A Product Market Fit Show | Startup Podcast for Founders
00:14:45 · Anurag Goel
[00:14:45] S0: Yeah. So one of the ways in which I started exploring these domains was, okay. Well, what are the big problems out there right now? And what can I do to help? Right? Like, it has to be the intersection of my talents and skills and interests and what the problem means to be solved. And, you know, by some measures, solving world peace is perhaps one of the biggest problems there is. But I'm not the right person. I just didn't think I was the right person to do that given everything that I'd done. And it wasn't interesting in a way that I would, I think, find, like, personally fulfilling on a day to day basis, even when things are not going well. And so that's what you have to think about. Okay. Are you working on something that you are excited by and will continue to work on even when things are not going well? And are you going to get up every day for years? Like, in the best case, many decades. But in the worst case, at least a few years, assuming you raise funding for enough of those years. It doesn't matter how successful they are on the outside. Pretty much every company is chaos on the inside when they're growing. Lots of chaos all the time. I saw this at Stripe. I see it at Render. You ask any hyper growth company right now, there's tons of chaos. I love how from the outside,
00:16:09 · Pablo Srugo
[00:16:09] S1: from the headlines, you always just think it's like a smooth especially if you haven't done it or you're just starting, like smooth up into the right, would love to be that and you would love to be that. But when you're in it, it feels very different. And sometimes when you grow so fast, your runways get tight and, you know, the the difference between making it really big and actually just failing might be closer than you think. It's it's a pretty crazy world inside there.
00:16:29 · Anurag Goel
[00:16:29] S0: Oh, yeah. When you look at all these companies that eventually make it, there are several times during their journeys when things could have gone very differently and they would have shut down. So going back to what I was doing in the year and a half, I was really trying to find that problem domain that was big enough, ambitious enough. Because if it's not big or ambitious, then you don't have to spend a lot of time on it. Right? Like, you you build it and you you're done. You sell the company, whatever. And if it is big enough, ambitious enough, then then actually spending more time on it can get you outsized rewards because your work compounds. And so every new thing, for example, that we add to render, now it's there. And now it's there for the next decade of or the next several decades. Everyone who uses Render will have every new feature we add. Right? Like, it's something that we keep building, and it there's a lot to build because also the way people develop applications keeps changing. And how do we help them regardless of how they're building their apps? Earlier, it was traditional apps. These days, it's AI apps. And we want to continue to enable the platform to support application developers no matter what it is that they're building. And not just application developers. These days, it's also agents. How do we make the platform work really well with the coding agents that every developer is using so they can do whatever it is that they're trying to do much faster? But anyway, so I was going through this. I looked at health care for a while, learned everything that I could about health care, and then realized that I was not gonna help solve health care through technology. And I wasn't really the kind of person who was happy to have, like, three year deal cycles with large hospitals. I just wanted people to start using the product immediately and get feedback and iterate on it and, like, continue to build great products and win on the basis of that. Healthcare is not that.
00:18:27 · Pablo Srugo
[00:18:27] S1: I also think, by the way, just to pause on that, like, I think that that's a pretty classic failure mode of some of the more ambitious founders, which is you obviously you start with the biggest problems. And I think somewhere where you can add value is a pretty classic thing. But I do think people forget a lot of times, or I've seen this before, technology often is not the solution. When you really dive deep, you realize, oh, there are other problems that tech might address this stuff on the edges. It's not gonna fix that. And so you're tackling a big problem, but only the small little pieces of it, and it's a mismatch.
00:18:57 · Anurag Goel, Pablo Srugo
[00:18:57] S0: Exactly. So I sadly stopped spending time on health care after a few months. And then I looked at a few other domains, but then another thing I looked at became interesting. This was very early when deep learning was just starting to become a thing. And there were people online who were trying to get their deep learning setups on AWS when there was just one GPU type available in the cloud, and AWS offered it. But you needed to set up your NVIDIA libraries and all the other libraries on top in a way that was very finicky. And so just especially if you were a data scientist trying to do this, it was pretty much impossible for you to do it on your own. And I was taking a course on deep learning online, and the the person who was teaching the course, he had, like, a lecture that spanned two or three weeks of work to get this GPU backed Jupyter notebook online so the data scientists could actually work on the things that they wanted to work on. And I thought that was a big waste of time for everyone, obviously. And so I ended up building something that gave everyone a single click Jupyter Notebook in the cloud backed by a GPU. This is just as part [00:20:13] S1: of the course? Like, you did it for the classmates sort of thing, or you did this as a as a product?
00:20:13 · Pablo Srugo, Anurag Goel
[00:20:13] S1: of the course? Like, you did it for the classmates sort of thing, or you did this as a as a product? [00:20:17] S0: I mean, I started building it for them, but I built it as an independent product. Got it. It actually took off. Again, through word-of-mouth because it was so easy to use, I quickly got to, like, 10,000 data scientists on the platform. Wow. But I realized, after spending some time on it, that I care more about the needs of application developers than those of data scientists. And so I could have continued to focus on data science infrastructure or just, you know, eventually, that would have become AI infrastructure and so on. For me, it felt very clear that the broader problem of just getting anything online was still unsolved in a lot of ways. [00:21:03] S1: Was that more personal or was that more market size driven, the the decision to to focus on application versus data science? I think [00:21:10] S0: it was more personal because I'd never been a data scientist myself, and I knew deeply what the issues were around application development. And, yes, the market size may have been bigger, but who knows? Because these days, the market size for AI infrastructure [00:21:30] S1: Yeah. It's gone pretty big.