our early indicator what happens to everything outside of engineering. And so we're starting to kind of see the shift and it started six months ago. And yeah, it's like, it's accelerating and we're starting to see more and more hands go up when they ask like, you know, is it 100%? Let me ask about another fear that people have about a world where the engineers aren't writing as much code. The fear is that people's understanding of their own profession will atrophy and that might be dangerous in various ways. you haven't written any code in six months. Do you feel like that atrophy has started with you? And how do you feel about it? There's a, you know, there's one engineer on the team, Lena, that was still writing like C++ on the weekends by hand just for fun. Because she like still enjoys like writing the code. And I think there's like, there's always room for this. I think for me, this is part of like a much broader transition and it's not about atrophy at all. It's just about like programming is always a thing that is in flux. like Mike Rampoff programmed in punch cards, like back in the Soviet Union, you know, like 70 years ago. And for him, that was programming. Like there was no JavaScript. There was no, you know, Python that didn't exist yet. For him, those punch cards was like piece of paper. There's a machine that like punches holes in it. Then you feed it into this mainframe. It processes it like a few lights light up. And when you talk about programming, like that's what it was. And then, you know, like before, before that, like, you know, like the Apollo program, it was like, there was a room full of people, you know, like often like women like doing doing math on on paper sometimes like by hand that was called programming right and nowadays you know like this this this changed like programming became writing machine code then it became writing assembly code then it became like javascript and python you know java all these languages that people use nowadays and now it's like changing again it's now you talk to the agent and it's actually i think about to change one more time where you you talk to an agent that talks to agents that does the coding. But, you know, it's just always changed like this. It doesn't feel like atrophy to me. It feels like a, it's like a sea change in the technology. My feeling about it has been that like, I'm sure that using a graphing calculator, like, caused some of my math skills to atrophy. But my solution to that is that I will just continue to use a calculator, you know, like, I'm sort of fine to seed some of that stuff. You know, now, if over time the calculator becomes super intelligent and tries to undermine me in subtle ways, like that would sort of freak me out. But, you know, maybe we haven't like crossed that bridge quite yet. Let me ask about another criticism that I sort of feel like is in this realm, which is it seems like, you know, every time a new model is released, we'll hear people say, this is really good. And then you check Reddit a few weeks later and they say the product has massively regressed. My sense that sometimes this is like a real issue caused by bugs. Other times, it's just sort of a vibe. But I feel like people are concerned that because it's all just sort of AI generated right now, there isn't maybe sort of the same craftsmanship that we once saw in code. So I'm just curious what you make of these periodic backlashes we seem to see. Yeah, it's actually sort of an open question what causes it. There's been a couple instances where it was real. And there was two that I know of, and we published like engineering, you know, blog posts on Anthropic blog about it. Because, you know, like if it's real, like we found it, we fixed it, and then we want to talk about it so people kind of understand exactly what happened. But I think in almost every other case, it's sort of like maybe it's like a honeymoon period where you kind of get used to the model. And, you know, at first it's magical and then you kind of get used to it. Maybe it's something like that. But I don't think it's really about craft because the model's code at this point is just much better than the code I would have written. If you talk to me like a year ago, I would not have said that. I would have said the model is like kind of sloppy and like the code's not really good. You have to like triple check everything. It can make silly mistakes all the time. But that's just like simply not the case anymore. And again, it's just the model keeps changing. It's really weird because every other technology we use does not change this fast. But there's a, you know, like if you tried the model for the last time, you know, like a year ago, the model now is completely different. And so if a year ago you had to handhold it and you had to triple check every line, Now I just generally let Claude do its thing. I ask Claude to double check the result. I ask Claude to open the app and test it by itself. And then while I do that, I have 15 other Claude's running that I've also asked to do stuff like this. But that's kind of what it is now. The code is actually just much better than what I would have written. Let me ask you about your Claude swarm then. You asked me earlier whether computers make me more productive. I think it seems clear that Claude is making you more productive, but it doesn't seem like it's actually reducing the amount of work that you're doing. And I think this is kind of like an important thing to dig into if we're curious, like what AI means for jobs, because, you know, it sounds like you believe companies are going to need fewer engineers. And yet, at least for you, like you're never running out of things to do. So like, how do you think about that? Like that, that question of, it's making me so much more productive. I'm not working any less. Will there ever be a case where making me more productive actually means I'm working less? Yeah, I mean, there's like a name of, there's like some paradox. I forget what it was, but there's someone that named this thing. Okay. Yeah, I don't know. I think that it's actually really individual. There's some parts of it where it's up to the company, because like depending on the business, there might be more need for people or less need for people. But I think actually a lot of it is individual preference. there was like um you know like when the laundry machine was released i'm gonna give like historical you know because you know for me i just this is this is a crazy technological change like i need history to kind of anchor myself no i love it i love the stories okay so this is just how i think about it yeah so like when when the laundry machine was released it i think the average person to do a load of laundry it took like five or six hours and you had to walk uh there was some estimate of like 3 000 feet or something from a load of laundry because like the way that it worked You have to walk outside and you have to, you know, like collect the logs and the colds and you go back inside and you start a fire, you blow the water, then you take it out. You put in the laundry, you like scrub it on the scrub board, then you have to wring it out. And then you might have to like repeat this for like your entire family, maybe like every day. And it was just like a lot of work. And at some point the laundry machine appeared and it just took it down like a lot. I think it took like three hours off the time it took to do a load of laundry. And this was one of the factors that let women enter the workforce. in mass, like without this, like usually it was the women of the house, like not always, but usually doing this work. And then this means like you were just like stuck at home and you couldn't do anything else. But now it's like three hours freed up every day. And so like different people could choose how they want to spend this time. And maybe for some people, the choice was, well, I just want to like hang out with my kids or like, I want to go like, you know, walk the dog or read a book or hang out with my friends. But actually for a lot of people, the answer was, okay, I'm ready to enter the workforce. Like I want to go like work at a factory or like I want to go work in an office job. And because the time was freed up, now you have this choice of what you want to do with this time. And I think it's kind of similar right now. It's like it's similar for any technology. It gives you more choice. Yeah. Hmm. A couple of last questions about the software engineering. I've been asking all of our guests if a 22-year-old just finished their CES degree this month and came up to you and said, okay, now what? What do you say to them? Like, is there is there an entry level job that is out there waiting for them? Or do they need to, like, think differently about this, like, first part of their career? My advice would be if you're a person that wants to work at a big company or you want to work like at a company, you can totally still do this. Like there are entry level jobs there. You know, there's a lot there's a lot that you can do. But actually, if you're at all entrepreneurial, go start a startup. It's just like there has never been a better time in history to start a startup. Like it's just absolutely the golden age. It's like you and your agents can build a giant company. Like people are building like billion dollar companies. And it's just like a few people. You know, like for Cloud Code originally, it was just a few of us. And, you know, we have so many customers where they're building like really big businesses and just really amazing startups. And it's just like, you know, like one or two or three people. And one person with the right idea has so much leverage. I just couldn't imagine a better time to go and do it. That's interesting because I feel like a lot of the view that we get from the AI world is like, you know, model capabilities are advancing so quickly that like, will we even have companies in five years? But like you think that, you know, at least for the next bit, there's still plenty of room to start a company, get into business, make a product, all the rest. At least for the next few years. I mean, if you really trace out the exponential, it gets really weird. I mean, there's like a version of it where like the idea of jobs kind of doesn't make sense anymore. The idea of companies doesn't make sense. The idea of software doesn't make sense. But, you know, as you trace the exponential, it just gets weird. But I think in the meantime, there's just, there's so much to do. And, you know, we're all just here kind of figuring out what the model means and what this thing can do. So might as well be one of these people that's exploring the frontier. Yeah, all right. Last one on the engineering. If three years from now, do you think we will see more engineers fewer engineers or will it be impossible to answer because we just might not be calling them engineers anymore um okay I think I think okay let's define it yeah um I don't think we're gonna call them engineers but if we talk about like people writing code or people like using agents you know like people using Quog to write to write the code I think there will be a hundred times more engineers than there are today that's my prediction wow okay all right super interesting uh well this seems like a good time to return to Quog co-work which you brought up earlier and i know that you helped to develop i would say this is the anthropic product that i actually use the most now um i use it as a kind of editor on my columns i use it to help me produce a podcast i've been using it as a kind of financial advisor uh basically you know you just create products you add some skills and it can kind of successfully imitate lots of different you know roles in in a workplace um also as a non-technical person i just find the UI, very intuitive to use because it mostly just involves like dropping documents, you know, into a box. Talk to me a little bit about the road ahead for co-work. And in particular, I'm curious whether you think it can solve for other jobs the way that coding is maybe now partially solved. Co-work is just so exciting because we first started to build it when we saw people like abusing quad code essentially for things that are not coding like you know like someone like installing the terminal so that and and like opening up their terminal installing quad code in the terminal so they could do like their tax returns like this is like crazy it's like that's not what a terminal is for but it's like amazing from product point of view because like people really want this yeah um i i think the journey the next few months is going to be figuring out like how do we make this work really well for people that are not engineers and this is something actually kind of new for us because, you know, most of our team works on coding and, you know, now part of the team works on co-work and we're trying to kind of figure this out. For coding, it's actually pretty easy because the people building this are engineers. And so we just kind of build for ourselves and it's really useful for everyone else. With co-work, one of the things, one of the harder things has been it's useful for just like everything that's not engineering. It's like it's useful for accounting and, you know, finance and also for legal and also for, you know, like I used it to like buy it like a clamming license so I can go clamming with like, you know, like the state of Washington. I used it, you know, like I said, to like book flights and like concert tickets. It kind of does all this different kind of stuff. And so how do you make sure it's actually really good at all this stuff? And so I think the biggest thing is like we use it all day, every day. We talk to customers all day, every day. And then, I don't know, we like, we just, we just do the things that, you know, a lot of customers are asking for. So I would expect it to just like keep getting better at this stuff. I would expect it to keep getting better running for long periods of time. And this is something that we saw for coding. Like a year and a half ago, quad code could run maybe like, you know, 30 seconds without me having to interrupt it because the model just wasn't there yet at all. It just would go off the rails after 30 seconds. And now, you know, like I run it, I run quad for just hours and hours and hours. And every night I have like, you know, hundreds or sometimes thousands of agents that are just running 5, 10, 20 hours. and this is just how engineering is done now. And I think the same thing will happen with cowork. I don't know like what the kind of work is where you would want it running that long, but I think we're going to figure it out because I think people are going to start to like knock down the doors and demand it because like they really want this. It's an interesting question to think about. Like honestly, it makes me feel a little bit insecure because if you're like, Casey, like what's a task that you wish Claude could work on for 20 hours for you? Like I would sort of flail a little bit, you know? It doesn't take 20 hours for me to like report and write a story or make a podcast. But, you know, then again, I think once that becomes available, maybe I, you know, figure something out. So maybe that's just sort of brings us back to the capability overhang that you were talking about earlier. Yeah, I mean, in some ways, it's so hard to predict the future because this technology is so weird. But I think once you actually use the technology and like you, like you use it every day, you kind of feel like you know what's missing because you feel it. So it just becomes more obvious. I think if you asked me that same question like a year ago, So like, would I have run an agent for 20 hours? I would have been like, no, that doesn't make any sense. I have no idea what I would use it for. But now I do this like all the time. And, you know, I suppose one thing that it can do, maybe that it doesn't today, is just like anticipate your needs, right? Like as it develops a better sense of who you are and what you do, as its memory improves, it can probably do a better job of guessing like, okay, like Casey, you know, you got four more episodes of this podcast. You haven't booked, you know, the final two guests. I'm going to brainstorm a list of those guests. I'm going to draft some initial reach out emails. Like I'll put them in your drafts folder. Let me know if you want me to send them Like these are the sort of things that i could imagine you know something like a human producer or co doing for me and i can imagine a world in the not too distant future where something like co-worker is doing something similar yeah yeah it maybe it also kind of depends like how how do you define a task like maybe there's like a horizontal way and like a vertical way like maybe one way of thinking about it is like a task is everything related to design um or everything related to engineering or everything related to you know, like finance. But maybe like another way is like this kind of more vertical way. And like something that I experience every day now is I use Quad to do something. Let's say build a feature. And then, you know, Quad will like, you know, build a feature, it'll test it and then it'll merge it. And then it'll kind of like launch the feature for me. You know, like maybe something in Quad code or something in cowork. And then something that Quad started to do with Opus 4.7 is it started to become a lot more proactive. And I think I've started to see it do is it'll launch the feature and then it'll be like, okay, great, I'm going to schedule a reminder for myself in 12 hours so I can see what the user feedback is. And so if there's any bugs, I can go and fix that. And this is something I would have had to do and remember to do and I might have forgotten to do it. And so when Claude does this, it's just delightful. So I'm like, okay, thanks, Claude. Thanks for thinking ahead a little bit. So yeah, there might be a little bit more of that. I also want to toss in a question about Claude Mythos. This model is not available externally yet to most people. What we have heard about it from you all so far mostly concerns coding and cybersecurity. But I imagine that some version of it will come out before too long, and it will probably be good at stuff beyond coding. So I wonder, like, in your opinion, is Mythos relevant to the jobs question? Like, does this feel kind of like another step down the road of, like, the jagged frontier being able to do more work? yeah I mean it's like another it's another point along the exponential this one I think this one's a little further along than most of the jumps we usually make this one is a particularly big jump but yeah it's like you know Mythos was particularly strong with cyber and with coding and maybe there's some future version of it that's you know strong and other capabilities also yeah well as we start to like zoom out a little bit and maybe talk about the you know implications of a world where more jobs are disappearing due to automation or maybe just transforming quite a lot. I think you've said in the past that you do think that this transition is going to be painful for a lot of people. Ananthropic is in a unique position here, right? Like potentially it will be a source of unemployment among software engineers or people in other jobs. Does the company have an obligation to those people? Is that something that the government needs to be paying attention to? Or what do we do about this world that seems to be coming into being? Yeah, like I said, I think it's going to be mixed like it is for any technology. There's going to be good effects and there's going to be bad effects. And we don't know the exact timing. We don't know the exact mix. You can just never predict it in the moment. I feel like I do feel this pretty immense obligation, just like as an engineer, that, you know, there's always more that we should be doing to tell people about kind of what's coming, make sure they're able to use the tools, kind of educate them and bring them along into the future. So I actually feel this very strongly. And it's something that, you know, like the team and I actually talk a lot about. I think that, you know, broadly, this is not a problem that we can solve. This is bigger, you know, than one company. And you actually like really don't want one company to solve it because it could be the wrong solution. So I think like this is a society-wide question. It's something that we should be talking about. It's something that we should be debating. And I think the thing that Anthropic is trying to add to the mix is we put out economic reports. We talk about policy and we generally just, you know, try to make it really obvious what we're seeing so that everyone else can decide what we do about it. Yeah. I mean, I do think the number one thing that has gotten people to take these issues more seriously is just improvement in the quality of models, which I guess makes sense. You know, like when it's just purely at the realm of the theoretical, people have a hard time thinking through, OK, what do we do next? But then, you know, once you like see a computer use itself, for example, I think a lot of people have that moment of like, OK, it's it's time to develop a strategy here. Yeah, yeah. And this is this is, I think, like one of the reasons why we build product at Anthropic. You know, like we're we're AI safety lab. It's actually kind of weird to even build product. it's like really not obvious like why we should be like doing that at all but actually one of the biggest arguments for building product and this is like actually a debate like early on in anthropic days and one of the biggest arguments was we want people to experience it so that they can understand it so that they can kind of play a part in figuring out like as a society what we should do but if you like keep this technology locked away so you know no one knows what it can do and no one can experience it it's much harder for people to form a point of view about it yeah i also wonder if you've thought at all about like the let me let me start that one again our uh our most recent guest on the show is james mannika from google who's kind of studied um like technology at the level of economies and whole societies and he's really worried about what started out as a kind of digital divide so not everyone having equal access to technologies like you know the internet or maybe a good laptop and he's worried that that's about to transform into an ai divide and it seems like the data that we've seen so far shows that the people who are getting the most out of ai are the ones already sort of like near the top of the income ladder um from your perspective like does cloud code make that better or worse like who do you see actually using it and not using it and are there efforts to try to you know maybe get it into the hands of um people who haven't always had access to these kinds of cutting edge technologies. Yeah. So I think there's a couple of programs to do things like this, to kind of like just spread access and Anthropic does this. I think the thing that I've been just continuously surprised by is the people that get the most value out of cloud code and use the most are just not at all the people I expect most of the time. Like we just did this hackathon for, you know, Opus 4.7 release. And the people that won the hackathon are not actually professional engineers, largely. There was like one person that was a there was an electrician uh there was a there was a doctor there was a carpenter like you that that used it to build an app and we actually saw the same thing with our last hackathon with a 4.6 hackathon before that i think it would have been like mostly engineers but i think now the models are sophisticated enough that it's actually like not even engineers a lot of the time that are learning how to like really harness them and this is a thing that we're seeing at kind of big customers also. So like as companies think about how do you adopt AI tools? Obviously you have to think about how do you do this kind of like business process change? How do you put quad at the center? This is like the biggest question. And every company is approaching it kind of differently. One of the ways that I've seen work the best is you just give everyone tokens. You make everyone feel safe experimenting and the ideas will come from the people that you don't expect a lot of the time. It's not actually the like super senior engineer that was like the most productive in the past. Actually, the best idea might be like an accountant somewhere in the corner of the org, or it might be like a GTM person somewhere in a different corner of the org that like built some amazing internal dashboard that just like sped everything up a lot, or, you know, like solve some important problem that like no one even realized the business had. So I think actually like this is, this is the future. And like, this is the thing that we're going to start seeing a lot more of, and it's going to keep being surprising. And so like, yeah, it's, it's important that people learn how to use the tools because it's not necessarily the people that are the most productive and the best with the tools of today that are going to be the best with the tools of tomorrow. All right. Well, we've talked a few times so far about how difficult it is to think through what life is going to look like in an exponential. So I'm going to try not to ask you to predict out too far. But if we had you back in a year, how much of the world of, let's just maybe keep it to software engineering, will look very recognizable and how much would look a little crazy to us from today's perspective a year from now, is there a world where you think we do start to see some of that automation kicking into gear, some more big layoffs at companies where they're citing AI as an example. What do you think we're going to be looking at? I think there's going to be a lot of disruption in the next year. I think there's going to be a lot of big players that try to figure it out. and I think a lot of them will be successful. I think we're going to see some like traditional business modes go away. You know, like there's all these different modes that companies have and some of them are here to stay, even despite AI. And then I think some of them are going to matter a bunch, you know, less because of AI. And, you know, so like a mode like network effects, that's actually like not going to go away. Like if you have like a, you know, an app that gets more valuable with more people being on the app, it doesn't matter like who's building the app or, you know, whether, you know, AI exists or kind of is it an important force or not? Like that's, that's still like just as important. If you think about like scale economies, so like, you know, as you manufacture the, the marginal cost of goods kind of goes down over time. There's like kind of a natural advantage your business gets from that. And I don't think that's really going to go away, but actually other modes are going to go away. And so this is, for example, like switching costs. So like, you know, you're on vendor A and you want to move to vendor B, actually quad can probably just move you from A to B and like the, you know, the switching cost is not a giant moat anymore. So I think businesses that depended on some of these moats that are going away are, you know, they're not going to do well. I think a lot of them are going to figure out new moats. And actually, if you look at like all the big businesses today, like the biggest companies, they actually have a number of moats because this is something they think about all the time. Like, how do we have a defensible business? So it's actually like not that new to them. And then I think, you know, some companies are going to continue to do well. And I think there's just going to be, if I had to predict something that would be surprising today, it's that there's going to be a lot more innovation than we expect. I think there's going to be a lot of new ideas coming from not big companies, but tiny startups of like one or two or 10 people. And I think the number of these startups is going to explode. There's going to be so many new startups exploring these ideas because again, just like the leverage of one person is going to be insane. Like at this, um, at this talk at a Y Combinator, there's a bunch of companies working on a bunch of different stuff. There's one startup working on like, material discovery. So they had this deck and they were talking about how there was the Iron Age and then there was the Stone Age and there was the Iron Age and now we're in the Silicon Age and if we can discover the new material, then this is how we get into the next stage. And this is something where as a tiny startup, there is just no way you would have been able to fund this years ago, but now they're just using Quad to kind of discover this and kind of like scan over all the possible molecules and kind of ways to design these materials. And a team of just a few people can maybe make this kind of breakthrough where, you know, 20 years ago, no way. Right. So the companies may be smaller, but maybe some of the ideas are going to get better and we're just going to see rapid growth among these small startups that have some kind of insight. That's my bad. It's like someone that knows the domain really well and has an army of clods can just do so much more than, you know, what they could have done before, even if they had like an army of humans doing this. if you have sort of sold one message consistently on this podcast Boris it has been buy as much Claude as possible at all times for all things that's sort of what I'm getting coming across but it makes sense for the creator of Claude code final question for you we're just like so this is like hard to explain I work at Anthropix so I have to kind of sell Claude but I think actually in a lot of ways we just use Claude for everything and there's just this there is this divide between kind of the way that we use it and i think the way that everyone else does and so this is also kind of part of the job is you know to be like actually like you put cloud at the center and it can do totally new things that you didn't expect i mean it speaks to what you said earlier which is that um if you want like ai to make you or your company more productive you probably need to redesign your workflow around it and not just try to staple it onto some existing process but you know and i think that there's evidence that that that is true last question for you in addition to everything that you are doing with cloud code and co-work you are also amazingly omnipresent on x and threads i see you every day troubleshooting users problems offering tips on how to use claude if you could automate that part of your job would you and how close are we to that world i have automated it but i prefer to do it myself okay All right. Okay. I actually, the way that I did it is I have like a loop set up in quad code. And now I've actually moved it to a routine. And it just runs every 30 minutes. And Threads has an API, you know, X has an API. And so it just, you know, uses it. And it's really easy to aggregate the feedback and look at it. But actually my favorite part of my job is just interacting with people. Even if they're saying something is broken or something doesn't work or it could be like 10 times better. That's still my favorite thing because that lets us make the product better. and there's this thing in product where sometimes people look back and they're like, wow, this was like a moment of genius or, you know, like this product was, you know, like someone conceived of it and just like built it and it was like perfect and that's just like never the way it ever works. And, you know, like Cloud Code has so many flaws. It's so far from like being the product that it could be. And the only way to make it better is to listen to people, especially when they say something doesn't work and to like keep improving it and keep making it better. and this is actually the way that awesome product is built and this is the only reason that quad code you know it's still so far from amazing but it just it improves a little bit every day and that's the reason i mean it strikes me that's also a really like human part of your job right like when you're talking to somebody who is using a product that you made it's probably reminding you why you started doing it in the first place right and i feel like those moments of connection like feel really important at a time when we're not always sure what our value is going to be that we're bringing to like an AI enabled workplace. Yeah, yeah, that's right. And yeah, like this is all, we're all trying to figure this out together. This is sort of the thing, like we have some hypotheses about where this is going. There's a bunch of stuff that we're building because, you know, we think we know where it's going, but, you know, I'm actually often wrong. Like not all of my guesses are right. Often good ideas come from, you know, all sorts of people and, you know, you never know. So you just have to listen and you have to kind of try a bunch of stuff and sometimes it works. All right. Well, Boris, thanks so much for joining us. Yeah. Thanks so much, Casey. to rovo.com to learn more today.