Sahil Lavingia: Founding Gumroad, The Minimalist Entrepreneur, and our AI LLM future — #27

Sahil Lavingia founded Gumroad at the age of 19 and built it into a leading digital commerce platform. He is the author of The Minimalist Entrepreneur and an investor in early-stage startups.

Steve Hsu: Welcome to Manifold. My guest today is Sahil Lavingia. He's the founder of Gumroad, an e-Commerce platform, and also the author of The Minimalist Entrepreneur. Sahil, welcome to the podcast.

Sahil Lavingia: Thanks for having me. Excited to be talking to you.

Steve Hsu: I am as well because I've listened to you in some other contexts and read your book. And being an old guy who's founded multiple startups, I'm always interested in hearing from the younger generation. One of the things that I think impressed me the most about your story was that I think you got funded by Kleiner when you were 19 years old for an A round, which is just unbelievable.

I'd like to start with your early life. I think you mentioned that you grew up in Singapore and attended USC. Maybe you can just tell us what your background is.

Sahil Lavingia: Totally. So, my parents grew up in India, moved to the US in the late seventies, early eighties. I was born on Long Island when I was five or four. We moved abroad. my dad got into, they, they, they came to the US for their, their master's degrees, went to Baruch College. and then my dad got a job in banking, and we moved to Singapore, which is where I primarily grew up.

And when I was, yeah, 17, graduated high school, moved back to the US in 2010 and for a computer science degree, and only lasted four months, before I dropped out and ended up moving to the Bay Area joining a startup called Pinterest, working there, building Pinterest for iPhones, starting my own company after that.

Sahil Lavingia: And yeah, it was pretty crazy because you know, and basically in a year and a half I went from like, yeah, graduating high school to raising a series A from Kleiner. which was, was not on my yearbook goals or anything, you know,

Steve Hsu: Incredible. Now I just have a couple of questions. So, you know, what you told me answers one question I had, which is that you know, you had a very American accent as you have a very American accent. And I thought, wow, kiddo grows up in Singapore, that's a little bit unusual. But you've answered my question there.

When, when you went off to SC were you thinking you were going to just complete the program? You were not thinking at all that, you wanted to be an entrepreneur right away?

Sahil Lavingia: I don't think I was thinking about it, you know, I feel like I spent a lot of time trying to figure out what I was thinking at all sorts of points in my past. And I think it's so tempting too. Insert your present self and all of your existing sort of biases and, and you know, knowledge that you have.

But I think at the time I just wanted to; I think my goal was to get a job at Google. I'm pretty sure that was like the sort of, you know, the ideal, the reach, right? Where, if getting into Stanford was the reach and I failed to do that and got into USC. The goal was to, you know, to get a job as a software engineer at Google anyway, right?

I did not expect to drop out, I don't think. I think I was really there to like graduate, you know, do a couple internships perhaps. And then, and then, you know, eventually I did, I think I want to start a company. I felt like that was, you know, making apps independently was something I was already doing.

So, I think that was on the roadmap, but I felt like you needed more information, more knowledge in order to actually do it well and succeed. And so, yeah, I, I don't think it was on the table. And just, just like, frankly, I think when I joined Pinterest, I didn't think starting a company was on the table either.

Sahil Lavingia: I think these things kind of happen, you know, pseudo randomly.

Steve Hsu: Yeah. I wanted to ask you, which was the bigger leap leaving college for Pinterest or leaving Pinterest to start your own company?

Sahil Lavingia: That's a good question that no one has ever asked me. So, I don't know if I've ever considered that before.

Steve Hsu: Well just think about how you felt at those various instances.

Sahil Lavingia: Yeah, I mean, yeah, that's true. I mean, I do think that leaving USC was probably the bigger leap. And statistically too, you know, I know very few, even though the stories are popular, everyone knows the college dropout stories, but there's not very many examples I have of people actually doing it.

And so, I think that that is probably like the sort of thing that was harder to do. It was the thing, you know, I had to do on my own. There was nobody around me that was agreeing with my decision. I was the only person that thought this was a good idea. Even Pinterest was like, are you sure you want to do this?

Which is kind of, kind of crazy in hindsight. because I feel like we tell people all the time that this is a great idea and, and I think it is. And, and yet, you know, everyone was against it. So, yeah, I think that was the big decision. I think that was the one that is the one that, you know, you kind of have to face your parents and say, Hey, you know, that college degree that, you know, you moved halfway around the world for me to get 30 years later is not happening.

At least not in the timeframe that you kind of expected. So, I think that was the bigger, the bigger relief. Once, once you’re kind of disconnected, it's kind of like the umbilical cord is severed. You're in San Francisco, you're in the Bay Area, you know, there's like people going to Burning Man.

Sahil Lavingia: Like, you know, you start, you, your, your openness I think starts to go up, you know, as a, as a product of your, of your environment a little bit. So, yeah, I think, I think leaving, leaving college was definitely, I think the bigger, the bigger one.

Steve Hsu: I could kind of imagine that, you know, your, if your dad was a financier and you know, doing international banking or working in Singapore, he might have this attitude, which a lot of financiers have, is that, you know, the grass is always greener. They always think the startup guys are the ones who have it easy and they're going to get the quick billions.

So maybe he thought this is a reasonable risk from the sun. Yeah, go ahead.

Sahil Lavingia: Yeah. Well, I, I've thought about that a little bit, right? Where I was, you know, growing up in Singapore, going to an international school, you know, content warning for privilege. but you know, like I did have that feeling of like, wow, I really don't want to go into finance, which was the default path because everybody and their parents were kind of, you know, doing that already.

And I didn't, it didn't seem fun, you know, it didn't seem fun, like just being on a plane, you know, three or four days a week and, and just doing, you know, they just didn't, didn't seem appealing. Like me, I still never really know what. what, what the value add is necessarily, you know, sometimes like talking to the folks in, in that industry.

But, you know, and so yeah, I kind of agree with you. Like, you know, I think for me it was like maybe software and tech will be, will be different, but when I apply that, it seems like, Well, that should have been the same decision that other people would've made too. Right? I was not the only person who grew up in that environment and would've felt the grass is the only greener thing.

But yet most of the folks that I know did go into finance. Right? So why, why is that? Like, why, why me? You know? And I don't, I don't, I don't really have a good answer, I think, for that yet.

Steve Hsu: My, my experience in, you know, I, because I come from this weird background, I'm a physicist, but I've started multiple tech companies and most of my friends from physics work in finance, hedge funds and things like this. The way I would describe it is that finance is still the best risk adjusted route to getting to a net worth of $10 or $20 million by the time you're sitting in your forties.

And maybe not the fastest finance could. I mean, startups can be faster, but I know very few financiers who love their jobs. I mean, they do their jobs, but they're doing their jobs too, to make the coin, whereas a lot of startup guys love their jobs.

Sahil Lavingia: Yeah, that's true. And that, that, that was always super important to me. I remember the first time I hung out with my high school friends after, you know, a couple years of college. I had, you know, been working at Pinterest, I think at the time. And they were still in college and, you know, some, some of them were maybe doing internships and yeah, I was like the only person there.

Loved what they did for work, you know, or, or like, had that expectation even that I should, and it was, it was, it was definitely kind of jarring, to be like, no, I think, yeah, I'm probably optimizing like risk adjusted yet, you know, joining Pinterest at the time didn't look maybe the smartest thing doing Gumroad certainly for many years did not look like the smartest thing after Pinterest did, you know?

And, but yeah, I, I just, I don't know what the, you know, what the sort of the big five personality that aligns with that is, but I just, I don't have a lot of patience, I guess for, you know, people I don't like and, and, and processes I don't like. And the finance industry and frankly the secondary, you know, education system generally is full of that.

So, I, now it feels startups were kind of an inevitability for me, you know? but again, it's kind of the story that I tell myself, right? You can't help it to a degree.

Steve Hsu: I think, you know, from an evo psych perspective, if you're a, you know, open to experience and the kind of person who's some, something of a leader and creative, you know, leading a small team that you hand-picked at the age of 20 and not having a boss. I mean, you have a board, but you don't really have a. That's a peak experience. A lot of people will go through their whole lives and never be able to do that.

Sahil Lavingia: It's true. Yeah. I mean, it's kind of crazy because I, I grew up in, you know, I, I feel like I grew up in San Francisco, you know, culturally, even though I, I didn't, because I was on Twitter so much I think as a kid. but now I live in Oregon, and it is interesting to like, you know, talk to the average 19-year-old here and feel like, oh my gosh.

Like, I really didn't really, did not appreciate how unique my experience was. You know, being able to raise $7 million at 19, it is just like, yeah, it doesn't happen very often. but for, you know, when you, when you do it and then there, you know, three other people who've done it, you know, because you had dinner yesterday, you know, in Soma, yeah. You just, you tend to think that it's a lot more, it's a lot more normal, than it, than it really is.

Steve Hsu: You know, one of the reasons I wanted to interview you is, I read your article on Medium and we'll talk a little bit about this. I know the listener may not understand what we're talking about just at the second, but, when I read some of the stuff you had written and listened to your talks, I, I realized you're a super grounded guy, even though in a way, your experience thus far in your life is easily one in a million, right?

It's not, I don't think there's more than one in a million people who actually have gotten to experience what you have over the last, you know, whatever, 10-ish years. So, super awesome that you're humble still about it, and you have a very developed, humble philosophy because what happened with, with Gumroad, let me just jump in there and just, get some of the basic facts, which I learned from, I think from your article or from listening to your talks just out there for the audience.

So, you founded Gumroad, which is an e-commerce platform, which allows, I would say, you were going to say this better than me, but roughly allows creators of digital things or services to monetize them by, by directly selling them on the platform. Is that fair?

Sahil Lavingia: Exactly. Yeah. So, we started in 2011, and I think that, the key insight and you know, nowadays there are lots of ways to sell content on the internet, CK, Patreon, Gumroad, Teachable, et cetera, and all awesome products. But I think at the time people started to build, basically started to build audiences before they were building websites, which is very strange because, you know, not too long ago you needed a website to have any presence on the internet.

And so that was kind of the insight, the epiphany was like, wait, what, what, what happens when you have a hundred thousand followers? and no website, right? All of a sudden you have like, basically the hard part about building a business, but not the easy part about building a business. And that just felt kind of, you know, the, the easy part felt too hard in a way, for me.

And, and so yeah, that was kind of the, the insight was like, we just want to be this the simplest way for anyone who wants to sell, you know, music or a PDF or a set of PDFs or a zip file or any, any, any binary, you know, kind of object that you, you can make on your computer. We want to make it really easy for them to just make a dollar on the internet.

And I think even, even, I didn't really realize why, I think I picked Gumroad specifically of all the things I could have worked on. Cause I had a bunch of side projects. but I think I picked it because it was very reminiscent of the app store. And, you know, growing up in Singapore, accepting payments online was basically doing freelance and having a PayPal account.

And then trying to use that PayPal account to buy other things on the internet that, you know, accepted PayPal. And the app store was the first time thatI would like, become an entrepreneur because I could actually sell things on the internet and then Apple for their 30%, you know, would take care of everything else, right?

Legal and tax. And I would just get a check, you know, three months later, you know, with, with, with my 300 bucks or whatever, right? and I felt like, oh yeah, like this, the internet is, is for this, you know, it's for connecting people of course. But once people get connected, you know, payments is, is, is a big, is going to be a big part.

And so that's kind of what led well to Gumroad. And, and we, we still like to think that we're like the simplest, fastest way to get set up, you know, if you want to sell something on the internet and you've never done that before, you don't have a website or, or any presence beside maybe a Twitter account or an Instagram account or something like that.

Steve Hsu: Once you connect, next you have to be able to transact.

Sahil Lavingia: I think so, I think, I think that, you know, that might be sort of a, a lesson from human history or something where, you know, it's sort of speech, and then money, you know, And, and, and money. Money is sort of a form of, of speech, a form of communication, that, you know, took, took, took a lot longer to develop, conceptually, I think.

But yeah, to me it's sort of, it's sort of like the internet without, you know, peer-to-peer payments, you know, would, would be like society, you know, without, without money, right? Like it would not really function super well, it would be a lot, a lot of IOUs.

Steve Hsu: So, you had left college, you were working at Pinterest, but you were still doing some side projects. Then you had an inspiration moment where you conceived of, you know, this building, this functionality so people could transact. And then you set about raising money and you were able to raise an A round of seven or 8 million from Kleiner Perkins, which for the audience is one of the tops, you know, one of the most prestigious funds in Silicon Valley.

Steve Hsu: Could you just talk about that fundraising process? Like at what point did you decide you go from, okay, the built this project, it's going to be a company, now I got to go raise and 19-year-old kid raises an A round from the top fund.

Sahil Lavingia: Yeah, so I was at Pinterest, this is April. So, I'd been there for maybe four or five months. And this was, you know, like I dreamed about living in Palo Alto, like one does when they're 16 years old.

Steve Hsu: Yep.

Sahil Lavingia: But I was finally here, and so I, you know, I was, I was spending weekends building stuff, hacking with other people on stuff, meeting people.

And so yeah, I just had a like, just like a, so many just different side projects. Ideas and Gumroad was the first one that I sort of worked on two weekends in a row. and so, I think that was probably the first signal that there was something kind of more meaningful here, where I kind of wanted to work on it, to work on it, instead of just to learn something new or to build something generally.

And then really what happened was, like, I started getting emails from people basically saying like, when are you starting a company? You know, just presuming that I was a founder or had that aspiration or desire to, and I remember getting an email from, I think it was Craig Shapiro from Collaborative Fund, who ended up being the first investor in Gumroad.

He saw Gumroad on Hacker News. I had titled it My Weekend Project. And yeah, he sent me an email and he basically said, Hey, if you ever decide to start a, you know, I'll give you 10 grand. And I think I replied saying, what if I started an LLC? You know, like me, I have no interest in raising a venture. I'm happy at Pinterest.

But you know, if I spun up like an LLC on LegalZoom, you know, would you take a million-dollar valuation, is what I said. And I think the reason I said that was because I wanted to Angel Invest and Angel investing requires a net worth of a million. And so, I was like, oh, if I can like to start the, you know, I didn't realize at the time that this doesn't matter at all.

No one checks this information. But I thought it was, it was almost like a hack, in order to start angel investing. and really just like to deepen my, you know, my tentacles in Silicon Valley as quickly as feasibly possible. I thought Angel Investing was the best way to do that. And he said yes. He said, yeah, if you send me a bank account.

And so, I created, you know, I went to Legal Zoom, spent 300 bucks, created little big things LLC, and then, you know, took his 10 grand, did nothing with it, really. But I got, you know, two or three, four more of those emails and, and you know, yeah, I sort of realized like, wait a second. Like I could actually be a founder, you know?

And so, yeah, you know, eventually that virus kind of got to me and I said, you know, why would I, even though Pinterest was awesome, you know, why? Like I, the, the, the, the, the experience of being a founder, being able to experience that, as you mentioned, at such a young age, effectively skipping like 10 years of career growth, being able to raise a, you know, and, and the people wanting to give me, you know, to write money into the startup were really great people.

There were people there. If I had done every single thing, right. I didn't dream of them ever being on my cap table, right? People like Max Levchin, the co-founder of PayPal, who I had read about, you know, but didn't expect to ever meet. and so, when I was able to do that at 18 or 19, it was, yeah, it was kind of a no-brainer, in a sense where I was like, why would I, how could I say no to something like this?

Which to, to be honest, I think foreshadows some of my problem, which is like, so, or, you know, generally maybe some problems with society is like you, you kind of just go for it because so few people do get that shot. You sometimes don't actually question like, should I have taken venture capital?

Should I have committed for, you know, that long? But I think at the time it was like, wow, I really can sort of, you know, I, I think of it almost like climbing or something where you can fall, but you can only fall so far back. And once you raise a Series A from Kleiner, like that's, you know, forever in a way, right?

And so, if I, even if I fail even, you know, it's sort of going to anchor me to being able to do something. that I think will, will sort of have perpetual value. and so, I think that was, that was also part of, I think why, why I did that. but I think along the way I was very clear with people, and I think I still am about like, look, this is who I am.

Like, I like to build stuff. This money's going to allow me to build stuff without having to think about making money. So that's primarily why I'm doing it. You know if this thing starts to work, you know, we'll see what happens. And I think as long as you, you know, you do, you have the skills to actually build what you say you want to build.

The sales pitch is not actually that complicated. I think what happens when people say that, you know, they struggle to raise money is often that they don't actually have the skills to raise money or, or to actually build the thing, which is why they're raising money so they can go hire people to build a thing.

And that, that's tough, that, that often, you know, leads to a no. Right. because VCs don't want to feel needed. They want, you know, they want to feel like they are not needed. and, and that's when they really want to participate. So, I think I was able to give it to them, right? A young founder who knows how to code, can design a little bit, is not reliant on anybody else.

You know, young also means, you know, my cost of living was, you know, super low and, and things like that. So, I think I was like, yeah, like. I think in hindsight, you know, I'm the, I'm the founder that I now look for, right? Now that I am in my position as an investor, I realize, like, holy crap. Like, I'm much rarer than I thought, like trying to find this kind of person.

I thought I would start a fund and I would go find hundreds of me, like, you know, all around the world. you know, there's this quote, you know, talent is evenly distributed, but opportunity is not. and so, you know, with covid and remote, like, I think, I think people actually, you know, and this is a lot of my growth in the last year, I think has been about like, why, why do I continue to be surprised?

You know, you when, when, when the data doesn't seem to line up with that. So yeah. Anyway, it's something I think a lot about right.

Steve Hsu: I'd say talent is rare, and I would say you're kind of a unique triple threat. So, you, you are able to actually do the coding and build the product yourself, but you're also super articulate. And, you have leadership qualities and those are roughly independent things, right? So sometimes the CTO can build stuff or the first engineer can build stuff, but he can't articulate the business plan very well.

And maybe he doesn't have leadership qualities and you, you could shuffle around like the CEO usually can't build anything, right? So, so, yeah. Yeah. I think you have three very independent qualities that are unusual.

Sahil Lavingia: And I would say like most people who I know who've raised, you know, that, that age do it with a team. and I did it, I did it completely by myself. I had no team, no co-founders. I didn't know a single person in Silicon Valley. Less than a year before that.

Steve Hsu: But were you, at that age, as articulate as you are now and as able to talk to, you know, super powerful venture capitalists or did that, was there a little bit of a learning curve for you?

Sahil Lavingia: I would love to think that I'm better now, but honestly, sometimes I go back to those emails that I wrote, and I'm like, holy crap. I was way more articulate in a, in a, in, and this just, and, and to be honest, like I, I, again, I sort of feel similarly today where like people send me emails and they're like friends of mine and they say, hey, Sahil, you know, like, just like, there's just so much fluff in, in human discourse.

And for whatever reason, I just never signed up for that and, or try, you know, try not to. And, and so I've all looked back at these emails and I'm like, wow. I was emailing like Mark Cuban. An email with just the word no in it, you know, like, or whatever. Like, I, I just have, and, and, and to be honest, I'll tell you where that comes from because I, I actually have unpacked this quite a bit.

And maybe it answers for the humble, grounded stuff too, a little bit, which is, all men are created equal. And I think that has just my parents ingrained that into me so deeply, which I think is part, part of why, like, you know, now the conversation about privilege is quite prevalent, but I like didn't know that word, even though I certainly grew up, you know, in a sort of a top, you know, sort of a quartile family in, in, in Singapore, blah, blah, blah, all that kind of stuff.

But like, yeah, I just think they, they never would. I ever expected that I was better, smarter, harder, working like, than anybody else. I was just completely in the middle of the bell curve on everything. And even when I got to, you know, even like, even after, even, you know, I think I had a conversation with a friend that I would consider a mentor, and this is like 2020, and he was like, look at your track record.

Like, you cannot, you have to stop thinking that you're in the middle of the bell curve. And I'm like, well, I don't know if I can like, I, I feel like that's so, like, it's almost a religion to me, of like, no, that's like, that's what I signed up for, you know? That's why I'm here. and so anyway, I forget exactly why I bring that up.

But, yeah, I just, I've always felt like part, part of the reason I think I can communicate with people articulately is because I think of them like myself, you know? And I'm going to treat you like I would treat myself and I treat myself decently well. And a lot of people don't. A lot of people talk up, they talk down, they, you know, they, they, they change the way they speak no matter who they talk to.

And I'm sure I do that to some degree too. Like, talking to my grandma's a little bit different than talking to like, you know, somebody else. But I think at the end of the day, like, I really think, just like I treat everyone, I try to treat everybody like an equal. And so Max Levchin is just like, hey, it's so cool that you built PayPal.

Like, what the hell? You know, that's awesome. And that just like, that's just like how I talk to everybody, you know? Like if I met Obama, I would, you know, I'd be like, wow, like your president. Like, that's kind of nuts because at the end of the day, everyone knows, everyone else is freaking out in their head, right?

Like, like the first night you sleep in the White House, like no one is, no one thinks that I hope, and no one is like, oh yeah, finally, you know, my day has come, you know, like everyone should be freaking out. Like, life is crazy. Life is super weird. But yeah, anyway, I really think that's important.

You know, even when I, you know, a big moment for me was like hiring people when I was 19. You know, I'd hire 40 year olds, managing a team, you know, at, and like, yeah, it was just like, that forced me to consider that, you know, like I have to treat everybody like an equal, because if they don't treat me like an equal, I'm going to lose, you know. We have to establish this early here, culturally.

Sahil Lavingia: So yeah, I've always been a big fan of that, and I think tech generally is pretty good about that. I guess the word for it is meritocracy, right? Which is like the idea, sort of the best idea should win, the most competent person at whatever skills should sort of be in charge of that.

But now I think meritocracy sometimes has a negative slant to it for some reason.

Steve Hsu: Yeah, I can go too far, but it's refreshing that Silicon Valley still has a kind of core value meritocracy. I want to go through a little bit of just the mechanics, the numbers of how it worked out for you for Gumroad, and you've talked about this elsewhere. So let me, let me just try to summarize it for my audience and then you, you tell me what I get wrong.

So, you raise an A round, which was seven or 8 million and started building the company, but the growth while healthy was not what venture capitalists really looked for. If they're trying to say build a unicorn. And so, when it came time to raise your B round, you felt upon advice from others that, you know, might be difficult because you're, you at that point, you had enough of a track record, maybe a year or two had gone by where, the growth really wasn't where the VCs really would want it to be.

And so, for your B round, you took a couple million more, but the investors, and again, it was maybe led by Kleiner, took a four-x preference, which for the audience means that if the company is sold, they're guaranteed to recover the investors four x on that amount. That's 2 million that they put in in the B round.

So that creates a hurdle if you're tracking this seven or $8 million in the A round and then another eight when you include the preference. So, Sahil had a hurdle of $16 million or so, which the preferred shares the investors in the company were entitled to before he would see a penny if he were to sell the company.

Are those numbers about, right?

Sahil Lavingia: Exactly. Yeah. We had about 16 and a half in, in preferences and I think at the time Gumroad was maybe at a million and a half or 2 million in, in total sort of revenue. And barely making profits on that. Because, you know, half the fees kind of went to Stripe and PayPal. So yeah, it was, it was like, yeah, I don't, I have no idea when the, you know, when this, you know, I think we, we had an acquisition offer for about a million bucks, from somebody.

So maybe, you know, a path to be, you know, selling for 16 million, but then, you know, yeah, all that money goes right back to investors and so yeah, it was not, it was not fun to be in that tunnel.

Steve Hsu: So, at that point you decided though you weren't going to try to wind up the company or liquidate it, you were going to try to continue running it. And for that, you know, in order to do that, you had to lay off a lot of your team.

Sahil Lavingia: Exactly. Yeah. I think this is also like one of those decisions that is not taken, or one of the paths that's not taken super frequently. Most of the time folks will either sell the business, you know, VCs invest in lots of companies. So, VCs will often help do this. They'll help sort of sell a fledgling company into a more mature company.

For you, you shut down, you return the, you know, money to investors, investors come right back and say, here's some more money if you want it to go, start a new company. because that's kind of really what they want, right, if they want more bets that they can take. And if you think, you know, if they think you're a great founder working on a b company like the, you know, they kind of want you to start again.

And yeah, I just said look, like at the end of the day, like Gumroad is, it's working, it's driving revenue. It's a valuable product to tens of thousands of people. Like whom am I to just turn that off? but yeah, it's not growing super-fast. We can't really afford a big team. And so, yeah, the only thing I felt I could do was lean out the staff.

We went from about 20 people down to three or four. Eventually just down to me. I started hiring only contractors and I basically said, I'm going to run this thing. indefinitely, right? It's just going to keep growing. Presumably it's like nice, lovely software. It mostly runs itself. I just have to have a couple support people, someone doing fraud and risk.

We'll hire, you know, a contract designer, a couple engineers, and, you know, T B D, right? What happens, we still have 16 million in preferences. and then everything really changed, I think it was December 2017. So, this is like two and a half years of just doing that, right? Kleiner emails and says, hey, you know, we were interested in buying, basically selling our position back to you for a dollar, $1.

And I said, you know, are there any strings attached? And they said, Nope. Then I, you know, I didn't really even ask why, to be honest. I didn't want to sort of bite the mouth or bite the hand that feeds. And so, I said, yeah, sounds good. And they ended up writing off their investment. I'm sure they, you know, got some nice tax, write us out of it.

Sahil Lavingia: But basically that, you know, changed the game because basically that preference stack, as you mentioned, went from 16 and a half down to two and a half, I think, something like that. It drastically, you know, changed the picture because of that, that four x that they were a part of, and then the original $7 million that they put in.

And so, all of a sudden, all literally almost like an overnight [unclear] moment, you know, go all of a sudden looked like a, a valuable company instead of a, you know, a basically a, a, a worthless company that was still providing a lot of value to the, to, to the actual people using the product.

Steve Hsu: Yeah, I mean today it sounds like the guys who held onto that, whatever that stake was that. Recliners to sell you for a buck, those guys could be quite happy. I'm guessing Kleiner, you know, I can think of multiple reasons why they'd want to dispose of the thing, because one, you know, partner time is super valuable, so you don't want to have partners worrying about some investment that probably is not going to turn out anyway.

They get a tax deduction, as you said, and they may have to wind up a fund and actually like issue a, a P&L for that fund, you know, an ROI for that fund. So, there are lots of reasons for them to do that. It was great for you, of course, but by persisting you've now built it into a, you know, it's not a, it's not a unicorn, but it's a, it's a super successful company as far as I understand.

Sahil Lavingia: Yeah. We did a crowdfunding round in 2021. We were raised at a $100 million valuation. More, more, more importantly, we do about $200 million in GMV for our creators. We do about $12 million in revenue off of that $200 million. And we're profitable, you know, well we're, we're trying to actually work through issuing our first dividend to our investors next year.

So, we'll see how that goes. Quite an unusual thing. Another, you know, that's kind of the. Step on my, you know, unusual journey with Gumroad is, you know, doing the boring thing of issuing dividends and seeing if that actually works for us.

Steve Hsu: Yeah. I wanted to ask you about the crowd fundraise, but it seems like with your numbers, you don't really need it. Right? So, the law changed to allow non-accredited investors to participate in ventures, and I think the limit which you hit is 5 million. You could raise 5 million a year through that kind of route.

Steve Hsu: But do you really need the money now? It seems like you're, you're profitable.

Sahil Lavingia: Yeah, I mean, we don't, we don't, we don't need the money. and we didn't either. I think for us it was about showing people, Hey, this new thing exists. A lot of people wanted to invest in Gumroad, and so, you know, why not use this tool to allow people to do that? And we'll sort of be able to hire 10 more people and build out the product faster than we originally anticipated.

Sahil Lavingia: And, you know, selling 5% of the company in order to kind of take that bet, I think, felt worth it to us. And going forward, yeah, I think, I think we're going to see if we can figure out this dividend thing and if we can, the goal is to sort of start returning in, you know, capital to investors, right? like, investments should eventually.

And yeah, we'll, we'll see how that goes. I, I think generally startups don't do that because interest rates have been so low, that, you know, there's just been too much incentive on a sort of evaluation secondary sales i p o basis, like it just dividends would never compare right.

Sahil Lavingia: To just sell a small percent of your, of the stock that you own. And that I think has, you know, has changed quite dramatically. and we'll see what happens in 2023, but if the interest rates continue to go up, you know, 5% or more, I think people will, will start to say, Hey, maybe we should turn these businesses into.

Sahil Lavingia: You know, profitable businesses, right? The, the, it doesn't make sense to, to pay $400,000 a year for, you know, a thousand engineers just because you can, because at the end of the day, like they're not actually driving any additional growth, right? They're just, they're just being spent on, on, on meta stock instead of, you know, solving, you know, all the problems that we see day to day that software engineers could make a large impact on.

Sahil Lavingia: Right? Societally.

Steve Hsu: So, you did that crowdfunded raise in 2021. and, for one of the startups that I founded, that's, that's going right now, I'm getting a ton of emails from different platforms that do this crowdfunding thing, and I honestly, I didn't really know much about it, until I, you know, actually answered one of the emails and had a conversation with, the, the guy representing the platform.

Steve Hsu: But what do, how do you feel about that as a viable alternative to.

Sahil Lavingia: Yeah, I mean, at the, at the end of the day, historically venture raising venture capital has become so easy operationally, right? In terms of it doesn't take a lot of time, it doesn't take a lot of money. They're effectively all standardized documents. and interest rates have been so low for people, there just hasn't been an issue in terms of raising capital privately from accredited investors.

That, so crowdfunding has, has sort of been the, the sort of ugly duckling, I guess, where people generally have a tendency to ask, like, why are you doing that? You know, like, what's wrong with your business that you can't raise money? The other easier, cheaper way, from better, more famous people or whatnot.

And yeah. But I'm hopeful that over time people will realize that it's not about just raising from professional investors. It's about allowing. the average person to have equity in a business, right? At the end of the day, equity is how people get rich. People complain all the time about inequality, diversity, equity, and inclusion.

Sahil Lavingia: And I just sort of think about, well, the answer is equity. It's in the word, you know, phrase like, if everyone has equity in Amazon and Tesla, guess what? Those people get rich. and should we force people? No, probably not. But how do we encourage more and more people to actually own equity in businesses?

Sahil Lavingia: Businesses, they work for businesses, they can invest in their local coffee shop, et cetera. You know, I think, I think that world seems pretty interesting and compelling to me. It doesn't seem like a world that, you know, has tons of lottery tickets turning into billion-dollar outcomes. But you know, in terms of, again, getting 4, 5, 6, 7, 10% return, I think it would lead to a much more stable growth sort of mechanism for society.

And so, yeah, I think it's still new. I think I'm hoping that it gets easier to do. I'm excited to show people maybe a path from crowdfunding to dividends because that, I think, is a big question mark generally in startups. But certainly, if you're an average investor and you're like, why would I invest in a startup that makes no money, you know, and pays engineers a lot.

Sahil Lavingia: Like what, what? This doesn't seem like a good business. But Gumroad is different, right? If I say, Hey, you don't actually have to care about what we build, who we build, how big the team is, all you have to care about is free cash flow, right? And you can decide what your discount rate is, just like you would a traditional investment and to say, hey, gum reds worth a hundred million, they're issuing 5 million a year to their investors.

Sahil Lavingia: You know, I want some of that, right. And then, boom, the valuation goes to 120, or, ah, I don't want that. Valuation goes to 75. Right? And so, I just think basically that's what I think, like that price discovery, like, I think there's a lot of value to that, to that liquidity, to that knowledge and insight and ultimately to like the shared alignment, that just very, very, very, very few people get to participate in.

Steve Hsu: Right? Are you in the secondary market? So, are your shares actually liquid right now?

Sahil Lavingia: They're not, or you know, they're technically, legally allowed to be liquid so people can sell them to each other. and that's allowed, and we've, you know, we've done six figures worth of those transactions in the last six months as people want to buy and want to sell. So, we kind of do them OTC. But there isn't a formal secondary market, which I think is a big opportunity actually.

Sahil Lavingia: I think, my guess is in the next five to 10 years, if Gumroad is able to show folks, hey, dividends work and this, and. It probably doesn't make sense to go listed on the Nasdaq. but maybe there's a way to build in like, you know, a yearly secondary event or, you know, buybacks by the company or, or, or other mechanisms that allow people to exit their position.

Because that will be, I think, really, really important. And there may be people who are like, I don't want dividends, you know, cash me out. And they should be able to do that.

Steve Hsu: One of the, one of the platforms that contacted me does actually have a secondary market, and I think they, they make their money from a, I think it's. Probably something like a 3% transaction fee for trading of the shares on that secondary market. So, it's kind of, they're kind of building toward all this stuff.

Sahil Lavingia: Yeah. It's coming. And I think, you know, that's one of the nice things about a bear market is people will just get back to building a little bit more and it takes time. So, it's not going to be evident immediately. But I'm sure there's, you know, a lot of people building a lot of cool stuff right now.

Steve Hsu: I wanted to, we have about 20 minutes left, so I wanted to shift gears to something else, which, you know, is part of the reason why I discovered who you are. I actually kind of knew what Gumroad was, although not really that well. Mainly I think because of some podcasts that I listened to, you know, having, you know, preferred content for people who subscribe via Gumroad or something like this.

Steve Hsu: So, I knew what you guys were, but I didn't really know much about you personally. And the way I actually discovered you is that one of the new startups that I'm working on has to do with large language models and trying to focus them on a particular corpus and try to force them to confine their answers to a particular corpus.

Steve Hsu: And I discovered a, I think it was a video of you building something like this based on your book, your book, the Minimalist Entrepreneur. I think you were using the Open AI APIs, and you had built a question-and-answer query engine, which was in our language, focused on your book, which was the corpus.

Steve Hsu: Can you talk a little bit about that?

Sahil Lavingia: Totally. So, like a lot of people, I've been tracking all the AI stuff in the last few years. The generative AI stuff, you know, starting I guess with GPT-3 was probably when I was like, okay, this seems to, this seems different than, than what I saw before. Maybe there's some commercial value, not yet, but getting close and, and then Dolly and, and all that kind of stuff.

Sahil Lavingia: And yeah, I just felt like I wanted to build stuff, you know, I like building stuff. I still do. And it, it sort of, AI got to a point where I could, you know, similar to pre-App store, I couldn't accept payments on the internet, but the app store made it possible for me to build an app and, you know, just upload it.

Sahil Lavingia: You know, that's all I had to do. And it worked, it, you know, people could buy it and download on their phone, et cetera. And so, I feel like AI got to this, this year where a developer like me who isn't super sophisticated in terms of like, you know, backend computer science data structures, algorithms type stuff, but can like hack together some Python and some HTML and some CSS on the JavaScript.

Sahil Lavingia: And I was like, wow, that's a moment. You know, like I remember this happening with payments on the internet and that led to Stripe. So, you know, I should pay attention to this feeling I have. And so, I started building stuff, sort of almost like a metal detector, right? Like, if I want to build something interesting in ai, it'll lead me to all the tools that, you know, I will use.

Sahil Lavingia: And then I'll email all those founders and ask to be on the capable and blah, blah, blah. And so, you know, this was kind of mulling around in my head and I had some ideas. Actually, the first thing I built was a. Was a thing trained on, images from progress, pictures like weight loss progress, pictures from Reddit.

Sahil Lavingia: And I wanted to train, fine-tune a, a stable diffusion model that would basically generate progress pictures so you could effectively see your weight loss before it happens. and if you've tried fine-tuning a model, like you probably are listening to this being like, well, this guy's an idiot. Like, how would he think that is good enough to be able to do something like that?

And the point is like, you know, I knew that I, I was 99% sure it was not going to be good, but I would have to do it. You know, the cool thing about AI is that it's not deterministic. And so, the only way I'm going to know how bad it is to actually just go through the exercise. And I was actually surprised at how it was bad.

It certainly isn't functionally useful, you know, to, to like to look at yourself, you know, and, and then, you know, it's not accurate at all. But it certainly was accurate enough where I could say a hundred pounds versus 500 pounds and it would do, it would kind of, Get a sense of, of that. but anyway, I started just, just hacking on that.

And then I had this idea where I wrote this book, the Minimalist Entrepreneur came out on, on 20, 21. And, you know, it's done decently well. It's like selling 10,000 plus copies, et cetera, all that good stuff. but I just felt like, and I've, I've been feeling this for a while, where, where, like books don't feel right, like I love reading books, but most people don't read books.

Not that much, not that frequently. And when they do, it's like Catcher in the Rye or Harry Potter or something. And nothing wrong with those.

Steve Hsu: Well, you know, [10,000] copies sold is a very good outcome for a non-fiction book, but you know, you might reach people more, more people with one

Sahil Lavingia: Exactly. And so, I just, I felt like, you know, they're books are still great. I, you know, and I still read them, and people should continue to write them, but maybe there are different kinds of formats that, you know, we can explore. And I felt like one of those, and I saw this. I forget exactly how I saw it, but I, I sort of realized like, oh wait, like, you know, one issue with GPT-3 is that you can't, it's not focused, right?

It's like, it's just like you can't ask questions like a book, or something like that. You can ask if it happens, have read the book as part of, you know, it's training maybe, but, but, but not really. And so, what I discovered was this concept of embeddings, which blew my mind. which was just basically this idea that pretty simply I could take 50,000 words from my book and effectively find the most relevant 500 words.

And that means that that's enough that I can just put those words directly in the prompt and effectively, you know, all of the code is, all the sort of conditional logic is just actually part. Prompt window that the user doesn't actually see. But it basically says, you know, Sahil is, you know, the founder of Gum Road, he wrote a book, this is a question, answer It, right?

That sort of thing. but what you can do with embeddings is you can actually say, by the way, here's some context from the book that might be useful. Insert all of that in the context and then answer, you know, sort of have that, have the AI complete that prompt. And when I realized that insight, I was like, wait a second, this is an amazing use case for AI and chat.

This would make chatting about a subject interesting. And, and at that point, this is, you know, before chatting GPT. So now chatting with AI is popular again. But that was a pretty, I think, key insight I had, which was like, there's a better format for like, question answer. The question answer is, you know, really AIs become really good for question answers.

People don't know this yet. And so, I'm going to build a chat UI. you know, for my book basically to, to be able to talk to my book. and so, I built that into my It's, you know, like maybe 200, 300 lines of Python. it sort of takes the, the, the, the PDF of my manuscript, turns it into a CSV file of pages, and then takes that page’s CSV file, you know, creates embeddings on it using open AI's API, and then uses those embeddings to, you know, stuff the prompt.

And so, it's effectively like, sort of two, two things happening. But yeah, it's, it's, it's kind of magical. It's kind of magical when you realize like, oh wow. It's just like, it's a very simple thing, right? but it, it, it feels like you're sort of standing on the shoulders of giants, you know? It's like I built Google search without having to build Google search, you know, just by building the, the, the box, the input.

And so, it's pretty, it's pretty crazy. It's pretty, it's pretty cool. And I have some, you know, stuff, stuff I'm working on now that I think will be pretty interesting too. I'm happy to talk about, but. I do think AI's getting to that point where, you know, people can build consumer facing applications with it.

Which is important, right? Because at the end of the day, Google Assistant is great, but like you need a thousand engineers to do that, right? Or Alexa or Siri. but like, I'm just Joe Schmoe. I just, I'm one dude who likes to build stuff. and

Steve Hsu: We had a very, we had a very similar realization to yours pre also pre-ChatGPT and for the audience, let me just unpack a little bit what you just, because you just explained what you did. And by the way, if you, I'll put the link in the show notes, but if you're interested to see how this works, I would definitely.

Suggest you go to ask my book and see how well this works. It's only, as Sahil said, a few hundred lines of Python. I believe you chunked your book into page length chunks, and then through the embedding engine, you basically map that page of natural language into the word vector space. So, all of these large language models work by taking the semantic meaning of human natural language and mapping it to a vector in an abstract space of human concepts.

And so, is that right? Your chunk size was about one

Sahil Lavingia: Exactly.

Steve Hsu: Yeah. So, then, if you type in a query, it can find, it can take that natural language query, map that also into the embedding space, find the vector, which is closest, which has the smallest inner product with the vector corresponding to your query.

Among all those one-page vectors that came from his book. And then he can use that as a prompt, hidden prompt to the AI saying, Hey, take this into account and answer this guy's question. And you get extremely good results. I mean, I've played around with your, on your, with your, ask my book, functionality. It's extremely good.

Sahil Lavingia: It's, yeah, it's crazy. Good. And one thing that I think is so empowering about it is that it is almost like coding with natural language, right? Where you're sort of able to say like, if this, do that else, do this by actually do, by actually just writing it out. which is kind.. It's almost like a transformer, right?

Where you're able to effectively write pseudo code and then like its output, it sort of does what the proper code would have done if that code had been written.

Steve Hsu: Right. In a separate setting, like some listeners may know about Codex and co-pilot and stuff like this in like we're talking about tricks to allow you to ask questions and then the AI answers using information that's in Sahel's book. But the other thing you could do is just give pseudo code to the AI and it actually returns well formatted Python, you know, executable Python.

Sahil Lavingia: It's crazy. so, it's, it's almost, it's, yeah, it's very, very, very exciting to see AI get to a place where consumers will start to use it. Because what that means is that consumers will start to give AI all their data and reinforcement.. Learning will start to happen I think at like, you know, 1,000 x when you are of all these people starting to like upvote and downvote AI picks and all that kind of stuff.

And so that's going to be, I mean, I can see the feedback loop, right, or like, you launch an app, you get 10,000 users. The model's kind of crappy, but these people need it badly enough. It's the people who are the best and most forward thinking that probably will use this tech anyway. They're going to inform the model and then the new model comes out and all of a sudden, the model is like 50 times better and now works for everybody else too, you know?

Steve Hsu: Yeah, the key, the key is what you just said, that you're going to suddenly have huge corpuses of reinforcement learning data. which previously hadn't really been in, used in GPT-3, but was used in ChatGPT. There was a lot of r l stuff in ChatGPT, and I

Sahil Lavingia: And yeah, that's, that's what it leads to. I mean, that's pretty epic, right? I mean, the believability. I, I think the information, I think they said this where like the, the accuracy is roughly equivalent, which is why they don't sort of give it a four. Maybe they intended to make this four and they, but like, I think the human rating of the responses, like it just feels so much better.

Right? it just, it, it basically has like learned to like, speak like how you would expect a human, like kind of the librarian to kind of speak. which I also think is by the way, just like a fascinating, thing to think about, which is like, you know, you kind of mentioned my American accent, right? But like, what if like, like I kind of have an international accent.

It's like, what if this is like the accent of AI, right? like it's just the most monotone, boring, de-scented accent and. And you know if you're training in ai that's sort of generally, you know, general purpose. Like this is sort of like the, the, the mean, you know, the, and, and it sort of talks like that, right?

Like it sort of, for some reason it sort of reminds me of like talking to a college admissions counselor or a librarian or like, you know, we've kind of picked like the, the middle of the bell, you know, like we've picked a, a voice, and which is, which is kind of interesting. You know, it’s kind of

Steve Hsu: Well, I think, there's a lot of, in GPT specifically, I think there's a lot of Reddit and Twitter and you know, there, there is a certain influence of the corpus on what, what the thing actually sounds, sounds and feels like when you interact with it. The thing we're interested in, like one of the problems still with ChatGPT, is that it has the right feel. It does seem like you're chatting with a real human-like thing, but it will occasionally insert false information and say things to you authoritatively, which are just completely wrong. And probably because the original huge training corpus, you know, that, that, that tuned those hundred or 200 billion connections in the neural net, some of that information is not true.

And, and it would be better if you could force it to, when it makes authoritative statements draw from a focused corpus and ignore that earlier stuff that was used in its training, which isn't necessarily true. And so that, that's what we're working on now is, you know, because like I, I never had this experience when it was talk, I was talking to your book, but I could imagine it saying something about startups or, or, you know, financing or something that was just wrong and didn't come from your book. Right? That's, that's actually possible for it to do. Or at least ChatGPT will do stuff like that.

Sahil Lavingia: 100 percent. I totally, yeah, I can totally say stuff like, you know, saw, you know, yeah. You could, you could, you could get it to say things that are, that are not true. I try to avoid that, you know, I can, you can set the temperature and things like this, but yeah. At the end of the day and, yeah, you don't want that.

Steve Hsu: You can’t stop it from saying like, Sahil was very sad to drop out of the NASA astronaut program, but he really wanted to start his company or something.

Sahil Lavingia: Yeah, exactly.

Steve Hsu: You could probably could get it to say something like that.

Sahil Lavingia: He's the best bullshitter alive.

Steve Hsu: Yeah, exactly. So, if you can make it less of a good bullshitter, but actually better at going through everything Bloomberg knows about this particular new drug development, than an investment banker can use it as something that just reads all the things that he doesn't have time to read.

But it is an extremely useful tool for him in doing his job. And so that's what we're trying to do is make it more reliable.

Sahil Lavingia: That's awesome. Yeah, and I think, I don't know if transparency fits into this, but I've always thought that transparency has always appealed to me generally, like just being open about Gumroad and financials. And I think that's also kind of an opportunity that I think will unfold over the next several years is, is, you know, we, we may not be able to tell you exactly like where this information comes from all the.

But I do think being able to cite sources and say, Hey, this is kind of where we, you know, these are the top three URLs that we think we may have sourced this stuff from, or, you know, things like that. And, and certainly you can do it on like a, on that sort of focused corpus, right? Where you could say, Hey, page, this is from page 178, this, this is from page 271, et cetera.

And I do think that, I mean, like the, the, the key insight that I think I had with asking my book was it's actually a very simple productivity gain, which is that the book is 250 pages. And I am telling you based on your question, the page, right? Like, or the three pages or the four pages. And so, even if that's all that's really happening, that's still a massive productivity gain because you basically went from, you know, 250 pages to two, like a hundred x, right?

And so even if all you're doing then is still. Saying, you know what, you know, answer this question based on what's above and return the page number so the person can actually go read it or whatever. You're, it's still massively valuable, right? It's kind of like the glossary at the end of a book. We should, we should just completely stop printing.

It makes no sense. You know, it's like, so, and by the way, someone is, I, I learned this like with my book, like there's a job and like someone manually makes that list literally. And I'm like, that is an exact, you know, that should not be like the only reason that exists because you can't control f. Right. Like an actual

Steve Hsu: Yep.

Sahil Lavingia: But yeah, like that, that's such a sort of, sort of a, a vestigial, you know, tale in a way. and, and yeah, it did really make sense to me that, you know, you buy the book and if you, you know, have rights to that book, you should be able to ask it questions. Just like you can read the book and ask it questions, right. You just, you don't actually have to read.

Steve Hsu: But now imagine an app where, it scans a hundred books that you tell it that you want in the corpus that are on your hard drive. And it basically uses all of those. So, say it's a hundred books on data science or something. and you're a data scientist and you, and, and maybe in addition to a hundred papers, you have a hundred books.

You have like a thousand papers and that's your corpus. And now your little buddy, your data science AI buddy who's on your, in the app on your machine, basically navigates that literature for you. I think that I think there's a market for

Sahil Lavingia: I think it's very interesting and I've thought a little, like in the context of asking my book, I've thought a little bit about what would be really cool is, yeah, being able to ask multiple books at the same time, the same question, and maybe even figuring out, could you get these folks to argue , you know, Hey, Peter Thiel says this.

You say this. Go, you know, I think that's one thing that I really think there's going to be a huge unlocking is when people figure out how to get the AI to talk to humans, right? Because at the end of the day, TikTok and Netflix, they're appealing because you can just sit there, you know, you can just like, kind of conk out.

Sahil Lavingia: And that will always be a human desire to conk out. And, you know, I think right now it's just like reading books. It's still a lot of energy, you know, asking questions, having a chat that's getting better, it's getting more interesting. I assume that 10 or a hundred x more people are going to do that than read the books themselves once this stuff is really out there.

But even better, it would be like, tell me what, you know, like, based on my Twitter account, you know, tell me what's interesting from your book, right? Like, and then I can ask you a question, you know, like, it can, or ask me a question, you know, like, I think there's a, there's a lot more there that will start to happen, once we get the reinforcement learning.

Like once we get the layers of. You know, it's going to start to happen and it's going to be, I'm just so excited. because it feels like podcasts. It genuinely feels like there will be new formats, you know, that, that will have words. Like, we will have new nouns, you know, ask books or some, you know, like there will, there will be a new way of thinking about this sort of thing.

I'm very, very excited.

Steve Hsu: So, last question, because I know you got to stop. Do you ever think you'll start another company? Right now, you're doing investing, but do you, do you ever get the itch to be a founder again?

Sahil Lavingia: I don't know if I ever, I think if I started a new company, it would be because it is not software consumer software, like it requires a completely different team. I think I would, I think I would potentially co-found a company, like if there was somebody who I felt like, wow, this person probably like a scientist of some sort, needs help, needs a co-founder and like this, you know, the, the sort of the thing I want to work on for the next 10, 20 years.

Like if it's some crazy, you know, synthetic womb startup or, you know, something like that, I think I could get really excited about. but I don't know, I just love building stuff and I, I think I'm good at this certain thing, and I'm going to, you know, I can invest and hopefully that, you know, you know, satisfies my FOMO.

But yeah, I mean, I think at the end of the day, what I would love to do is figure out how can I eat my cake and have it too, you know, like, can I make Gumroad, you know, we're building a new product at Gumroad now, which is like, kind of like all the fun parts about building a company. You know, I don't have to hire anyone.

Sahil Lavingia: I don't have to fire anyone. I can just say, Hey, we're building a new product now. and so, you know, we'll, if that works, we'll spin it out and that, you know, may be a new company. and I might be the CEO of two things. And you know, Montgomery kind of becomes more of like an incubator lab type deal.

But, yeah, I just, I, I'm always just kind of thinking about like, do I need a company to do that? Right. because at the end of the day, that's just a legal concept, right? and, yeah, I don't, I don't know, but I, I'm, I'm, I'm open to it, but I also think, what I've learned in the la especially even with building stuff for the ai, what's so important is just, just solve your own problem, right?

And I think a lot of people, especially when markets were really good or like, I need to go start a company, you know, what can I go do? What can I solve? What skills do I have? And I think the best companies are kind of like toys. They're kind of like, Hey, I'm just going to solve my own problem. I'm going to make it better and better.

Oh crap, this VC wants to gimme a bunch of money, or, oh crap. Like Sony is like, we, you know, we want this to power, you know, this thing or, or whatnot, right? And, I don't know, I think randomness is always like a pretty key thing. when you hear stories sometimes it feels like it wasn't random, you know, it was all planned, right?

But, yeah, I don't know. I think I, I, I like to just sort of stay focused on just solving problems. And if it makes sense to start a company, then, you know, Doing that. I think luckily, like nowadays, starting a company is the easy part. Raising money and audience building and all that stuff. What's hard is like science, research, tech, engineering, you know, that's the stuff that I think is really hard.

Takes a lot of time and you don't need a company necessarily to do that, you know?

Steve Hsu: Well, as a guy who's primarily founded deep tech-type companies, I can tell you, yeah. It's a different situation than using off-the-shelf software to build a product. It's quite a different thing,

Sahil Lavingia: Yeah, I mean, just the idea that I could have, you know, built Gumroad in a weekend and then raised a million dollars off this like 20-hour project, right? Like, it's crazy. It shows what supply and demand look like for consumer sorts of software, product engineer type, early-stage CEOs. But I'd rather do that.

I'd rather build the thing and then raise the money because that's kind of the thing that I tend to build anyway, right? Like they don't. a lot of people, a lot of, you know, a lot of, staff or capital, you know, expenditure. They're sort of simple apps, you know? but yeah, I could totally see myself like, I'm, I'm not opposed to it.

I think a lot of people are kind of like, I'm done, you know, and I'm like, I'm definitely not done. Like, I will be working in some capacity, you know, on stuff until I die. Like, I, I just think it's, it's just so fun and rewarding and, yeah, it's what, it's what I do for fun, you know? Like it's, it's, it's, it's more fun than it's ever been, to be honest.

Like, reading patents for like one click checkout is pales com in comparison to reading AI papers where people are tearing like 2D text prompt, like text prompts into like 3D models, you know? So yeah, it's like a video game, you know? It's better than any video game that exists.

Steve Hsu: All right. Well, Sal, I know you've got to go, so let's just stop here. Thank you very much for your time. It's been fantastic. I really look forward to seeing some of the future projects that you work on.

Sahil Lavingia: Thank you. It was super fun to do this. I really enjoyed the chat.

Creators and Guests

Stephen Hsu
Stephen Hsu
Steve Hsu is Professor of Theoretical Physics and of Computational Mathematics, Science, and Engineering at Michigan State University.
© Steve Hsu - All Rights Reserved