Q4 2024 Anti-Work All Hands

Awesome. And I'll put it— I'll tweet it out. Awesome. Good morning everybody, or afternoon, or night, wherever you happen to be. We are here for the Q4 2024 Antiwork All Hands. Lots has happened since the last one. Last time it was called the Gumroad board meet— public board meeting. So, uh, lots has happened, uh, including, uh, we changed our name, uh, of the company. So Antiwork, uh, Antwork.com is, yeah, some people are snickering. Why are you snickering? Grab the mic and explain yourself. She refuses. Do I need this or no? It's better. It is better? Okay, good. Yeah, so Antwork is, some people may be familiar with the subreddit Antwork, which is basically where people complain about their employers. And I like that idea. I like kind of thinking that the future is basically like everyone is like infinitely rich and like you can kind of just predict the future basically in that way. Like everyone, you know, how does like an infinitely rich person act? They probably don't have an employer. They probably hang out in Bali. And I think anti-work is kind of a good mission. It's the mission of the, actually the tagline of the subreddit is Unemployment for all, not just the rich. And so that's kind of, I think, a really nice through line for what we're trying to do and what we're trying to build and why we continue to work. And this is actually all AI-generated, but this is our new website. And so, yeah, this is kind of like the more public-facing tagline. Gumroad started, was our kind of our first product. It's our biggest product. It drives effectively 100% of snaps currently. It was about helping creators spend 100% of their time doing what they love. Creators like to make things, and if we can make it really easy for them to make money and make it really sort of trivial in terms of how much time, money, and energy they have to spend doing that, that they can spend more time creating, and they sort of the anti-work portfolio. Expands that to a bunch of different categories of people. So entrepreneurs, business owners, freelancers, engineers. Can we kind of replicate that, but for all sorts of different categories of people? And the goal, the sort of pseudo-goal, we'll see how serious we are based on if we actually accomplish this, is to effectively get every single letter of the alphabet. Not a really original idea, alphabet. Which is technically the name of the company behind Google, also kind of did a similar thing. But I think it's like a nice, it's like a nice framing instead of just trying to like squeeze all the juice out of a single thing unnecessarily in a way that maybe like maybe piss creators off because you're trying to do all these things that like part of it is just like a simple, we can kind of go more horizontal and build new stuff. So we have a bunch of products I'll talk more about. If you actually want to see these slides yourself, you can, you can go to antiware.com/2024/Q4. And the slides are actually a little Next.js app. So this is something we built, also generated with AI. Pretty, pretty cool. There's a theme, there's a theme here and using AI. I also want to talk briefly about the logo. My dream job is a logo designer, which is not a great dream job because like how many logos do you need? Also why maybe having 26 projects is interesting to me. So we can have 26 logos. We have to come up with 26 brands. I think I'm a pretty terrible engineer, slightly better designer, but I really like to think I'm good at brand. You know, hashtag, you know, it's like the meme. There's a really good Onion article. One nice thing about doing everything here is I can just show you this stuff. Personal brand. This is a very— one of my favorite Onion article. You think it'll get it? I have to say The Onion. Worm. I love this. I love this Onion article. I'm a brand. Everyone's trying to build their brand. So this was written in 2012 around the sort of the, you know, social media taking off. And I am my own product, the little worm said while staring at a laptop and depressingly shuffling between his Twitter, Facebook, LinkedIn, Google, Google Plus. You guys remember that? I don't know, I just think it's a, it's very funny. It's good to be self-aware. But yeah, I want to talk about the logo because I love logo design and I think it's a really good one. AI also helped me come up with this idea. And there's a bunch of meanings. The first meaning, anti-work. So I think there's sort of like the, sort of like the kind of communist uprising view of anti-work. But then there's also the physics terminology of work. What is work? Work is force times distance. So if you want to move an object, you know, 100 feet, that requires work, requires like real energy to be expended. And so anti-work is kind of like trying to reduce work, trying to reduce like physical work, right? Like Gumroad, for example, is really cool because you can take a class on the internet. You can just open up your laptop, you like all the work that you have to expend to move your body to another location, to go to college or go to class or commute or all these things. You know, we can provide tooling to sort of make that less work physically. And so I think a really good sort of metaphor for work is climbing up a mountain. And so I like this logo because it's actually, it's actually like an anti-mountain. It's an A cut out of a W. So there's a mountain there, but it's like, it's in the negative, the negative space, which I think is pretty cool. And the A is also anti, so the A is actually negative. And then there's also fangs, which I like. It's pretty cool, kind of like subtle fangs. So, you know, we're trying to kill a bunch of work. Or kill a bunch of competitors, deplete profit margin. I think there's kind of a nice— and it's also kind of friendly. It's not, you know, it's not like a scary, you know, it's like kind of a cartoony set of teeth. And then the last thing that someone brought up is it's— I think it's really important to take one step back to move forward. You often have to like like a new mountain every once in a while, you know. You have to like, uh, you have to decide that you like, you're at some local maxima, but then you have to go back down to Earth and then go back up another mountain that's much better and taller and has better view. You have to enjoy that. You have to enjoy the process, uh, the journey over the destination. And, uh, and so this is kind of that. It's, uh, it's one step. In this case, it looks more like forward, but imagine there's there's two arrows pointing one way and one arrow pointing the other way. So I think it's a nice reminder that you, you know, sometimes you have to go backwards in order to go forwards. Ideally, the goal is to still go more forward than you're going backward. But I think a lot of people and companies as institutions only go one direction. They get larger or they get bigger or they get slower. There's more technical debt. There's generally not less technical debt. It's very hard to sort of change your mind as an individual and then imagine how how hard it is to change your mind as like a group of people. Very, very, very hard. And so yeah, I think it's like, it's really, it's really nice. Just like the name, I think it's nice to have these like ideas that are kind of controversial, like embedded, like unavoidable. Like they're like in our, they're all, you can't, like if you have to apply to a job at a company called Antiwork, you have to acknowledge some aspect of the almost hypocrisy. Of the present moment. So yeah. Oh, the other one that someone also brought up was it's like car headlights. You can imagine a car right here. These are the two headlights. And that's also, I think, pretty nice because I have no idea what next year is going to look like or what 10 years from now is going to look like. So it does feel a bit like driving a car with bright headlights in the dark. So yeah, it's kind of a cool, cool logo. Pretty, pretty, pretty happy about it. It's the first logo I've ever designed that I liked, and I've been trying to design logos since I was 13. Actually got my start designing logos on State Point Marketplace, and I never won a single competition. It was like the 99 designs of the year, and Yeah, it was very sad. So then I had to learn how to code so I could like get a job. So it goes, so it goes. Yeah. And so, yeah, so, so yeah, another, another way of looking at what we're trying to do is basically transforming work. I'm not anti-work on a universal principle. I'm anti the work that we do today. The work we do tomorrow will be a different kind of work. We can also be anti that work when, when we're there. And then maybe one day we'll live in a world in which there's truly no quote unquote what we would consider work done, maybe at least by humans or conscious beings. Conscious beings can spend their time being fully creative, doing what they really like doing. I really like designing logos and coming up with names and brands. I don't like fixing bugs. Any of you guys like fixing bugs? If you do, we can artificially introduce them into the into the, uh, into the software. Uh, you know, the AI will be introducing bugs and the humans will enjoy fixing them. Um, whatever, whatever the right relationship may be. Um, so yeah, that's kind of Antiwork. That's kind of the new holding company. Uh, so that's one big change we made in the last 3 months. Another big change is I wrote this article a few years ago, No Meetings, No Deadlines, No Full-Time Employees. You guys like the animation, by the way? Gonna do it again. It's no longer true. This is a meeting that we're having. We do have deadlines. This meeting is happening at a certain time and place. And we do have full-time employees too, so far. Me and Connor over there. He's very excited to be featured. And yeah, so I think it's kind of also just important to like very publicly say that, like, we don't follow this anymore. And the reason is actually in the last sentence of that little intro of that essay. Also, by the way, what's really cool is that this is actually text. Pretty cool, right? When you have everything in code, you can actually do all sorts of crazy stuff. If I had more time, what I'm gonna do with the next one is I'm gonna actually add content editable to all this. So I could, like, make changes, like, as I, you know, if I want, but just delete Can't do that if it's just like an image that you've exported from, from Figma. So yeah, so yeah, so this, this last, this last bit. Oh, that's weird. We're growing 85% year over year when I wrote this article, you know, COVID, we were kind of in the heyday of COVID creator economy company. Just really easy to, when you have a product that's, you know, super old. And you kind of know what you need to do, you have like a kind of a roadmap, a design system that's very established, you can kind of work in this like super asynchronous fashion where effectively we had this big Kanban board in Notion, which you can, if you wanna read about this, you can, the essay's still there. I'm not deleting the essay from the internet. But we had this big Kanban board and things would go from scoping to design to building, and then we'd kind of repeat that cycle. And that was fine, but it did, and it worked very well actually. Helen is featured in that article as she shipped memberships, government memberships, but we didn't have a single meeting. Like, pretty crazy, like the complexity of what we were able to do, and we still can do that if we wanted to, but I think just with the way things are going, we need to move a lot faster and we kind of need to do those same steps, scope, design, build. Kind of at the same time. At the, you know, instead of over a sort of 90-day window, we should be doing it over, you know, like a 90-minute sort of window. And that kind of requires like a lot more synchronicity and you can get a lot more of that in person with an office. I also wrote another essay earlier this year, which is kind of like an update, I guess, called God Mode. And it sort of references— you guys, originally, have you read Founder Mode by Paul Graham? Or at least, like, you think you have because you've seen enough tweets about it. Paul Graham wrote this thing called Founder Mode, which I guess I could just click and it'll work. So here it is. It's Honestly, not very, uh, I mean, it's really short. Uh, there's not much really in there, but he basically just talks about Airbnb and, uh, how Brian at Airbnb gave this talk at Y Combinator. It's like an essay about a talk about a guy who was inspired by another guy. It's like very indirect. Um, basically he's just like a Steve Jobs in view of building stuff, which, uh, is, I think, is kind of basically micromanagement, which is a word that people don't like, but basically just means that everyone who cares about stuff should be able to like see that stuff happening and it's kind of like 100% observability. And so I kind of wrote this article about trying to be more specific basically, like, okay, if there's like a, if there's something there in which this article is clearly resonating with people, but there's not actually a lot of depth to it. And that's also kind of interesting. It's like, why not? I think it is because it's sort of things like micromanagement are generally not favored by most people, which is why most people don't start and run companies. There's a lot of like unfun stuff you have to do in order to build like the best products on Earth, which is what I want to do. I want to build like the best products on Earth. Pretty cool that we get to do that. And so GOD actually stands for Growth, Optimization, and Destruction, which if you're familiar with like Hinduism, which is relevant because we're in Bali, which is the one island in Indonesia, which is a Hindu majority. Uh, it's, uh, basically you have 3 gods. So there's kind of a godhead similar to Christianity, but it's 3 gods. Uh, one god is responsible for sort of like the creation of the universe. I think Vishnu is their name. Uh, one god is, uh, sort of responsible for like the persistence of the universe. Uh, Shiva maybe? Brahma. Awesome. Uh, and then, uh, and then one is responsible, uh, for destruction of the universe. Uh, that's Shiva. All right, Shiva tricked me. Uh, and I, yeah, I just, I really like that. Uh, I think in sort of the Western startup VC, it's like there's a lot of hero's journey, there's a lot of this like, uh, this, uh, sort of up and to the right thing. I actually have a little whiteboard, so do this. But basically it's just like this, right? Like, beginning happens at the end, right? And this is like kind of like startups and IPOing and, you know, Star Wars and all that kind of Western sort of ideology. And then there's like this this more sort of Buddhist, Shinto, Hindu kind of like growth optimization, destruction, growth optimization, destruction, growth optimization, destruction. I think there's like a synthesis of these two ideas, which is kind of what I'm aiming for. I want the business to do better and better and better. I want us to have more products, but it does require like a constant— or not constant, but you know, you're, you're always kind of And as you zoom out, this will end up basically looking like the above chart. You remove all the noise. It's like when you, you know, you get in the plane and you're like, those cars look so smooth on the road. There's just like, just can't see the bumps, you know, every car just looks like— and so the goal, you know, I feel like, I don't know, at the end of the day, we live in a cool time where we have a diversity of ideas and trying to incorporate them into a single into a single methodology, sort of a universal principle of building stuff, is kind of what I was trying to do here. But the most important part is the destruction piece, I think. The growth and optimization are kind of obvious. Like, you go to Harvard Business School and they'll teach you about those sorts of things. You can read Zero to One. You know, it's called Zero to One. It's not called Zero to One to Zero to Two to 1.5 to 7. You know, it's not— it's not called that. And yeah, I think the sort of the tricky part, the thing that is hard to do is destruction. I kind of think of it like surgery. Surgery is really hard because you don't actually know, like, potentially the ramifications of what you're— what, you know, what are the sort of second order, third order effects of doing surgery? That just basically requires like a lot of knowledge, a lot of knowledge about a system in order to make a change. If you add something to a system, Right, like for example, like let's say we added a wall to this room. Pretty easy to do. Moving it's a little bit harder, but removing a wall, like what happens if I remove this pillar? I don't know. I don't have x-ray vision. Even if I did, I don't really know. Like it's, it's, there's a lot more risk sort of associated with the sort of destructive aspect. And it requires a lot more knowledge, which I think in the context of gods, like omniscience, right? Like you kind of— the ideal sort of system is one in which everyone in the company ideally understands it. And that way you can make really, really good decisions. It's also why the offices work is because it's sort of a way of replicating that omniscience where you have everyone in the same room. It's kind of like a big team brain. And you can actually do those sorts of things. You can design, you can scope, excuse me, scope, design, and build in parallel. And that leads to really, really, really good products at the end of the day, because I have no idea how good my design is until I actually have customers using it. And so, you know, trying to speed up that feedback loop, just like this, you know, allows us to grow, to grow faster, and build better things. So just in the last 90 days. These are all the things that we destructed at the company. We got rid of PayPal, kind of controversial, but it was a pain in my butt. And they actually kind of were annoying. Like they basically like banned our account for a day. And then I was like, oh, I guess we could just ban it forever then. So, you know, self— so yeah, so we got rid of PayPal. We are basically 100% on Stripe. We're getting rid of all our CSS. There's no CSS as part of this presentation. I mean, it is, but it's kind of like outside of the company. We're just using Tailwind. And so it's just Tailwind is just— when I say CSS, I mean like the .css files. It's just HTML. And actually, it's not even HTML. It's actually TypeScript files. It's like, you know, rendered by JavaScript. So effectively, we're trying to like move everything to a single language, where we were trying to delete all of our own components. We're starting to use ShadCM, kind of outsourcing that. We got rid of Figma. We deleted our Figma account basically. And we, we, we moved— I mean, designers can still use Figma, right? Like, I'm not saying you're like, it's illegal. But, but I'm not paying for it. And I think it's generally like, you want to set really strong defaults of what you think is kind of the right behaviors. Like, I think of Figma now as a kind of a scratchpad. Right? Just like people have notebooks, there's no company notebook that we all use. It's, everyone kind of has their own thing and when it's worth sharing, you can share it and get feedback and all those sorts of things. And because we're kind of a really public company, we, we, we basically think the right way to use Figma is, is publicly, not privately. So if we do have a design system in Figma, we can still do it. We still have it in the company account. It just should be publicly accessible to everyone. And that's really helpful because then AI gets to learn from it, makes the AI better, and then we can use the AI to actually ship our own products, which is pretty cool. Zoom, we also got rid of Zoom. We now use Google Meet like this. I personally stopped using Chrome. I switched to Safari because it's the default browser I got with my laptop. I was like, I mean, I already have a browser. So I started using Safari again. This is Safari. Notion, got rid of Notion basically, started just relying, and this is kind of annoying honestly. There's, there's some pain by the way associated with all these things. There was no pain, then everyone would be doing these sorts of things, which would not be good for us because then we'd have a lot more better competition. But luckily we don't. No, so we got rid of Notion. We moved to Slack and Apple Notes is what I use. For myself, and it's way, in my opinion, it's better. Obviously Apple has like gazillions of employees, and there's all sorts of, you know, there's all sorts of really cool stuff that you get with that. It's also completely free. Buildkite, we deleted our Buildkite account, which is our CI. We moved to GitHub Actions. Iffy, which was this internal product, another nice sort of approach similar to the Figma thing is we can still have it sort of internal. We can still build stuff, right? We just make it public. So we took our internal IFFI app that was basically our risk tool and shipped it so you can now use it yourself. And that makes it a lot better. We also started, started using Stripe Radar. We already used Stripe. We used to use a thing called Bill.com. We switched to using Mercury. Wells Fargo, I deleted our bank account and moved to Mercury. Wells Fargo, by the way, was a pain in my butt too. Like, they would like require, like, I mean, think about the technical debt of these, some of these companies. They would like, I had a physical key that I needed to like log into the bank, which is like, I guess how probably people listening are like, I still do that. And my recommendation is to switch, switch to Mercury. It takes like, you know, 5 minutes. It's not that hard. Elasticsearch, Transcoder, S3 service. Or sorry, AWS service, or AWS, I guess, technically. And let's say ATM machine. And we have this new thing called, just internally called Gumroad Media Converter. We may also do a similar thing, buy a nice fancy domain, like a design one more logo and sort of offer it to the world. Nice thing with software, it's like very low maintenance costs. And we also deleted, almost effectively deleted support@gumroad.com and we're basically doubling down on in-app live chat, which I think is way better, is faster. Does anyone enjoy writing emails? No. I don't. I, I hate opening my inbox and seeing 30 emails I have to respond to personally. No, generally you wouldn't ever need to talk to support because there's, it's like, Gumroad is perfect and does everything you want it to do 100% of the time. If it doesn't happen, it's generally because you tried to do something. And so I think the best way to approach that is like, we should, one, detect if you've done something that, you know, you wanted to do something and it failed, and we should kind of like auto-open the in-app chat. You should just be able to just be like, hey, why did this happen? And then we should be able to tell you and actually like help you, or an AI will be able to kind of help you and assist you in real time. We also got rid of our help docs. I think help docs are kind of a pre-AI phenomenon. We went to this play in Bali and they gave us like a sheet, a handout sheet of like, you know, the playbill effectively. And Sean was like, "Why can't they just make the play better?" And like, more under— like, couldn't they just— couldn't the— I'm paraphrasing a little bit. Like, why do we need a study guide for this play? What did you say? Why do we need a study guide for the play? Yeah, why do we— yeah, it's just a study guide for a play. And I was like, well, that's kind of like our help docs. You know, like, ideally our product is so good that you don't need, you know, in the case of like an API, like the product is the documentation. So I think it's a little bit different. Like if it was just the Playbill, that was the product. Different, right? But if there's like a— just if you're sort of just like supporting something else, ideally that thing is just like subtitles, you know, while you're watching, there should just be Subtitles of what the frick is happening. Who is this monkey man? And why are these 3 other monkey men chasing him? And then he like always somehow shows his genitalia. Very big thing here. That's universal culture thing, I think. And then to be disrupted. So we still have a lot of stuff to do. Sort of optimistically, I think we can get all these things done in 2025. But we'll see. We're gonna get rid of Swift and Kotlin. So native apps. We have an iOS app and an Android app for Gumroad, and we don't currently have any mobile apps for the other products we've built. And I think that's largely because they require completely separate codebases. It should be kind of like designing for mobile. You should just get, get it for free. You shouldn't have to think about it. Actually, we should be mostly probably be designing for mobile first because way better to do things on your phone. Most people don't scroll TikTok on their laptop, right? Like, so generally people are addicted to their phones. They're not generally addicted to their laptops. And so I think we want to get people addicted to our products to build better ones. The way to do that is to embrace— and a lot of this was also kind of like impossible. Even arguably today in 2024, it's impossible. Expo. Kind of unlocks a lot of this stuff, React Native, Electron. And there's gonna be enhancements to all these things. There'll be new things that beat them out. We'll switch to those eventually. There's always room to keep going and delete more and more and more stuff. Ultimately, the goal is to like delete everything. There's nothing, nothing left. We're all just living life. And every time you want something, you just get it. We have teleportation, we have telecommunication, we have telepathy. Uh, everyone just knows anything they want. And why would you use the internet if you just knew everything, right? Like, why, why do you use the internet? Why? Like, I, I find it interesting, like, when I use, uh, the internet on my computer, like, what am I— like, what, what? Uh, it's like, oh, it's because my friends aren't next to me. Well, what if they were, you know? I think there's sort of like Pareto superior solutions to a lot of technology. I think the internet is an interrupt technology. People will— it's going to be fun to tell like our grandkids about like the internet. Like, yeah, we went like, just like we typed http://antiwork.com. It's like, what? Why com? Oh, com is short for company or commercial. Why the dot? Why not just Antiwork? It's like, oh, well, that's, uh, uh, never mind, you know? Uh, so yeah, I'm very excited for the, for the future. Uh, AWS we're also going to try to kill. Uh, we're going to move to Cloudflare, Vercel, and a head server. Vercel's really, really awesome, uh, and basically, uh, allows us to effectively outsource a lot of stuff that we've done internally. Uh, in at Gumroad because we had to, because Vercel didn't really exist when Gumroad started. Otherwise, I'm sure we would've built on that right away, 'cause I'm lazy brand designer pretend engineer. So obviously that's perfect for Vercel. That's like their ideal client, people who pretend to know how to code. But soon everyone will be able to do that really well with AI. That's ideally everyone is a pretend engineer. We all pretend, right? None of us really know. Really know how programming works on a deep fundamental assembly level. Right? And so yeah, I'm super excited about that. And then Ruby on Rails. Ruby on Rails is a form of technical debt, in my opinion, and we have to act like it. We're gonna open source Gumroad, so we're gonna contribute to the Ruby on Rails ecosystem, but eventually we're gonna move to the entire app being written in TypeScript. Which I'm excited about. I think maybe I changed my mind. Maybe something else wins. Maybe it's Rust, who knows? But I do think Ruby on Rails is a form of technical debt, not because it's not awesome. It is awesome. That's why we use it. I love Ruby on Rails. I love lots of things that have technical debt, like books and pen and paper. But they're, you know, and horseback riding. We went horseback riding. And I was like, oh, this reminds me of when I used to, like, type code. I think that's what it's gonna be like. You're gonna, like, go somewhere and, like, you'll go to Bali and there'll be, like, a little, like, 7 PM to 8:15, we're gonna, like, go to this studio and, like, type, you know, for fun. So yeah, just preview. We're also moving primarily to in-person. So that's a really, really big change. Also very scary, but I woke up one morning and realized I was having a lot more fun and I was being a lot more productive. And work is frankly, lots of parts of work are not fun. That's why we're anti it. And it helps to, it's like going to the gym. I don't know. I think people work out in their bedroom. I always think I will like buy, you know, the convict conditioning ebook. I'm like, I'm just gonna like do 100 pushups every day. And never do it. Instead, I go to a place, uh, that's like built for the function and the purpose, and my brain's like, oh, you're here to like work out, you know? And then you do it a lot better and longer than you would have otherwise. There's a reason like offices and temples and churches and gyms and schools exist, um, otherwise everyone would just chill at home all day long, uh, and then people would go crazy because we need to hang out with other people and We kind of— someone needs to force us to do that. So I'm doing that. We're not going to go fully in person because our team is really awesome and many of them don't want to. So what am I going to do? So yeah, the goal is about 10 people in person sometime next year. We have 3 or 4 people now in New York and then 20 remote. I think one advantage potentially of remote, uh, we'll see how it manifests. It's like, you, you know, humans have to sleep. So one nice thing about remote is we could potentially move a lot faster if we have like, you know, the kind of 24-hour, you know, if you're working 8 hours a day, let's say that's sort of 3 shifts. So the idea of sort of parallelizing, you know, scoping, designing, and building 24/7, uh, is pretty appealing to me. I think the most important roles to bring in person will be the sort of the most high variance So like design engineering, you know, it's, it's, it's like the sort of the front of that sort of iteration cycle. Systems is probably the thing that is probably more, most friendly to remote. However, I do think it's going to get harder and harder and harder probably to justify it. It's like AI can do more and more stuff generally. Like, you know, if AI can, and if I can just ask Claude and it gives me an answer in 3 seconds, it's kind of hard to You're now competing with like this, this thing that didn't really exist before, right? Everything changed in my, you know, June 25th or whenever I started using Claude and Cursor and I was like, holy crap, this is, this is kind of insane. Like I'm just bottlenecked by my own energy at this point, my own ideas, which is the best place to be, right? You don't want to be bottlenecked by typing, you know, that's weird. Like I can't get this thing done because I have to type it all. Why? It's, you know, yeah, it's weird. The other nice thing about having an office is I can do whiteboarding sessions. I love doing these. So this is just one of them that we did when the Vercel kind of team came over and we talked about how we wanna switch and do this, a lot of these technical rewrites. And at the time we had these 4 products, Gumroad, Flexile, Helper, and Iffy and sort of the backend, the frontend. The S, which stands for servers maybe, or styles, one of those two, maybe both. That's why I just did it one time. But yeah, you can kind of see that, like, you know, the newest products generally have the least technical debt. And you can kind of just infer from the decisions we're making with the newest stuff, what the right quote-unquote answer was. It's kind of like if you want to know if you need to break up with somebody, you just have to ask yourself, like, if I met them today, would I want to date them? Very scary to ask, but very important. Every day I wake up and I rehire everyone at Gumroad. Scary. And that's, you know, That's how you get rid of technical debt. And so, so yeah, so if you know TypeScript, fully TypeScript, Next.js for sale, shout out CN, et cetera, kind of like the sort of quote unquote perfect picture. There's still technical debt. There's still a lot of janky stuff in the ecosystem. I'm not saying it's perfect. Helper, we actually made progress on this. So Helper is now fully in TypeScript, I think, right? It's pretty insane how fast we're able to do that thanks to the humans and AI. And a lot of other stuff. And Flexile, we're working on Rails. So Helper was Python, now TypeScript, and then Flexile was Rails, is Rails, and we're moving to TypeScript Next as well. We did try TanStack for a bit, but I just, I basically felt like we were just replicating all the Next behaviors anyway. So it's like, why not just use Next and double down on it? One less thing, you know, one, one less, one more thing to disrupt, one less thing to think about. Uh, actually Guillermo from Vercel, he has like a good concept, uh, which is basically you wanna minimize the number of concepts you need to understand in order to ship. Uh, like if there's every, and every single new thing, I think of it like if you get a job at like, there's a reason the McDonald's menu is very simple. It makes it really easy to train new people. Uh, it makes it really like every, every, imagine like every every new thing they would have to add to the McDonald's menu, like how much complexity that would add in terms of the supply chain. Every country is different. They have to source it. They have to teach people how to produce it. The ice cream machines are always broken. You know, it's just, it's a lot. So reduce the number of concepts, I think, is like a helpful exercise. Like we don't have to think about assembly. We've kind of abstracted a layer on top. And my thesis is that that effectively TypeScript is, and Next is kind of just like a new, you know, on Tailwind, like this new layer of abstraction on top of, just like we don't think about assembly and CSS now, we're like adding new things to the things we don't have to think about. We have, it's sort of 100% error corrective proof. So you actually don't have to think about it. We never need to go to assembly. It's perfect. Like the translation layer is effectively perfect or maybe it's not, but somewhere it's imperfect. And those people get to think about that sort of thing. You have to work at like NVIDIA or something to do to do that kind of work. And I kind of think that basically everything is just gonna be a TSX file or a TS file, and even, you know, no HTML, no CSS, right? We're producing those things just— and those sort of gets all this stuff gets sort of compiled all the way through to assembly. And but, you know, these languages should be, should be, should be good for, like, you know, being humans being able to make things. Ultimately, It's probably just English, right? Because these are all just abstract, like the, you know, like I should just be able to say like, hey, like give me a slide. I mean, this is how I talk to AI, right? Like I talk to AI through English. I don't type code. I say, hey, give me a slide, 50/50. On the left, title is Welcome to New York. This is the content that I pasted from my Apple Note. And then on the right,, put this image that I'll fill in and make it so that when I hover over it, it shows like the whole image. And that's kind of fun. Uh, I have no idea why, like what it chose to do that. Is it, is it CSS animations? Is it Framer Motion? Now renamed Motion because boom, one less concept to learn. You're like, what is Framer? What, what? No, it's just Motion. Awesome. Way better. Uh, world is getting simpler and simpler. So, so exciting. Yeah, it's— you don't have to think about it. The beauty of software is you can use the product, right? So it's like if you— if it does what you said you wanted it to do, then it works. Like if it's too slow, then you do have to like look at the code and figure it out, or if there's a bug or whatever. But if it's actually working, then I think it's totally fine to like forget and not have to think about like those layers of abstraction. Even like I look at a keyboard and it's like, why? Like, why, why QWERTY? Why ASCII? Like, why not ABCD? You know, I don't get it. Uh, you know, it's just, it's all gonna go away. Uh, it's gonna be fun. I don't know exactly how, because kind of annoying if you're filling an office just like people talking to the computers all the time, but, uh, I'm sure we'll, we'll figure that out. Yeah, one step back, two steps forward sort of thing. Awesome. So these are our products. This is what we're working on next. Lots of stuff going on. We're not just destructing stuff. We are doing things. There's another nice thing. I was able to combine 5 slides into 1 because I can make it interactive. Pretty cool. So we're going to open source Gumroad, as I mentioned, and we're going to have this very new business model behind it. Which is we're going to collect— it's a licensing model. So we're going to, we're going to collect 1.5% fee. So if you imagine Gumroad is currently 10%, effectively, if you self-host it, you can save 8.5%. What is that going to do? It's kind of scary, but I think it's the right thing to do and we'll see. So the way it works is, yeah, after you make a million, but it's actually completely free. Once you make— start earning a million dollars a year, quite significant, then we'll basically just send you an email. Ideally, an AI will send you an email and say, hey, you owe us 1.5% of fees. How much money did you make in 2024? And, you know, click this button, fill out this form, type in your, you know, bank credentials, etc., and wire us the money. And effectively, it's like 100% gross margin business for us because they're doing the hosting, they're doing the support, they're doing the risk, they're doing everything. Maybe they're even using our products, hopefully, to do that. And they can just focus on what they like to do, which is probably not, you know, run and think about millions and millions of lines of code of a sort of a marketplace software. We're gonna open source kind of like progressively. So we're gonna open source like the current Rails stack and then also the fully TypeScript and people actually be able to see all these people that pick different sides of the debate can just see, oh, this is how much simpler it is or not. Maybe I'm wrong about certain things. And then also this is great because AI will effectively be able to learn, you know, if you really love Ruby on Rails, if you're like, oh no, Ruby on Rails is amazing, it's the future. Well, open sourcing government will help make that true, right? Because one One of the big issues with Rails is there's just not that many open source code bases. I think I might be the third or fourth largest Rails code base that's open source once, once we do them. I think Mastodon and Kickstarter, or not Kickstarter, Diaspora, I think are the two that I know of, but you know, Shopify or GitHub or Stripe are massive Rails code bases, but they're not open source. The nice thing about TypeScript is kind of runs in the browser. So like inherently it is open source. Just kind of default, which is pretty cool. And then you have v0 and Claude and React and Claude artifacts and all this stuff has really, really made it, you know, basically we have like AGI for like building Next.js apps. Now we're just working on AGI for everything else. Pretty cool. I also think that we'll probably have a new compensation model associated with working on Gumroad where basically, and it's not that new, other people do it, We're gonna probably use Flexile to sort of improve the UX, make it really, really awesome. But there'll be some sort of bounty reward system. Performance-based compensations, kind of like the business school term for it. But basically, you know, this task is not really worth us doing it, you know, like translating this from, you know, old CSS design system to new Tailwind component would be nice, but doesn't really maybe there are better, more creative things that the core maintainers should be working on, which is what everyone at the company then becomes in this world. And so, yeah, I think we can in theory have hundreds of people make money. And ideally, because of Flexile, not just make money like, oh, I made $100 or $1,000 or $10,000, but actually get dividends. And I'll talk about that later. But you're getting equity and you actually get to choose your split. So the way we work at Flexile, you get to choose your split and/or we'll offer a similar sort of model for every contractor, every contributor, sorry. And so imagine you do work and you get $500 a month or $500 a year for life. I think that could be a really, really cool way to, that maybe, you know, a lot of people talk about what the end of software, like, why do we even need any of this software? Why would anyone pay for software? Why would I pay anyone to work on software, etc. I do think these, you know, we're progressively learning the answer and trying to build, build, build the answer as we go. So yeah, Flexile, as I talked about, a lot of stuff we still don't really even have product market fit for. It's one reason we're coming back to the office is we need it. And, and yeah, the biggest bottleneck is literally not even self-serve. So you can't sign up for Flexile. All this amazing stuff. I can, I can talk about and people can listen if they want, but at the end of the day, if they can't like go to flexcel.com and do it, then what's the point? So we're gonna do that. That's a big focus for us. Helper is something we started working on a year or so ago, which was basically like RAG, if people are familiar with that. Actually, by the way, fun fact, I published the first RAG on the internet. AskMyBook.com was the first. I, I, I, uh, I— and the story behind it is that I, I, uh, was using Playground, OpenAI Playground. A friend of mine— at this point they didn't really allow like a lot of usage, uh, it was just— this is actually pre-ChatGPT, uh, November— that was October of 2022. And, uh, and so I was using it because my friend was working on this app called Sudowrite, and he's like, hey, You tried like this new AI, like GPT-3, really good. It's much— yes, obviously GPT-4 is way better, but GPT-3 was really good. And I was like, oh, how do I use it? And he's like, oh, you have to go to like playground.openai.com and like type in this prompt. And then it just, it just like gives you more, like it just produces— it's like not a human-friendly sort of approach. And, and I was like, oh, that— and then I started messing around with it and it was really, really good. And I basically, I was like, oh, this can like basically make it so that people can have a conversation with my book. That's really cool. Like, why would I write a book? Why can't I just have a chatbot that like answers everything on my behalf? That's like way better. Why would, you know, in that, for that specific use case. Books are still nice, you know, as paperweights and stuff. Just kidding, I like books. Most attention. But yeah, but the issue was that there's a context length problem. So you could only put like, I think at the time it was like 1,000 tokens or something. So like basically like a few pages worth of prompt stuffing. And so I DM'd this guy that followed me who worked at OpenAI and I was like, hey, like I can solve this problem. The only thing I need is like more context length. And he's like, that's hard. Research, more of a research thing than a product change. So it's going to take some time. But, and then he sent me this Python notebook of basically how they did RAG. And it's just, it was just in GitHub. It was like how you, it basically, the example was like asking a Wikipedia page about the Olympics, you know, questions about that. And it was really cool. And I looked around for like any examples of anyone actually doing it. I asked him, he's like, no, no one really. It's just a cool, like, research kind of thing. And so then I had a manic episode and built askmybook.com. And then, yeah, I was reading cool. And then so Helper anyway, the reason I say that besides, you know, my ego wanting credit is Helper is basically the same thing, right? There are still help docs. It's just that humans don't want to read help docs. There's still documentation. It's just, you know, there's still assembly, there's still CSS, there's still all this stuff. It's just humans generally don't— it's not natural, right? Like, I don't see a monkey doing that stuff outside, right? The monkeys are just like chilling and hanging out and like messing around with other monkeys. I feel like that's what AI will allow us to be, is more monkey-like. And so Helper basically is— it started off as basically Ask My Book, except instead of my book, you know, it's just the help docs. And now everyone calls this RAG. And so it's just RAG. But I think there's a lot more we can do beyond what, you know, this very saturated space. So I think we have sort of two approaches that I think are going to be pretty interesting. One is we're going to help government creators use it. And so it's not just about doing support, but it's like, hey, imagine you land on a government creator's ebook and you're thinking about whether to buy it or not. And you have a question, right? Like, I think that's a, that's a room to, to add the kind of how startups think about, about, you know, their Intercom sales kind of cycle. But for every Gumroad creator, so every Gumroad creator is going to be able to turn it on. I hear it's listed as $10 a month for that feature. I think we're going to change this actually and just say we get a commission. So like, we're like your sales agent. And if we help because you make a sale, maybe we'll take like 10%. And then on the other side, I also think that there's room to massively improve the user experience of using something like Gumroad because Gumroad, as our tour guide likes to say, is like, for example, if you wanna create a product, let's say you wanna do a course and it's like the next 4 Mondays from 6 to 7, it's $50. There's like 25 seats except on like the third class. There's only 15 seats or you're taking one, whatever. Like, imagine I set what I want and then I have to spend— how long would that take to like type into Gumroad, right? Like gumroad.com for it. So like, then you click products and you like type the name. You have to come up with a name. You're like, my class, you know, should I type my— should I make it class? Should I be like coding class? Why do I even need class there? Should I come up with a new name for this? For this object, like all this stuff when you're just like, I just wanna teach class, you know, the next 4 Mondays, uh, for $50, 25 seats each, except the third one, it's $15. And then you just say that and then boom, you're done. Like it just, Gumroad creates all that for you. Uh, I think this is how a lot more people will use products generally. Uh, and so I think Helper could be like a tool that helps a lot of companies do all that without actually having to do all of it. Kind of like a command bar is kind of how I imagine it. It may be like refund this customer, right? Imagine you get an email from a customer and you just need— they just ask for a refund. Like, why can't you just CC, like, why, like, why, you know, support@government, why can't it just be like help@government.com and then say, hey, could you refund this? And then you get a response in 1 second that's like, duh. So I think that's going to be way better. It's going to make the user experience of of making Gumroad really, really, really, really way better than it already is. We had Iffy. Iffy is basically our internal risk stuff productized, more focused on content moderation now. And so there's a lot more we need to do to make it really a lot better. But basically Gumroad should not even have to think about like whether someone is a bad actor or not. There's like thousands of lines of code that have to do with like suspending users, refunding, appeals, you know, all the admin dashboard stuff. Ideally, we can delete all that from Gumroad and then spin it out as a separate product called Iffy, iffy.com. And very happy with that domain name. And the goal there is just to get more customers. And that's— we're going to do that by hiring an in-person sales team. So if you're interested in potentially doing sales for us, let us know. And then fifth product, shortest. So this kind of came about because I was building stuff and I realized I never write tests for my own stuff because it's not fun. I like building stuff, coming up with ideas. And so my sort of, you know, it's not a new idea that people have had, which is how do we just make testing super, super simple and easy? And in my view, like the simplest test, a short test, would be a single English sentence, like just sign up to Gumroad and sell a product. And right now, if you have a test, you know, the equivalent test in Gumroad, if we even have one, because, you know, we don't really truly have end-to-end tests in that way where they kind of are comprehensive all the way from like, you know, going to slash and then doing everything. You know, it's like, it would be hundreds of lines of code and you'd have to be creative and think about like, okay, what's the email that user's gonna have? What's the price they're gonna sell the product for? What, like, why it should be different every single time. Right? Like every single time you should use a different email and sell a completely different product, uh, and kind of, uh, chaos monkey your, your product all the time. And, and this now recently became possible because AI is finally getting good enough where you can have this kind of a computer use agent behavior where you can effectively spin up an agent that can just test your, your product for you. And I do believe this is possible. We currently have Shortest running on the shortest repo and we'll kind of propagate back and have Shortest do tests for Iffy, for Helper, for Flexile, and for Gumroad eventually too. And Gumroad being open source will be a really good way for us to like basically promote all of these other products that make it even faster and easier and more fun to ship world-class products. So that's the, the roadmap. Yeah, then AI, AI, AI. Just for the SEO. AI, AI, AI. Why work for someone who's so focused on automation? I think that's a very good question that I think it's important to ask and answer. You know, it's like, why, why, why, why am I working to fire myself? It's like, well, that's the job. So you don't have to do it, but that's like what I'm paying you to do. But yeah, basically, I think the way you align the incentives is in profit sharing dividends, right? Like you basically say, hey, you know, you automate this thing that saves us $1,000 a month, you get a percent of that, right? So that's basically— I mean, it's pretty simple, but that's, that's how it works. Like, Rachel from Friends doesn't have to act in Friends anymore. Like, she doesn't show up on set every single day, but she gets paid like she does, right? And that's because she owns a percent of profits. All the Office actors I see all the time doing random stuff, probably because they don't have to show up to work every single day. Uh, and so that's kind of the goal, is to like build products that are really great, make money, but do not really require humans to run them, right? The reason we can have 5 products and maybe 26 products is they're kind of like pretty low maintenance costs, and actually building these new products only makes it easier and easier and lower effort. And there always will be— I mean, you know, the nice thing about me is that I do like to build stuff, so I don't— I'm not probably going to run out of ideas. I'm just going to have more of them. Then hopefully bigger and grander ideas and incorporate new things that I want to learn more stuff. Like we're working on this like hardware product at the office, you know, completely, completely new. But it's going back to that growth, optimization, destruction. The destruction will allow us to then have more energy to then focus on doing new stuff. And yeah, the goal is to open source everything. That's my— that's kind of where I think a lot of this stuff goes. It's the value of open source. It's just too great in the AI world. And so everything is going to be open source. We all use open source software all the time. The open source software eventually wins. Eventually there will be a better product than this MacBook and it's going to be open source. You know, it'll eventually happen 100%. And so the plan is to optimistically open source all of our code and then sort of the core team and the contributors basically all get paid in this sort of profit sharing model. The core team is more focused on sort of most, most creative tasks, scoping features, QAing the features, shipping the features, like hitting merge. And then we'll probably ideally have thousands and thousands of contributors who are all getting paid sort of this, this reward bounty performance-based challenge sort of thing. And yeah, as I mentioned, like the goal is not to end all work. I don't think that's really probable, at least not in the next like maybe a couple million years, you know, eventually I think we'll get there. We'll like reverse entropy and like prevent the heat death of the universe. Pretty optimistic about that. But at least my contribution to that will be trying to focus on ending some of the work that we do today so that those people can use better B2B SaaS. And so, yeah, lastly, I'm almost done. I had this tweet a while back, full-time employment is modern-day slavery. Uh, I don't— I did not say full-time employment is slavery. A lot of people took it to mean that. Uh, I could have said maybe modern-day equivalent, which would have been more clear at what I meant, which is like, if you had to pick one thing that resembles, uh, this old thing we used to have, like, you know, a lot of people say, oh, what's, what's something we're gonna think is evil today, right? Like, oh, eating meat. We're gonna like look back like And people are going to be like, you're eating meat? Like, it's weird, you know, it's like a dead carcass. Why don't we just— we should just refer to meat as carcasses. And, you know, so that's often like a, you know, an answer. I think— I actually think employment generally is true. And so this is just— oh, another nice thing with interactivity, you can kind of see how this equity split stuff works. And if you're interested in this, just email me and we'll We'll try to get you set up on Flex hours as soon as possible. But you can kind of see how this works. Let's say you're going to pay $250,000 a year. Sorry, Strong, I haven't merged your PR. There will be a comma in the annual compensation thing soon. It'd be funny if you merged it and then just like auto-refreshed. How cool would that be? Well, Vercel shows the thing like when it needs to— Oh, I could, I mean, I could hit here. Oh yeah, that's true actually. It should. Show a thing because I'm logging in my Vercel account. Well, we'll try and see what happens. And yeah, so, so, you know, you get paid, uh, $250,000 a year. Uh, yeah, you know, imagine, imagine no equity, right? Just keep it simple, like pretty, pretty simple math, right? You get paid $250,000 every year, right? Each year you get paid $250,000. Uh, if the company grows, like if you're doing a lot of awesome work to help me make more money Well, you're still getting paid the same, so what's the point? Why? What's your incentive, you know? Uh, versus an equity split. So let's say the company's not growing yet, right? We're assuming 0% growth. Let's say you're taking half of your comp in equity. Oh, you can kind of see how much you're— you know, in the beginning you're getting $125,000 in cash, half of of that, uh, $250,000, and then you're getting profit sharing. And then over time, your profit sharing should effectively— and this is kind of one way to think about it, is, uh, you're, you know, how I think about retirement, you know, if I truly believe in sort of employment is modern-day slavery, like, how does that look like? Well, the simple thing is to just figure out how long it's going to take until you're making more in dividends than you are from your salary, right? And so you can do that and see when it takes, like, let's say you take a 90% equity split, pretty aggressive, maybe possible if you live in Bali. And you can see that basically in 2033, you're now making $267,000 from dividends, profit sharing, et cetera, versus, and you know, you own, you actually have like accumulated 5% of the company at this point. So you own 5% of Antiwork. By doing a lot of work, ironically. And then ideally, let's be more reasonable and say 50/50. Ideally, we're growing, right? Like we're doing stuff, we're either saving money or growing the top line by doing things. And you can kind of see if we grow, you know, I don't know, this is pretty reasonable to me. I think 10% is reasonable. That's like less than 1% a month. It's hard though, certainly hard. Uh, then, you know, you can kind of get into this, uh, these numbers a little bit faster. Uh, and this is, uh, you know, this is how people do really, really well. Uh, I was talking to somebody who wants to make a million dollars a year. Don't we all? And so you can just kind of see— let's just keep that the same for now. Uh, you know, I guess right here, like it's possible, but we would have to grow. I mean, 40%, honestly, I mean, the S&P 500 is up 40% this year, which is kind of insane. But think about how big those companies are and they were able to do it. So it's possible, it just requires a lot of work. And of course it does because otherwise everyone would do it. And that's what inflation is basically. Inflation is effectively like the rate of sort of everyone's salary is going up. So basically everyone's in a race against everybody else. Inflation is just kind of like the current incline of the treadmill that everyone's on. And there's this thing called Baumol's cost disease, which if you want to go read about, you can check that out. And then lastly, this one other concept I want to introduce is basically I think how we will make decisions and run the company long-term going forward. I like my alphabet, ABCD. We spent a year, imagine, automating stuff, anti-working, making the business more valuable generally. That generates cash flow and profit. We use the profit to buy back, cash out early investors. I mean, it doesn't technically have to be early investors, but generally the people who are going to want to sell are going to be the ones who bought in the earliest, have the kind of the highest ROI to date. And so we, that's great for the company, because you can buy them out at a relatively good price. And then you crowdfund. You either use that price that you come up with via this reverse Dutch auction thing that we figured out to do at Flexile, or you can just say, hey, I think Gumroad is worth the S&P 500 multiple on our profit, or however, it's us, we get to choose. People don't have to invest. We crowdfund for people who are interested in sort of doubling down. So to sort of like a liquid cap table, right, where like early investors are leaving, new investors are coming in, uh, constantly changing, including all the team members who are getting paid a split in equity. Uh, and then dividends, uh, dividends, uh, to kind of keep everyone aligned and get everyone paid for efforts, including myself. Um, I think this is just, uh, unfortunately dividends is not super popular, so I think, I think the buybacks and other things would be just like in crowdfunding and all these things, just like public companies like Apple, Nvidia, Facebook, Google, they all do dividends. So dividends is a part of the business model. But at the end of the day, people are generally— investors are looking for more risk, more yield. And so we're going to satisfy that by building all these new stuff, new products. But I think the dividend is really important to make sure that we, one, can be profitable We're not like LARPing as a profitable company. And it just keeps us accountable. And I will take my salary effectively in dividends. So it's also, I think, a way of being clear that we are trying to make money. Otherwise, New York City is expensive. So my 4 years of runway will eventually run out. And no one wants to be there for when that happens. All right, 12 slides. I feel like I was better. I don't know what the time is, but I feel like it was reasonable. Oh, it's an hour and 6 minutes. That's not terrible. Awesome. You guys tell I'm hungover? No, right? Because if you haven't met me, this is just who I am. I'm just so much more bouncy. Any questions or comments or concerns or feedback? Nothing? Literally nothing? Anyone on the Zoom? Yes. Yeah, dude, you have the— you can use the mic. Oh, okay.

Excuse me, I have a question about payment processors and how you're going to work after the PayPal is cancelled? How are you working with the banks? Because, uh, just do Stripe. But in some countries, like, uh, for example Indonesia, where I have the bank, the Stripe is not available.

Stripe is available in Indonesia now.

Yeah, yeah, I found the You should ask Driven. Yeah, question about, uh, because the right now the payment, uh, go like 2 weeks or something. It's like, okay, or so it does work.

You told me it's not here, but then so it is here.

So the payouts, it's very slow.

Why? Why is it slow?

Maybe, maybe you can improve it somehow.

No.

Uh, what about the The cryptocurrency, we will add crypto.

Yeah, you're right, you're right. So Stripe is that, right? Stripe is adding crypto, so we're gonna— we will turn that on.

Okay, but about the Stripe, uh, can it be that it had the same problems as the PayPal? Because, uh, I don't know the reason why they cancel, but yeah, it's possible.

PayPal's just like a much shittier company, basically. Yeah, but like the likelihood is less with Stripe, and I, I'm friends with the founders, so that helps a lot.

Is Stripe going to have the same problems maybe, or not?

What do you think? I mean, what are the, what are the specific problems? Like the fact that PayPal just deleted our account without telling us? People almost did that and they stopped, or like they sent us an email and then they were like, we sent this email to you incorrectly.

So you're going to work with the cryptocurrency in the future?

Yeah, as soon as Stripe allows. Yeah, we're like a— basically we're built on top of Stripe, right? Gumroad is just like a Stripe wrapper, you could say.

Yeah, because right now, uh, I only had the bank payout to deposit.

Yeah, well, Stripe will— I mean, my— I mean, did you see they just bought this big crypto company called Bridge?

So in In the future, I had the option to pay out to my, for example, like USDC or USDT.

I would— that's what I would imagine. Yeah, exactly. Ideally, ideally creators.

Yeah, because if you can pay out to your cryptocurrency like address, you can cover any country. Because right now I see you have problem with covering the whole countries because, because of the PayPal.

And so, yeah, well, I mean, we can do more. There are still bottlenecks to covering every country because there's regulations that you have to follow. So it's not just about the functionality of supporting the crypto, right? You have to still do KYC. So you're still going to have to give us your government ID, all that. Like, you're not— you can't really make money without the government knowing because they need to collect their taxes basically and make sure you're not going to like, you know, blow up a building with your money basically. So like crypto is certainly like— I, I'm long crypto. I'm a big fan of crypto. I own crypto, etc. I don't sell my crypto. And I just think Stripe is a prop. I mean, I can't really speak on behalf of Stripe, obviously, but I think they're just bottom— they're trying, they're moving as fast as they can. Is my guess, and they would love to support all these things that you say.

So right now Stripe in Indonesia not working.

But you just said you got it worked. It just takes 2 weeks.

Yeah, it's work, but it's not perfect. Yeah, not perfect.

Yeah. And have you asked Stripe about it?

I don't have option like at my own Stripe account.

Yeah, I'm curious, like, what Stripe says if you, if you ask them why it's slow. What would they say?

Because, uh, before I saw I had the option to add, uh, like my Stripe, connect to my Stripe account, my personal Stripe, and get payout to my personal stream.

But right now I don't get this option, uh, but I have the Stripe account Yeah, we're going to, we're going to, we're, I think there are ways for us to make it faster to get paid. Like we're going to add an instant payouts feature. So you'll be able to like say, hey, I want to get paid right now. However, all we can do is trigger the payout, right? So then it's hard for us to actually say, okay, well, that means that like Stripe is like this, this money, you know, has to move from this bank to this bank. And some banks still have like a lot of work, like a lot of, you know, pre-automation work, right? Where there's, there's some human. Like every once in a while I like transfer money and then they, I have a, like my, I get a phone call, which makes me think I should leave my personal bank account and move to Mercury. And you know, and so I think, I think my guess is it's probably like Indonesia's banking system is not great though. I did see they just, I think it was Indonesia's largest bank just rewrote their backend in Rust or something. So, you know, that's cool. So I might, you know, there's change. I mean, I don't know, I'm optimistic. I don't know, I'm optimistic about it. The beauty is Gumroad will be open source, so you can add crypto yourself, right? Or we can do it together.

Should I host by myself or not?

It will allow you to do that. So you could basically just like deploy it for sale, and then all of a sudden you have like your own Gumroad account. And if you make a million dollars a year, then you owe us some If I host by myself, uh, and where the money go from the customers, it'll go, it'll go to yours, whatever you're like. Let's say you put in your Stripe API key, it'll just go to your Stripe account. Or there'll be, I'm sure there will be a crypto variant where you'll be able to, you probably could specify, just say, hey, this is my Ethereum wallet address, or this is my USDC wallet, or USDT or, or whatever. And then maybe it's an integration built on top of Coinbase. Um, I don't know, it could be— that's the cool thing is like we may say, hey, we want to do things this way, but you know, there could be a pseudo-competitor that comes out and says, hey, we're gonna like fork Gumroad and call it, uh, you know, Gumroad that sees for crypto, and then, uh, launch, you know, launch that.

What, uh, like a date or where When the crypto will be available, do you have anything like dates or time?

Um, Mike, if I had to make a guess— I'm pretty good at guessing, I think— I would say my guess would be April of 2026. That's my guess. That's a guess though, don't hold me to it, but that's my pseudo-educated guess.

Right now, I have two options with PayPal. It's, uh, first I connect to get my payouts, and the second option is I connect my personal, uh, account and get like customer pay to my PayPal. Like this, usually the most platforms like this, they like direct, uh, the payments to my personal PayPal account.

Yeah, but before, uh, yeah, PayPal required us to make that change basically.

Before, my personal account was banned because of, uh, after I like connect from, uh, from the Gumroad, I connect my personal account, and after like working about maybe 1 or 2 months, they, they banned, and they didn't even say Why they banned, they not say the reason why they banned. It's very strange.

Yeah, well, I'm hopeful that Iffy can help at least with that part where we can give you like really good reasons for why something was not supported or banned. But yeah, this is actually like a problem with PayPal generally. I think that the core issue with PayPal is basically they're like very big. And so, you know, they're, they're not really even like when, when they suspended our Gumroad PayPal account, like there was no manual review of that. Like they were just like, hey, we just— and I'm like, this is weird. Eventually the CEO like texted me about it and was like, sorry, but it took, you know, 2 days, 3 days or whatever, which is kind of insane. So it's kind of a one step backwards. It does suck a little bit that we're like removing PayPal, but I do think the long term is going to be really nice, which is like we can double down on Like, we want to add all these, like, local payment types. Ideally, we can support, like, in, you know, Indonesia may have, like, a certain way of doing things, or in India you have UPI, etc. In the EU, you may have other payment methods. I would love to see a world in which, like, PayPal is completely killed. I would like to contribute to that. I would like—

Okay, I don't have questions. I mean, my main questions are about the processors because it's the main reason why I use the Gumroad before. Right now it's not a lot like different platforms that work same like Gumroad, like only I know only the Patreon. Yeah, I can, I can recommend the same like If you connect personal PayPal account, you probably get banned after a few months.

I'm curious what you're selling, but we can talk about that after this recording is over. Um, any other questions? Yes, sir.

So, uh, do you have some internal in-house tools like leveraging AI or Internal.

Can you give me an example of an internal AI tool at a different company? A system powered by AI that's doing an important task in-house. I mean, I think there's no real need for that. It's kind of the messy middle where you would just go all the ways of straight automation, right? Like all, like there's, AI is just like a, like a short way of saying automation. Like nothing new really, right? And so everything is AI. Everything we do at Gumroad, every single thing we build is a form of trying to automate a human task, right? For example, fine-tuning an AI model for a specific use case in your company. We don't, We don't fine-tune anything. Yeah, we just use like prompt stuffing, basically. I personally think fine-tuning is like a, is not really worth doing generally because the smarter the AI gets, the more you can just tell it to effectively fine-tune at inference. Just, you know, ask it to do it, what you want it to do, and it'll do it. But those products I think are like good examples of effectively anything we think is really valuable internally, we'll just ship as a product. Right? Like, why not? Why would we keep it to ourselves? So like shortest.com, Flexile, they all have AI like built into them. Like, you know, it'll be kind of like the core way you use these products. We'll be sort of like talking to it, you know, like, do you like, I don't know, do you like talking? Like, do you like texting your friends or do you like hanging out with them in person and talking to them? Right? Like, I think there's like a superior form of communication. It's just we haven't been able to talk to computers, so we've had to create APIs. Why do APIs exist? It doesn't make a ton of sense. Why can't you just have ify.com and then a string that's like, just do this thing? Then the translation can happen on the fly basically. I'm very excited about a lot of these. Kind of ideas. Yeah, I'm trying to think. I mean, and the other thing is we do use a lot of tooling, like just, you know, other products like Slack, for example, where they're just getting better. They're just going to build AI directly into the products. So, you know, we'll kind of be using AI just by nature. Like when you search for something in Slack right now, it's not using AI, quote unquote, right? It's just doing some sort of like textual embeddings, sort of Elasticsearch or some, some, something. That's basically deterministic to some degree. And so, you know, I think eventually you'll say, hey, like, I said something in Slack, like, about 3 months ago that was about China. What did I say? And then it would just say, you said this, and then click message, right? Like, it wouldn't be like search. It would just be chat in a bit. So I do think chat is like a pretty dominant interface, like, of the future. Um, like, I noticed this, by the way, like, when every time, uh, you, uh, a plane lands and you look at what everyone is doing, uh, they're all chatting, like, they're all texting their friends, family, uh, you know, they're either using iMessage or, uh, WhatsApp or Instagram messages, uh, they all basically stop doing what they're like scrolling TikTok and start socializing. Like, you know, like TikTok is just basically like what you do when you like aren't able to hang out with your friends. Uh, and so, uh, or they call— some people just like call their friend, you know, just say, hey, I'm— my plane just landed, well, whatever. Uh, so to me that's like a pretty big sign, like that's what humans want to do. They want to connect with each other. Uh, and the more they can do that, the less they will want to do anything else. Uh, which is why I think all these things are kind of interim technologies and why dividends and profit sharing being profitable so important is because eventually our industry will go to zero, uh, complete zero, right? And so you don't— you can't rely on this sort of like, you know, this up into the right chart that I drew before because like that's not true, you know? It's true on like a societal scale, but every industry has a kind of a hump and then dies, right? Um, like horseback riding, you know? Or race car driving or chess or whatnot. Like chess is, you know, used to be actually like, it's much, you know, chess is like relatively small compared to like what it replaced, which was war, right? Like war was, I mean, it's like literally kings and queens and like horses and bishops. I mean, it was like, you know, like the Crusades, like chess is just like a tiny form of the Crusades. And think about how much more work that was versus like how much chess, you know, it's a tiny, tiny industry relative to what it used to be. Like that's all we have left of the war in, you know, in a— if you're in a safe country, it's just chess and other video games, that's the thing. Yeah. Yes, you know. Thank you. I have a— it's a question related to— I know what Big Link Day is. You want to host on Vercel frontend, but why did you pick Vercel? Why over Cloudflare Workers over hosting your frontend on Hetzner? Because you said you wanted to use Hetzner. Yeah. Or, wait, like, I don't know. I just think Vercel's like just the current winner in the speed. Like, I think if you value shipping fast, I just think Vercel's the best. Like, Hetzner is great, we use Hetzner. Cloudflare is great, we use Cloudflare. Uh, Vercel, I think, is what you would use if you're purely optimizing for speed of shipping. Uh, like you get preview deploy. I don't know if Cloudflare Workers has preview deployments for free. Uh, but Vercel does, you know, you can comment, you know, on a Vercel, like I can give you a URL, you can comment on it. Like I just, I, I, I hope that all these other things compete cuz I would, I have no real skin in the game. I have a tiny investment in Vercel. You know, I'm not gonna get rich off my $10,000 in Vercel, you know, anytime soon. And so it's not really about that. It's more about, I really just believe Vercel is like, just, I, I personally think like it's the best sort of platform for TypeScript. And Gumroad used to use Google App Engine back when we were written in Python, which is now renamed to like Google Cloud Platform or something. Uh, and so yeah, I, you know, I just believe strongly in this sort of like vertical integration of like language and hosting platform. I think it unlocks really a lot of really cool stuff. I just, yeah, I think, I think Vercel is by far the best. The only downside to Vercel is it's, it's not free. I wouldn't even say it's expensive. I don't think it's expensive at all. Vercel is actually free to use. So like, unless you're doing a ton of traffic, it's effectively free. What's expensive is Humans. So if I can use Vercel to save tons of time, then it's like an amazing, amazing deal. And just like DHH says, like, don't worry about optimizing now, optimize in the future. Well, that's why you should not use Ruby on Rails. Like, you should focus on a stack that is really, really good for shipping. And then if it does get too expensive, the beauty of TypeScript and all these things, it's so easy to move stacks, right? Because you have all this stuff. So eventually, the way I think about it, if I were to draw like a little diagram, Maybe it's helpful because I haven't explained this exactly. I'll put this back here and I'll draw this little thing. So, so, so, CF, uh, head start. I'm bullish on on-prem. I think that's gonna be fun. Why not just have a little Mac Mini, powergummer.com? I think Gumroad does about enough traffic to be run on 5 MacBook Pros. And if you think Vercel's expensive, AWS, it's way more expensive. Yeah. Right? And so, and that's why I kind of like took a shot at the whole the Unreal crowd because DJ said it was really expensive. And I was like, hey, but you're the one— your software makes me pay AWS $42,000 a month. And like, that's a website that can be run on 5 MacBook Pros. So like, I was kind of annoyed at him, so I tweeted. So that's— that's for me. But yeah, basically I think it's like speed and cost. Right? Like, those are the two metrics. And so basically start with speed, and then if cost becomes important, like, move stuff over. So I think the cheapest would be, you know, something like on-prem, right? Like, if you don't need mega reliability. But you need network engineers or SREs. Yeah, but fine. I think we need one. Yeah, you can. Yeah. Or you, you know, you work with like some co-located provider like back in the day, like Rackspace or whatever, Deft. But yeah, I think this is kind of my approach. It's like we should do everything on Vercel because it's actually free. And then eventually as things get to scale, you know, you can, you can opt cost optimize, right? Just, just like AI, right? Like for example, if you use Claude, by default, we use, you know, it's the fastest, or OpenAI or whatever. But then eventually, if things get to like we're spending like $100,000 a month on compute, then we should just have Llama. Yeah, Llama running on, you know, a big box somewhere. Yeah, yeah, one of, one of something else. I don't, I don't know exactly what it'll be. And you could, you could say actually that this is more like Hetzner should be over here. Maybe CF is here. Nice and here to the right. Something like that, right? So like Vercel, Cloudflare, Hetzner, on-prem, right? Like based on what matters, speed or cost, you know, but what matters mostly for us, speed right now. So, and every stack can be like this. Gumroad will hit Vercel first, and then the API call could be something else, right? But the, the, the, the, the front line, in my opinion, should almost all definitely be Vercel. I don't think it's a coincidence, like walmart.com or Washington Post. It's just really good, and you get— I mean, it's by far the best user experience, I think. If you go to gumroad.com page, uh, and you, you know, It's just not an amazing user experience. It could be a lot better if it were hosted on Next.js on your site. Have you tried Railway? Any good? I did briefly, like a couple years ago. It's been a while though. It was very buggy, but it's also been a couple years. I've heard good things about it on Twitter. Yeah, one nice thing about the office is people just come by and I just ask them about stuff. And that's— because the problem is that Twitter is not, you know, Twitter is like a fog of war. You don't really know what people— what actually, like, you just see like memes and like viral things. Uh, like if I just tweeted like Marcel is a piece of shit, it'd probably go viral. But like, that's easy for me to do, right? I can just lie. Uh, and you know, we have freedom of speech in America, so lying is legal. So we It's, uh, yeah, it is what it is. That's why everyone says they're like the number one burger in America. They're allowed to say that. And I guess the last one was, where are you getting your design? If you're— so is it like everybody has like the public Figma like files, prototypes right now? Like, or is it just like complete decentralization? Like, I don't care, just give me the PNG, whatever. Yeah, I mean, I think we're still figuring it out. Like a great process. I use v0 personally a lot, so I'm a huge fan of— I mean, the antiwork.com is v0. Like, most, most things I do, I go to v0 first. Um, I'm actually bullish on this idea. Like, imagine if you go to v0 to GitHub to Vercel to host your thing. Like, could Vercel just kill GitHub? Why can't, why can't you go from v0 to vercel.com directly? So I'm bullish on, on that. Potentially. I don't really know, just making a guess. That's something I would do if I were them. Uh, sorry, what was your question? Oh, design. Um, yeah, I think we basically want to open source as much stuff as possible, and then designers get to choose how they present that information, right? I personally have started using Photoshop again. Uh, I'm also a Figma investor, but I don't— I'm not right now. I'm like, I, I'm happy to say I use Photoshop more than Figma. Why? Because the deliverable is not Figma files, it's the product. So it doesn't matter. And your React to design to React or design to JSX is not that good. Yeah. Yeah. And I'd rather, like, what I like to do is I take a screenshot. Like, normally what is design? You're generally redesigning, right? Like, you're taking something and making it better. You know, one time you design and then end times, you're just redesigning the design. So what I generally do is I take a screenshot of the website. I either put it in V0 or Cursor depending on what I'm trying to do. But if I want like a scratchpad sort of designer visual perspective, I just paste in Photoshop and then I just take out the pencil or paintbrush tool and I just select the back— the background color and I like erase the parts that I don't want. And I just designed that, redesigned that part, uh, with pixels. Pixels are great. Like Figma is great for like React vector or whatever. But personally, I like, I think it's kind of constraining actually. Like I prefer like starting here with like pen and paper and then Photoshop and then V0 and then Cursor. Like that's my current stack. The beauty though is Figma has a lot of agency as a company, so they're gonna They're gonna be back in our stack at some point because they're gonna ship a lot of cool stuff. But currently I think they're kind of a little bit slow on the AI uptake. So they're currently out of our, out of my personal stack right now. And the last question is on FlexCell and Bounces. A lot of people do that in Tiny Core because they went through Spots and now— Oh yes, I know. You know, Replit are trying to do that and I do hear a lot at I know it's sort of a very open market in Bonsai Stereo. Did you think of integrating, for example, Replit? I know I'm a Java Stack investor, so do you think of working with them on Lightning? Oh yeah, I'm an investor in Replit. Um, nice. So instead of creating that, like basically using Sahil Lavingia's wrapper on top of Replit instead of like creating your own Lightning system, especially for code, like, did you ever think of that? Or did you think of like Ideally we do it and then they use us. So the thing is, I like building stuff. So the problem is like, if someone's asking me, they're like, what happens when like AI replaces the prompting? And I'm like, it won't because I enjoy that part. Yeah, like you can continue to do the things you enjoy. Like, so that's like where you said it in a podcast recently, AI is not gonna eat your ice cream for you. Like, you need to— yeah. Yeah, AI is not going to tell you what your favorite ice cream flavor is, right? You're gonna have to make that decision and try a bunch of stuff. Uh, this, this, uh, retreat we've been doing in Bali as a company the last couple weeks, it's like, I think it's very healthy reminder of like, like what's fun and what people like to really do. Hanging out, socializing, eating ice cream. Surfing, sitting on the beach, reading a book, engaging with other humans, uh, in the limited time we have on this— in this— on this planet. Any thoughts about— sorry, what was that Oh yeah, I can show you. I can show you guys, uh, the 6 products. Yeah, it's this. By the way, check out this logo. I told you I love this shit. Wee! It's a door. And it basically just plays your welcome song when you walk into the office. It's a Next.js app. So yeah, basically it's got a closer one shot of this one shot. Everyone knows the moves. Yeah, so basically it's like a door. Like my aspiration for this, oh yeah, for people who can't see this on the thing, I'll share my screen quickly. Basically, well, now I have to ship this, don't I? Yeah, so basically I would love— have you ever been to Burning Man? There are no doors at Burning Man, or there shouldn't be anyway, because people trust each other and it's kind of like this hippie-dippie thing, but I like it. I think it would be cool if there were fewer doors. I think there should be thresholds, those doors. And so I thought, yeah, we basically, it started as this, you know, a way of just like, kind of like, I don't know, just fun, you know, like you get a walk-up song, right? You walk in the office and it plays your song. And, but I think there's like sort of this, there's also this kind of social element to it, which is if you get pinged that, oh, like, Khan's in the office, Sahil's in the office, Shad's in the office, maybe you're more likely to come in and like hang out. And I think there's just kind of just like a nice health benefit to it. But over time, I would like to basically just like get rid of the door and just have a camera. And that camera effect— and one day it'll be a robot, probably it'll be like a bouncer that just like, you know, hangs out in your office 24/7 and effectively just like is like a bodyguard/receptionist. And then the door just becomes, you know, the camera is just an eyes, this future robot that doesn't yet exist. I can, can I show you guys this? This is also AI.