Claude Code's Creator Reveals "Claude Cowork"'s Setup

0:00Claude Cowork is here and if you understand how to use it you're going to be able to outperform 99 % of people on this planet. It is an easy way for you to use Claude Code. You might have heard of Claude Code, it's gone viral, but the problem is it feels technical. You have to go into the terminal and it's not fun for a lot of beginners. So Claude has come up with Cowork and it's a brand new product that harnesses the power of Claude Code in a UI that is simple, that anyone could use, that your dad can use, your mom could use. Hey, even you can use. So in this episode, I brought on Boris, the maker of this product.

0:36I'm so excited that he shows you the best practices for how to use Claude Cowork and at the end, how he sets up his Claude code to get the most out of it. You're going to love this episode.

0:59We are lucky. We've got Boris. He's the creator of Claude Code and I would say the co-creator of Claude Cowork today. And today what I want to get accomplished is everyone's talking about Cowork and I want to best understand what are the use cases? How can we get started? What are some non-obvious ways I can use it? So, Boris, thank you so much for coming on the show. And I have one question for you. by the end of this episode, what are people going to get out of it? Ooh, I think people are going to start to get some more ideas for how to use cowork. And hopefully they're going to tweet about it and maybe they'll even reach out to me so I could learn how they want to use cowork.

1:42It's funny. I feel like, um, for quad code from the beginning, it was sort of built to be, it wasn't actually built to be a product at all. But when we first started thinking about it as a product, we thought that the thing people would use it for is coding, but we've learned very quickly that people didn't just use it for coding. They used it for like all sorts of stuff. So I kind of feel the journey of quad code has been very surprising. And I have been learning so much just watching how people use the product and how they abuse it and kind of like what they actually want to use it for. And even if it's not designed for that.

2:14So I kind of feel the same way about coworker. You know, I have some ideas about what people are going to use it for. I kind of view these as hypotheses. And so, you know, happy to talk about it. Happy to be able to do a quick demo to show you kind of some of the things that we use it for. But I think that it's going to be pretty surprising. And I hope that I will be surprised when we see how people actually use it in the wild. Yeah, I think that's always the case with platforms, especially if you think about it. But when the makers of the app store created the app store, the initial apps was a beer drinking app and a bunch of random apps like that.

2:51Did they know that Uber was going to come out of that? DoorDash was going to come out of that? TikTok was going to come out of that? Probably not. So I think that what's really cool about the phase that we are in right now with Cloud Code and also Cloud Cowork is we're all figuring it out at the same time. And none of us, including the creator, have all the answers. But I agree with you. Like what's it's a good time to be sharing some, you know, sharing what's working, what's not working. And, you know, if you're open to it, let's let's screen share and get our hands dirty. Yeah. Yeah. Let's do it.

3:26So what are we looking at right now, Boris? Yeah. So this is the this is the cloud desktop app. So you just download it. It co-workers only available for Mac OS. You know, Windows coming soon. There's a few different tabs in the desktop app. So there's the chat. That's the default. well, there's CoWork, that's the new one, and there's code, and that's just QuadCode. CoWork, under the hood, it's actually just QuadCode. And so, you know, the agent that makes QuadCode awesome, we call it the QuadAgent. It's also available as the QuadAgent SDK, so you can use it programmatically. You can, you know, all sorts of companies build all sorts of cool things on top of it.

4:03We actually use that same exact SDK directly in CoWork. So it's like, you know, it's pretty cool. It's just kind of one layer across everything. We have the best agent, have the best agentic model, might as well use it. And so what I'll do is just to kind of show how to use this thing. And, you know, like when I think about agentic AI, this word agentic has sort of lost all meaning because it's just like used so much. So I feel like probably a lot of listeners have heard the word agent, but they don't actually know what it means or, you know, they think it's like some like cool AI or something.

4:37but it actually has like a very specific meaning in the ai world which i think has kind of been lost because a lot of the products that people have released and called agentic in the past are not actually agentic and so like when you think about the ai products that everyone's used you know like obviously quad code uh chat based apps where you know you just chat with the with app kind of back and forth send some messages the biggest difference with agents is uh it can take action and you know it's not just text and it's not just like web searching but it can actually use tools on your computer.

5:07It can interact with the world. And so for Anthropic from the very beginning, since before our models were good, like before like, you know, like Cloud 3 or whatever, this is a thing that we wanted to get really good at because we felt that it's very important. And so from the start, we wanted our models to be really great at coding and then really great at tool use and then really great at computer use. So it's kind of cool, like, you know, the last year, seeing how people have been hacking quad code, it's pretty obvious this is kind of the place that we should go. And so for people that have used quad code, you know, none of this will be, you know, like too surprising.

5:45These are actually things that you can do. So really what we're trying to do is make this something that everyone can use in a way that's safe. And so what I'm going to do is, you know, here on my desktop, I have this like receipts folder. I have a few receipts in it. So I'm just going to give co-work access to my desktop. And you have to pick like which specific folders they can see by default they can't see anything so you have to kind of opt in to see to let it access specific folders and so i'm gonna say um you know i can say uh i have a receipts folder can you rename the files to match the dates on the receipts so i think one interesting thing is when you're using co-work it's really like it's operating with your files like that's that's a big sort of mindset shift that people should have it's almost like your operating system is that right yeah exactly it's like it has your files so the ones that you give it access to um but actually the even cooler thing is it can use all sorts of tools um and so like files actually is like you know that's like that's useful but it's not it's not like that cool what's actually much more interesting is it can generate files for you.

7:01So, you know, it can make like presentations and things like this. It can interact with any tool over MCP and it has built-in support for Chrome based browsers. So it can actually control your browser to do, to do stuff. And so I'll, I'll kind of show that a little bit. So this is kind of the first step. When you're first getting started with cowork, the thing I recommend is, you know, do exactly what I just did, you know, dismount the folder, give cowork access to it and just like play around with it. It's super useful for cleaning up files, organizing, things like that. And so here it found these four receipts that I have.

7:35It's asking me if one of the receipts, I guess, is missing a date. So should I just rename the others? So I'll just say, you know, for one, it's, let's say it's like up to you. And then for two, I'll say, don't rename it. and we actually call this uh like reverse solicitation in the ai world so so what this means is when the model is unsure about something it's going to ask you for clarification um and we've sort of taught the model to be pretty good at this so instead of assuming if it's unsure about something it's just going to ask you and so yeah in this case so it renamed the receipts so i'm just going to open this up to double check yeah cool and so the receipts were named a little bit better organized.

8:23And so maybe what I can try next is, let's like put this in a spreadsheet. So it takes control of your computer, basically, in that sense, right? If you allow it to take control. Yeah, yeah, that's right. And so we put like so much work into kind of safety and making sure that as this happens, you don't like accidentally shoot yourself in the foot and like delete files or whatever. So there's just like a huge amount of work that went into this. and it kind of starts at the model side where you know for anthropic from the very beginning we were the ai safety lab and that's the reason that we exist um and so there's a lot of work into like alignment and mechanistic interpretability and kind of all these ideas to make sure that the model does what you want in a way that's safe kind of at the model layer and this literally means like studying the neurons kind of the same way that you would study uh neurons and you know like in the human um and so we you know you can like identify structures and you can kind of study in a very scientific way as a black box also to make sure that it's safe.

9:25So this is called alignment. And then we do a whole bunch of other stuff. So there's actually a whole virtual machine running under the hood. And this is just to make sure that any actions taken are safe and don't affect your broader system. And then as of last week, there's also deletion protection. So if you accidentally delete something, then you're going to get prompted first. So the model can kind of make sure that that's actually a thing that you want to do. Obviously, also, as we start interacting with the internet, something like prompt injection is quite scary. And so we built in a lot of protections against that.

9:57Obviously, you know, it's not perfect and it's something we're iterating on. But this is also part of the reason that we released this pretty early is we want to see how people use it. And a big part of making models safe isn't just studying them in a lab, but studying them in the wild to see how it's useful um and so this was like pretty cool right so we have this folder um first it kind of renamed all all my receipts i asked it to make a sheet it just made this spreadsheet so it's already here um maybe i don't want a spreadsheet maybe i want like a google sheet so actually that's so interesting right because like i think uh where people stumble with tools like co-work and cloud code is because you can do anything making it a of course that makes sense making it a google sheet but like that's where it's so interesting that you can you really can treat it like a teammate in that sense right like go and do this thing the world is your oyster whoa what is happening right now yeah so it's like it's opening the browser um and so here it's gonna ask me for for permission just for this demo i'm gonna say always allow for the site um but you can just say like allow once or deny.

11:05And so quad's taking the wheel. Um, it's making a spreadsheet for me. Um, this is going to sort of take a while. And so this is one of the things that we're iterating on is like making this kind of computer use a lot faster. I remember like, man, this is like when I first joined Anthropic and this was maybe, um, this like Sonnet 3.6 days or I forget what the, what we called it, like Sonnet 3.5 new. Um, and that was the first model that I think we really started to crack computers and i remember i was like sitting with my team and there was literally a researcher that ran into the room like in the movies and was like oh my god quad knows how to use computers and um we just had it like order a pizza and it uh it picked a pineapple pizza and then ordered it and it asked us for a credit card and then we got that delivered to the office but it was tedious it took like an hour or something for it to click around and since then we've been improving the model's ability to use a computer.

12:00And so you can kind of see that here. So here the, you know, quad is typing and it's interacting with the spreadsheet and it can see what's on the screen. It can interact with it. Anything that's in your browser, it can just use. And could it use like email, you know, let's say we wanted to send this to somebody, you know, finance team, like, can you, maybe we'll get into it later, but you know, with things like connections and mcps um is that possible yeah yeah totally okay so let's like let's make this like kind of a nice spreadsheet because we don't want to send like a badly formatted spreadsheet to our co-worker um but let's make it nice and then uh we can ask quad to send it and so here it's kind of cool it's like you know it's still early days and so there's some like formatting mistakes it didn't paste it exactly correctly but it noticed it so noticed that you know it's not actually split correctly and so now it's trying to format it um so kind of notice this and yeah greg like to your point it's funny these like these use cases and the way that people are going to use these tools like i said at the start it's going to be so surprising um i'm just so excited to see how people use it and sort of the crazy thing about them is they're so general purpose it's like it's sort of like a computer itself or like the internet or something like like you said Like when you first like had an iPhone, like you would never have predicted that, you know, there would be an Uber app at some point.

13:20But, you know, that's that's what happened. And I sort of feel like we're at the beginning of that. But for for agents. Yeah, I think it's it sounds so obvious in hindsight. Like, of course, you have a phone with GPS. Of course. Right. You're going to you're going to have things like Uber and DoorDash and stuff like that. But I feel like, yes, you're right. I think what I'm trying to figure out in my mind, I'm watching this and I'm like, this is really cool, but I'm thinking about, okay, how do I audit my entire company? What are all the tasks that the company is doing? How are they interacting with files?

13:59And how are they using the internet? And how are they sending things? And then what are opportunities I can use to make my team and also my life more productive? That's what's going through my head right now. Yeah, yeah, totally. That's a great way to think about it. And by the way, something that's kind of cool is, you know, as this is running, I'm just going to make a new task and do something else. So, you know, what are some cool episodes? Startup.

14:37cool and so while this one's running we can like we can let this one go too and i often have a bunch of tasks running in parallel too but yeah that's exactly it it's like you should just think about like what what's all like the tedious stuff that you do every day and you can just throw all the stuff to co-work um and i i feel like this happened for coding over the last year because of course, engineers and programmers are the earliest adopters. So when the tooling first became able to do that kind of work, engineers adopted it first. But now this is possible for everyone else to use too. So that's super exciting.

15:15And when we talk to engineers about the way they like to use quad code, they run a bunch of quads in parallel, they use it to automate the TDS stuff. And I don't know, man, for me, this is just the most fun I've ever had as an engineer, because I get to do the stuff that I enjoy and I just feel so productive because Quad does all the stuff that I didn't want to do. Okay, so now we have the spreadsheet. This looks all right. It's not too bad. The data looks correct. You know, one thing that's missing though is, well, we can add maybe like the totals, but I think for the sake of the demo, what I'm going to do is I'm going to see if you can email it.

15:56So just reading for audio listeners, can you open Gmail and send the sheet to Amy?

16:04And wow. Yeah. So I'm going to say always continue. So how is it going to know it's Amy, right? It's going to pull up the contacts within Gmail? Yeah, yeah, exactly. Yeah, that's exactly it. It's going to interact with it the same way that, you know, anyone else would interact with a computer. And it's the same thing. It can, like, click stuff. It can read the screen. Okay. So, compose window open. Let me type Amy. So, you have, like, multiple. When you're, like, locked in, so to speak, you're kind of, you have multiple of these windows open at, or multiple virtual machines open at the same time.

16:45Yeah, yeah. I usually have a bunch of these, you know, like five or ten or whatever. And so we can see the models trying to interact with this. And so it filled out Amy. And you know, by the way, just because this is going on YouTube, you know you're going to have some haters in the comments. So we're just going to address them right now. Someone in the comments is going to be like, but that's so slow. I can do it faster myself or something. What is your reaction? Of course you can. Yeah. I could definitely do this much faster. And it's actually the same thing for a lot of quad code, right? Especially at the beginning, I could do it much faster.

17:22But I think sort of two things happen. One is that the model just gets better at doing it quickly. And this is just something I would expect over time. But then the second thing is because you can do multiple things in parallel, it's actually a big time saver. And so usually my workflow is I'll kick off a few different tasks in parallel. And I'll just kind of go back and forth between them and kind of tend to my quads. make sure they're in a good place see if they have any questions or anything like that so it's really a different kind of workflow like i feel like now is the age of like kind of multi multi-clotting of of parallelism of like not going super deep on stuff but kind of being more of a generalist and more tending tending to your clods and so like this was this other chat that we kicked off earlier um and in this case it sort of did some research so that's pretty cool um so it didn't just go to you know like startup ideas podcast website but it searched the internet checked a bunch of different places checked like someone's notes about it yeah 10 10 rules for for quad code yeah i'll find that interesting too and let's see so it seems like this is the email is probably all drafted the draft has been saved please send and i also use this for like slack too like a pretty common use case i've been doing maybe a couple times a week is i have a spreadsheet where we track all the team's work for for the month or for the week or whatever and instead of having to bug everyone on the team to fill out the status what i do is i ask co-work to look at the spreadsheet any column that's not filled out, just message the engineer on Slack.

19:02And it does that really well. So, you know, I just ask it to do that. And then I go get a coffee and I don't have to do that anymore. Do you think that co-work is going to be the gateway drug to CloudCode? Because for a lot of beginners, non-technical people, CloudCode feels a bit overwhelming. But when I'm watching you do this, this feels, you know, approachable. Yeah, I mean, I think it's like probably two kinds of users. and it's also it's funny Cloud Code is built originally in a terminal and you know nowadays it's available in the Cloud mobile app it's available on the website in the IDE, on Slack, on GitHub so you know nowadays it's available everywhere but originally it was built in a terminal and I never thought that most engineers would want to use a terminal because you know it's scary it's like this thing that's sort of hidden away on the computer like you know only the most hardcore engineers want to actually work in a terminal day in and day out And so it's surprising to me that most engineers wanted to use it.

20:01What's even more surprising is a lot of non-engineers started to use it. And that was the craziest thing. And, you know, looking at like the sales team at Anthropic, like the GTM team, half of them use quad code every week. And looking at other non-technical people at Anthropic, like designers, product managers, data scientists, pretty much all of them use quad code every day. So that's been like pretty surprising. But my hypothesis is what they would prefer is actually something a little bit more like this. where you don't have to deal with a terminal and all that kind of thing. You have a nice UI because you don't need access to bash and all that kind of stuff because the model will just do it for you.

20:38Yeah, I think what I would love with something like this is, you know, automating some of these processes. It's like instead of doing one-offs, it's like anytime this happens, I want you to do this. That's my dream. Yeah, it's a cool idea. Yeah, I'm excited to see like how it's tied into skills. Also, I think it's something where we're thinking about. We haven't quite figured out yet. Yeah, because like a skill, what is it? It's essentially kind of a repeatable way to do something. And so you saw that a little bit earlier up in this conversation where when Quad was generating this spreadsheet, I don't know if you saw, but it actually loaded the skill.

21:17And so we prepackaged the skill for Excel. And that's the way that Quad knows how to do it. and so if you have some like weird file format i don't know if you work with like autocad or if you work with you know like sales for whatever it is the tool the tool that you use just make a skill and then quad can can do it for you that's interesting so if you have more skills co-work and you're asking it you know to do things related to those skills co-work is more likely to use those skills so you might be able to get a better outcome out of co-work if you you know do some upfront work to actually create some skills i think so yeah wow yeah i think i think that's exactly right yeah okay well that's good to know and and you know what else do we need to know about co-work because i know there's you know there's you know extensions there's their skills like walk us through some of, is there anything else that, you know, we need to know in order to get the most out of, of the product?

22:23You know, we, we tried to build coworker just to be pretty simple. When we think about the different audiences we serve for an audience, like engineers, they don't actually like simple, like simple is, is good by default, but engineers love to hack their tools. They love to customize their tools. So that's actually really important. So quad code is just like the most customizable coding agent. It has just like so many extension points. There's skills, there's custom agents, there's hooks. There's an insane amount of configuration and settings that you can set. There's a very sophisticated permission system.

22:56So there's all sorts of ways to customize it. And this is because we know every engineer works different and everyone has different preferences. Everyone uses a different tech stack. Everyone uses a different editor. Everyone uses a different OS. So we wanted to make sure it's really customizable. For some of my co-work, I think we're kind of starting in the opposite direction. and just keep it really simple at the start. And so if you use Cowork, I would probably not customize too much. Like if you have Cowork installed and you install the Chrome extension, that's pretty much all you need and it'll do everything else for you.

23:27I think over time, as you find yourself using maybe software that Cowork is not great at, like that's the point to think about writing skills, but that wouldn't be my starting point. Just start simple, like, you know, see how it works, see what's useful. And I think also the other thing, you know it's just it's so early this feels to me a lot like quad code a year ago where you know we released it before it was early it was not very good yet it was like super buggy it kind of barely worked like the model was not very good at coding a year ago and it kind of feels like that to me but I think our work is actually a little bit better than quad code was when we first released it because it's actually useful and you know it's already a thing that I use every day and um yeah Yeah, I think in the first week, we've seen so much more growth.

24:13It's been like a multiple of the growth that we saw with QuadCode the first week. So that's very exciting. It seems like this is something that people are already finding useful. You know, we're recording this, what is it, January 2026. If we come back here in 12 months, it's January 2027. How do you think people are going to be using CoWork? You know, I know you don't have a crystal ball exactly, but what sort of use cases and And what is what does the product look like? Like describe the world in January 2027 with with co-work. Oh, geez, Greg, I plan in like a one week timeline. The model is just changing so fast.

24:54It's so hard. And I just feel like, you know, like the model is it's advancing exponentially and just like my puny human meat brain, like can't grapple with exponential. It's like we think in linears. And so I think this kind of exponential is just very, very difficult to plan around. Okay, but if I had to speculate, a year ago, I made, and I think Dario also made this prediction that by the end of the year, people wouldn't be writing code anymore. And I think that was like sometime mid, you know, last year or something. in you know i i code every day uh i ship you know two three hundred prs every month something like that uh and in the last two months quad quad code has written 100 of my code i haven't written a single line by hand and this is something like i also predicted this you know like way back middle of last year it was sort of not intuitive because if you just think about kind of the experience at the time and you trace it linearly, there's just no way that the model would be at that point.

25:56So you really have to kind of believe in exponential and just literally like plot it out and kind of follow the way that the line goes. That's the only way that you would have predicted this. And it was just, it was absolutely right. And so I think for coding, this is something that we're going to start to see in more and more places for more and more kinds of code that the model is able to just do all of it. And when we think about co-work, I think it's somewhat similar. I think it's a little bit earlier. And I think what we're going to see is for all this kind of tedious tasks, like, you know, like connecting app A and B or kind of shuffling data back and forth or whatever, the model is just going to be able to do it.

26:32And it's going to get increasingly good at it. This is, I think in some ways, it's a little scary. And in some ways, I think it's really exciting because you don't have to spend your time on this toilsome work anymore. You can just focus on the work that you enjoy. and also everyone I think just becomes much much more productive because you have an army of clods that can do this. Okay I like that feature I like that feature a lot it's a hard question it's a hard question when you it's sort of like the genies out of the bottle and it's hard to predict like where the genie is going to go so I believe that co-work is the gateway drug to clod code.

27:11I think that people are going to start using it and they're going to develop vertical use cases for whatever it is their businesses. You mentioned AutoCAD. You probably weren't thinking about AutoCAD when you were helping develop CoBalt, CoWork. So I think there's going to be these verticals and I think that CoWork is going to be similar to like vertical job boards where there's very specific roles that you can hire to people. And I do think that there's going to be this combination between skills and these like digital teammates that you're going to quote unquote hire. Yeah, that's super interesting.

27:54Maybe we should make a bet and just see where it pans out in a year. You know, I have no idea, but it's definitely interesting to speculate because I just feel like the way this technology goes is so different than past technology waves. it's sort of similar to the internet it's sort of similar to computers you know like maybe like telephones or something before that but the speed is just so much faster and because it's kind of piggybacking on all of this like the internet could not exist without telephones uh telephones and kind of like phone lines being everywhere you couldn't have dial up without it you know mobile phones couldn't exist without the internet existing um so it's sort of like every layer of the stack it just gets more and more powerful and it spreads more quickly and so it's like on the back of all this that ai can exist um and yeah it's just gonna be very exciting to see to see where it goes i want to shift gears just for the last 10 minutes um because we mentioned cloud code you had a you had a post that went absolutely viral i think you know what post i'm going to talk about 99 000 bookmarks which is crazy and i'm going to share my screen i want to go over it a little bit.

29:02This is me learning to use Twitter. I think I made an account like a decade ago and I haven't really used it. It's funny. It's been fun to learn. You have just a few tweets, but this one, you figured it out. You say, I'm Boris. I created Cloud Code. Lots of people have asked for how I use Cloud Code so I want to show off my setup a bit. My setup might be surprisingly vanilla. Cloud Code works great out of the box, so I personally don't customize it much. There's no one correct way to use Cloud Code. We intentionally built it in a way that you can use it, customize it, and hack it however you'd like.