Episode 36
A conversation with Chai Choudhary
Published · 1 hr 13 min
AI agents now write the code, review it and write the tests. Chai Choudhary, solo founder of Workers IO, is building the proof layer that shows whether any of it works. A conversation about why "nobody's buying code" anymore, the first company he sold too early, killing ideas fast, and why founders should start solo.
What Chai talks about
Pick a question to ask it of this episode. Answers come from the transcript.
Transcript
Transcribed automatically, so names and words may be wrong. Each time opens the video at that moment.
0:00Every single line of code now being generated. It's mostly generated by agents. It's being
0:04reviewed by agents. The tests are written by agents. People are losing touch with what happens
0:09in their code base. We find all kinds of interesting failures in our customers and they
0:13are like, oh, I didn't think of this. I was working with this company right out on deck, which I
0:19thought I would work for ten years. They ended up having a co-founder conflict and decided to sell
0:24the company. What's the reason that somebody should consider not being a solo founder? It can
0:28be a little more challenging if you are the only person who's building and then also talking to
0:33your customers. What's the case for solo founding? I've never felt that, oh, things would be easier if
0:39I had a co-founder. Now building anything is extremely easy. If anything, you need to just build
0:44less and really figure out if what you're building is the right thing. It's just really
0:50incredible to to see the growth and to see all of these folks who are migrating away from, you know,
0:56extremely well-funded competitors to to your company. But what is workers today. And let's let's
1:03go from there. What goes IO is building deterministic simulation environments to verify
1:08mission critical software. What we are doing is we are basically taking this problem of verifying
1:15software and converting into a compute problem. So a team doesn't have to like, think of all the
1:19different ways your system can fail, all the promises that they should be checking. We convert
1:24it into more of a compute problem, where you systematically explore the state space of your
1:28system and give the guarantees to your customer that these are the things that that will never
1:32fail or like are like fall apart in your system. Now to put that into a different it, put that
1:39differently. Um, you know, my, my understanding of it is that, um, essentially, you know, you've got these
1:45extremely complex code bases, uh, you're constantly with generative code right now. You're, you're
1:52pushing all of these major changes, you know, like the average PR is, you know, many more lines of
1:57code, maybe tens of thousands of lines of code, much more than it ever was traditionally. And part
2:04of this is that it is very hard to actually simulate all of the different sort of
2:10potentialities of, of what would actually happen when you make these big changes in a code base. Um,
2:16and it's sort of reliability kind of takes a hit. And sometimes this could be absolutely
2:22catastrophic. Is that is that where would you correct that? What would you say is wrong about
2:26that? Um, just so just so, you know, different, different types of people who are listening to
2:31this podcast can make sure that they understand sort of the gist of what you're working on. The AI
2:36agents writing most of the code is definitely, um, like a big driver of the new wave of formal
2:43verification and basically figuring out all the new ways of testing systems. But if you think
2:49about it, historically, a lot of people do not know how low level systems work because a lot of the
2:56like. The good engineers in the old days did a really good job of building them up so we can
3:01rely on them. So nobody today worries about, uh, is my CPU going to work or not? Right.
3:07And you don't have to understand how the CPU works to rely on it because it's verified and it
3:13works and it's survived like, like decades worth of workloads. Um, now is the first
3:20time I feel where people are slowly like going away from the understanding that they have of
3:27their own code base. So when this happens, you definitely need a way to still have the
3:34confidence that what you're giving to your customers will work, right. Like nobody's buying
3:38code like people are buying that. Um, if I'm going to, like, implement this database in my in my bank,
3:44it's going to handle all the workloads that we see and we're not going to lose data. And that's
3:50like basically one of the things we are trying to solve for that, Even if you do not understand how
3:55your code is written, you still have the confidence to go to your customer and give them
3:59the guarantees that they care about. It's a really interesting time. Just the idea of talking about
4:05we don't understand the code that was written it, that feels like something that, um, you know,
4:12wouldn't have been said, you know, uh, fairly recently. Right. So can you share a little bit more
4:18about kind of the reason that customers are coming to Workers IO now? Sort of, it seems it
4:24seems like they realize that there are these big risks associated with the way. You know, there are
4:30obviously incredible gains from from being able to build with AI, but there's also incredible
4:36risks that kind of are opening up that maybe weren't there before. Could you share a little bit
4:40more about sort of if somebody is coming to you, why are they coming to you? Is that is that part
4:44of it, or could you share a little bit more about kind of how they actually think about working
4:48with you? Yeah. So today most of the customers that we have are people building, um,
4:55a lot of the core infrastructure stuff, like new kinds of databases, like there would be some who
5:01are trying to validate a very, very specific part of their system, which is, um, kind of directly tied
5:08to the revenues that they have. I think people underestimate with software that software usually
5:13have like a very long half life. But when you are starting to write something, you don't know if
5:17this is going to be the one with a very long half life that will be like critical to everything
5:23that you do for a very long time. Um, and this is a kind of pattern that I see. A lot of these people
5:28started off thinking that this will go nowhere, but now their businesses are working, and then now
5:34they have like a lot of people depending on them using their products like millions of times a day.
5:38So every single edge case that would only happen in like, um, with a very little user, it would
5:45probably happen once a year. But now those problems are happening every single hour. So you
5:49need like a much more rigorous way to verify it. And that's the kind of, uh, guarantees and like, um,
5:56um, confidence that we give to our customers. Um, and I think another new thing that has started to
6:02happen now is that you can be at a relatively low scale. Um, but now even if you have like a
6:09few customers using a lot of agents to use your product, that itself kind of spikes the usage as
6:16if they are like millions of users using your product. So the bar, the floor for reliability and
6:22the kind of workloads that you can handle as a business has kind of like is rising because it's
6:29no longer the humans using the product. You don't need to like scale to a million people using the
6:35product, uh, million people using a product before you have to like, fix these challenges. It's just
6:39like now 100 a team of 100 people using 10,000 agents could still, like, give you that amount of,
6:46uh. Which is why, like, a lot of the GitHub outage is happening, right? Um, people never really could
6:52operate at a scale and like, push out code as much as they do today. It's not like the engineers have
6:59increased like it's, if anything, like less number of engineers, like actually working today,
7:05shipping like ten times the code, which now changes how you need to think about your systems.
7:11Um, the kind of like reliability and, um, uh, like the workloads that it can handle. You have to
7:18think about a lot of them very deeply when you start working on some of these things. Hey there.
7:23We hope you're enjoying this episode of the Solo Founders Podcast. If you haven't already
7:28subscribed, you should make sure to do so so that you can get every episode as we release them each
7:33week. And of course, we always appreciate it. If you share the Solo Founders podcast with friends, it's
7:39what causes this show to grow and help more people. If you are already building a startup or
7:44considering building a startup solo, you should apply for the Solo Founders Program. We bring solo
7:49founders to San Francisco to build solo together for three months. We work closely with each
7:55company and we invest $100,000 as well. You can learn more at Solo Founders.
8:02And without further ado, let's get back into the Solo Founders podcast. You know, I've been building
8:08on deck off and now solo founders for about 70 years. The amount of
8:14customer pull that I'm seeing with workers, IO is really some of the most, you know, the strongest
8:21customer pull that I've seen in 7 or 8 years. Why do you think that is? Like why, why and why are
8:27people moving away from, you know, folks that they maybe are in a, you know, an annual contract with
8:34and deciding to pay for you even though they're already committed to a, you know, a competitor
8:38because they, they think this is just so important to them. What what is it that's actually driving
8:43that behavior right now. Yeah, I think there were a few decisions that I made early on, which now, in
8:50retrospect, I can say that probably that's what is driving some of it. Um, number one, we really wanted
8:56the product to be extremely easy to use, right? You wouldn't want only the smart, smartest engineer
9:03in your team to be using a product like, um. So that was one. Um, the second was how we priced it.
9:10I think, like, initially, I got a lot of suggestion around pricing it as at par with, like, what the
9:16competitor charges. Um, but then at the same time, I was like, okay, if we make it easy enough to use
9:22and like and reasonably price, it's just going to be used way more often in these teams. Um, and then
9:29it'll also be much more difficult to replace it, but we'll end up generating way more revenue. So
9:35those were the few things like also, um, I remember like initially when I started working on this,
9:41like we ended up building a lot more for our customer, and that kind of increase the surface
9:48area that they had to like, learn to start using the product. And slowly we realized, okay, like we
9:53need to shrink that down and basically build a lot of the tooling for ourselves so that we can
9:58make it a very hands off experience, um, where you just connect your product, you define the promises
10:04that you care about, and you get get the guarantees based on the amount of compute you
10:08spend. Um, and yeah, so basically a combination of a lot of these things, um, ended up adding
10:15to the now sales becoming much more easier than it was when I was just attempting formal
10:21verification stuff. You know, one thing that's also really interesting that you just talked about, I
10:26think it's worth us maybe pulling on this thread a little bit is this idea of kind of making it
10:33easier for customers, and by making it easier for customers, it's it's almost by doing the work for
10:38them, you know, instead of it being tools that they use, It's tools that you use. Um, and that really
10:44feels more like like we're talking about this sort of service model, uh, versus sort of your
10:48traditional SaaS. What have you learned about that over the last year of building? You know, for
10:55you've built for roughly the same types of people over the last year. And we're this this current
11:01thing that you're doing with workers, I o um, is clearly the one that's most resonant. But what
11:05have you learned about that idea of software versus service? Yeah, I think like when I was first
11:11building it, I even today, like I don't think of it as like a service thing. A lot of these decisions
11:18that we made were also for the fact that I wanted to understand what the challenges were when we
11:24were just focused on delivering the outcome. Um, like in this, in the early days, some of the
11:28customers would not share everything that what the bottlenecks are for their system to get the
11:34like the ideal output from the system. Um, so we were like, okay, we can use it for them initially.
11:40and then that will expose all the rough edges and then let the customer decide what more
11:47things that they want. And then now we incrementally expose. So like last week we
11:51released three more features in the system, uh, which is kind of a subset of everything that we
11:56had on day one. Initially. It's just we abstracted it all out and let like let
12:03people get comfortable with the with the shape of product today getting the same amount of value
12:08that they are getting, um, now and then incrementally expose some of these features. Uh, in
12:14that way it's like much more refined. And we are basically addressing the customer demand for what
12:19they need instead of just giving a bunch of features, and then they have to figure out, okay,
12:22now I need to like, use it all to just, um, get the value that I want. And it'll probably take me like
12:28two weeks to do it. It feels like, you know, that's that's such a great approach because there's so
12:33much to wade through. When you're getting onboarded to a product, the more you can strip
12:37away, you know, the better it is there. I think it would be awesome if we actually took this way
12:42back now. Um, you know, we'll talk about the journey, uh, of workers io, which is, I think, a really
12:48important journey. Uh, and a really good lesson for a lot of people, both in terms of persistence, but
12:54also in terms of, um, continually sort of moving towards the, the product and the idea that has the
13:00most resonance with folks. But maybe we just we just take this all the way back. Uh, you grew up in
13:06a really small town. And the reason, the reason I'm bringing this up is because I think that there
13:11are a lot of people who grow up in really small towns, and I think that it would be really great
13:16for them to hear about that experience. Um, and sort of how you ended up becoming a founder. I
13:22mean, this isn't your first company that you become so or you started, so maybe you can talk
13:26about that. What was the name of your town? Yeah. People who are watching this. I he's laughing
13:32because he knows I'm kind of planting this. Uh, his is a really funny story and an interesting story.
13:36What is the name of your town? And, um, when did you first start, like, building things either with
13:42computers or working on things that were creative pursuits? Yeah. So the the name of the village
13:49or a settlement is called 53 LMP is it's a small village in Rajasthan in
13:55India. Um, and the way, uh, the naming of villages work in India, like if is that
14:02if the settlement, the population count is less than 100. It's kind of named after the canal that
14:07passes by because there are so many of them. Um, and 53 is like the intersection and the LNP's the
14:14name of the canal. Um, and that's where I pretty much grew up. My parents still lived there. Um,
14:20and coming back to when I started building things, I think like that's one of the distinctions that
14:27I have with, like other people, I would speak with a lot of people here and they're like, oh, I
14:31started coding at like eight, and then I'm the I was not one of them, I. I first got access to
14:38computer in school and I was already, I think, about 16. And a lot of
14:45the programing that we would do at that time would like we would write it on a paper and then
14:49we would get a like a, an hour of week where you could go to the lab and then you can run your
14:55things. So, so that's where it all started. And then yeah, I got, I developed some interest there. Um, I
15:02would, I would write like these small tools to like capture keystrokes. And that was my like fun
15:09experiment. So I would install it in the computer lab, uh, computers and then people. And then the
15:15next, after a week, I would go and I would go through all the things that people have done on
15:20that machine. So it's like a and I would tell all my friends and eventually, like they realized that
15:25I was doing it. And then, yeah, they cut off that one hour that I had. So you want you went to zero
15:31hours? Yeah, I went to zero Obviously, we're keeping with the paper. Yeah, but. But I was lucky that my,
15:36my uncle, um, eventually bought me a computer and then, uh, but we still didn't have internet. I
15:42remember, like, I was in, like, the 12th grade, and the fastest internet that was available was at
15:47one of my friends place, and it was, I think, two Mbps. And it was so bad it would take three days
15:54to download a movie. So, like, I didn't use internet till I was in college. Um, and I that's where
16:01like I started my first, first company. Um, it was it was a friend of mine who I had met on
16:08Facebook, um, and he was also into startups. And I was like, I kind of like, learned that there's this
16:14thing that you can do from him. Um, and, like, I didn't know how to speak at all, like, I was so bad
16:21at it. Um, and he came from this very fancy school in Delhi. Um, but he was kind
16:27enough to, like, like, basically suffer all the awkwardness and like all the
16:34poor communication skills that I had at that time and basically helped me get better at it. Um, and
16:41that's when we started the first company. It was a it was a media company. We realized, like a lot of
16:46people were like suddenly getting access to the internet. Uh, but they didn't have, like, anything to
16:52watch, which was, um, relevant for them. Uh, they would usually watch, like, uh, videos from the
16:58US or Europe. Um, and then we started making content which was more regional. Um, and I think
17:05like that was the first time. Like, things just worked. And I had this false idea of, like, how
17:10startup should be. I thought, like, you just start things and they just, like, work. Um, because I
17:15remember we, we, we created like, our first, uh, skit of sort, and it got like a
17:21200,000 viewers in a week, and it was all over the news and I thought, like, that's how every
17:28company started, which was not the case the next time I did it, it was very painful. So. So you start
17:34off with a media company and the timing couldn't have been more perfect. It sounds like in terms of
17:39what what people wanted and sort of the lack of, of it's, it's just sound really
17:46old talking about this. Right? I mean, the idea that there wasn't any content for, for people in India
17:51to watch online. Yeah. Um, but, uh, I think that I think it's really remarkable how you can time
17:57things and timing is actually really important. What else have you learned about timing when it
18:02comes to to building things? It feels like it feels like that there's an important sort of
18:07connection between that first company and what you're doing right now with workers. I o yeah. Just
18:12to add to the timing bit, I think I also saw the opposite side of it. So we got the timing right
18:19to go into the market and things worked out pretty well. But my co-founder and I were still
18:25not happy with this IP business, you have to keep creating it and it was not very sustainable. So we
18:32transitioned more into product side. We built this platform for building interactive videos. We did
18:37that for a while and then things didn't work as well. And we had this fresh
18:44like experience of things working really quickly. And then um, this was not that. And we thought we
18:50were doing a lot of things wrong. So we decided to sell the business right before internet became
18:55free in India. So the same thing that we sold for like, like pennies ended up becoming way more
19:02valuable. And now it's like one of the most, like, expensive IPS that exist.
19:09Um, and so timing. I think I've seen both sides like getting it right and then getting it wrong
19:16in a very, very, like, bad way. Really wrong. Yeah. And I think that it's really interesting because like
19:22essentially what happened, what you're saying just for the audience is you distribution just
19:27unlocked in a massive way. So you're your potential sort of audience or customer? Uh, or
19:33viewer, uh, those that amount 100,000 x more than that. Way more than that. Yeah. So
19:40so it just completely changed the entire dynamic. Um, and that changed the sort of the the value
19:46of the potential value of the business. Yes. Um, I think I'm guilty of still doing it. I
19:53think, like one thing a lot of founders probably do is when they are making some of these, like,
19:59very important decision B, like fundraising or like starting business. They think of like
20:03everything as, like this is as good as it ever will be. Um, and then you end up making a decision
20:09based on that instead of like, being a little more optimistic where things could go and and keeping
20:15that into account when you're making such an important decision. I think we made the very big
20:20mistake of selling it off, thinking this is it. This is where it's going to always be. Um. Even
20:26though there were, like, signs of it. We could have. We could have figured it out. Um, and after that, I
20:33was kind of, like, bummed. I was like, we did this bad thing. I think within a year, we realized we
20:38made a very big mistake. Um, at that time, I got an offer from Harvard Law School to work on a
20:45research project. I ended up coming to the US. After that, I didn't really start a company for
20:51like, a few years. I worked at a bunch of companies on deck, being one of them. Um, and,
20:58uh, I think I was I wanted to be in a better, better financial position. I was like, If I'm going
21:02to attempt it the next time, it doesn't matter if I raise money or not, I should be able to sustain
21:06myself for like, uh, like a decade to see things through. Um, and I just spent some years
21:13building up that war chest. No, um, I think that that's it's interesting because, you know, there's
21:19all sorts of different types of advice for different types of people. Some people just say, go
21:23and build the start up. You know, life is short. Um, other people I think, would would agree. And I
21:27think they would benefit from what what you're talking about, which is get yourself to a place
21:32where you actually have the financial stability to not have to worry about sort of short term
21:38things, because I feel like what are the biggest kind of risks is when you talk to a founder and
21:44they say, yeah, I really need to figure this out in the next six months or otherwise. I have to start
21:48looking for a job and I don't that's not, not not disrespect to them or something, but I think that
21:55that even that having that attitude generally potentially can lead to the company not working
22:00out. Because if you're already thinking about it, even if it's just a kernel of thought in the back
22:04of your mind, six months from now, you're going to need a job. It's already sort of, you know, sort of
22:09invading a little bit of the the mind space there. And I'm curious, like how you've thought about
22:15that. I mean, we could talk about sort of, you know, uh, the, the journey, um, through to where you are
22:21today with workers. But how did you think about that when you were getting started? Did you? Did
22:26you say, I'm setting myself sort of a definitive period of time? Would you originally got started?
22:30Did you? Do you say, you know, I will I will decide to keep going until X. I'm curious. Like what?
22:37What you had sort of in your mind when you actually I think that we, we were, we were
22:42messaging on email or something. Um, and you, you said, you know, you wanted to start a company. And I
22:48was really excited because everybody that had worked with you would be really excited about you
22:53starting a company. We all really loved working with you. But what what was actually going through
22:56your mind at that time when you decided this was the moment? Yeah. So I was working with this
23:01company right after on Dec, uh, which I thought I would work for ten years. They
23:08ended up having a co-founder conflict and decided to, like, sell the company. Um, and at the same time,
23:15I was, I was I had a number in mind. I was. I can be a little pessimistic. Um,
23:21I'm guilty of that. I think Romania would would, would say that. And when I was, I was thinking
23:27about, like, starting a business, I was like, everybody is going to say no to me. So how do I,
23:32like, do what I want to do when everybody just says no? Like every investor in the world says, no,
23:37we are not going to give you any money. And then I had a number in mind. I was like, okay, this is
23:41enough for me to like, pursue it for, um, long enough that I can see this thing through.
23:48And that's, that's the that's was one of the mistakes that we made with the previous company,
23:53um, where we put ourselves in like a very short window that if things didn't work out, regardless
24:00of how good the business is, it would just eventually shut down because of, like, external
24:05factors. And I didn't want that to happen again. And that's when I reached out to you. I had a
24:11few areas in mind which I was deeply passionate about. Um, travel and verification.
24:18It's just I worked at a startup, which was sort of in the code generation space. So I was like, I'm
24:25going to attempt the travel first. So just to like, have a fresh like right out of a of a
24:31co-founders conflict startup, I wanted like a I wanted to get away from that for a while. Um,
24:37so travel was very interesting. I also had like a reasonable amount of following in that area. I
24:42would always share about like all of these hacks. Um, and I was looking at, um, the travel industry
24:49from a few point of views. Um, one of them being I wanted to, um, build an AI travel
24:55agent, uh, kind of a setup. Uh, I worked with this, one of the largest banks in India, ICICI,
25:02to help them, um, launch a credit card product. And, and I saw that, like, a lot, like a lot
25:09of banks in India were, like, doing really well with, uh, the travel OTA business. Um, and I
25:16think that was pretty interesting. Um, and things didn't really work out. I pretty quickly
25:22realized I was very much a pricing game. If you're competing with like Booking.com, Expedia spending
25:28like billions of dollars worth of on on Google ads, um, trying to get like, um, um,
25:35like selling like a, like a commodity. At that point, it's like a grocery store. You have like two
25:41grocery stores. You you get like the same thing on both of them. Obviously, you'll go to the one which
25:45has cheaper stuff, like if they're right next to each other. Um, so I quickly realized I was not
25:51going to be a easy journey in any way, and I was not really excited after that point. Um, then I
25:57pursued this pursuit, this email client idea, for a while. Um, I saw, like, people were complaining about
26:03superhuman. Um, and AI was, like, getting really good at, um, automating different
26:10workloads. Um, and then I was like, okay, maybe I can build something in the in the knowledge workspace,
26:15build build this client that can do some of these like tedious work for a lot of these travel
26:20advisors that I was speaking with at that time. Again, I think I quickly realized there was not a
26:27not a very painful problem for a lot of these people. So they didn't have, um, like a they didn't
26:33want to pay as much as I expected them to. Um, and that's when I kind of was like, okay, I think, I
26:40think verification is what I understand. Well, I think is a big enough Tam. Um, and then I saw like,
26:45then I doubled down on it and I just needed to figure out what, what was the right shape of the
26:49product that that will work in this space. So I distinctly remember this because, uh, as you were
26:55sort of getting ready to be able to come to the United States as you were joining the Solo
27:00Founders Program, we first were started, we started talking about this email client for for travel
27:06agents. Um, and I just remember, uh, saying, I expect like this idea will
27:13change and thinking I hope this idea changes. Um, but but like just knowing that
27:20for some reason I just had a sense that you would figure it out. You would figure out the right
27:24thing. I'm trying to think about why I actually believe that, to be honest. Um, I think it was
27:30probably just because of the way you talked about it, and you didn't feel like the type of person
27:36who was kind of delusional about, like, a specific product or a specific
27:42solution. And I'm curious kind of how you think about it, because I think that in the end, we're
27:48about to start talking about this, but this idea of kind of continually pushing
27:54towards the thing that is actually, uh, a combination of something that you're really good
27:59at and that the world really needs, and that if you're successful with it, it really matters. I
28:05think that very often one of the biggest failure modes for startups is that, uh, founders both don't
28:11really hone in on something that they need, that the world really, or that they're really good at,
28:15that the world really needs. And that will really matter. I think that you're really good at that. Uh,
28:20what what do you think it is? Do you think it's just like you've seen enough times where it
28:24hasn't worked? Um, is there anything that you would sort of give as advice for people who are who are,
28:29you know, watching or listening to this? Because ultimately, I if I look at the last year and
28:35change of of you building this company, um, that is sort of the, the thing that I would highlight for
28:41people and I tell people about workers I o all the time and you. Yeah, you can correct me if I'm
28:46wrong, but like I remember telling you this part, which may have
28:52probably given you this thought that. Okay, um, I'll eventually figure it out, but, um, I remember
28:58telling you that I'm being very, very intentional about, like, if this has
29:04commercial value and if I see Enough like evidence that this is not going to work. I'm going
29:10to change it. Um, and even though I loved travel as a space, it was very evident after like
29:17a month or so of working on an idea that there were clear signs of this not being like
29:24a hugely commercially successful business. Um, and a lot of people can, like, ignore a lot of
29:30it. It can be very like personal. Like if you are married to the to the solution, which is what I
29:37definitely didn't want to do. I know there are a lot of great ideas in travel today. A lot of
29:41people can attempt. Um, but I never wanted to like, get like marry a shape of solution.
29:48And the same thing with, like, the software verification, right? Um, when I initially started,
29:53there were a couple of, I think, like 4 or 5 different angles. I tried to like, address this
29:58problem, but it was very, very clear from the from the get go for like some of these solutions that
30:04they were not the right approach. I could have told myself no. Eventually people will turn around,
30:08but I think that's really the case. Um, unless you are building something extremely novel.
30:16Um, but again, even then, the idea should be like, very easy and, like, accessible for a lot of your
30:21customers. So you, you, you know, you've lived in, uh, in the United States before. Like, you've worked at
30:28companies that are in the United States. But you came to I think this is the first time that you
30:32were building a company of your own in America. Uh, and in San Francisco. Can you recall sort of what
30:39it was like when you arrived? Um, for SFP, like Solo Founders program? Like what? What was that?
30:45What was that, like, first day like. Um. And do you remember who you met and sort of what that
30:50experience was like? I'm curious is I, I, I for me, I think it would be interesting to hear, but I think
30:56for other people as well. Well, I think I had some idea that I was going to be there. So I was
31:02already like kind of feeling like an imposter because I knew he. Him from online.
31:09He was very active, very energetic, um, building a lot of cool things. And
31:16I thought, okay, the other people are going to be just as good. Um, and then I met them and they were,
31:20like, very special in all kinds of different ways. So I think, like a first few days, I was a
31:27little overwhelmed because you don't really see that density of like very interesting people
31:34with who are very driven, very passionate about what they are building and really understand that
31:39the space that they are in, um, in like one room, I mean, you can obviously like go to some
31:44accelerators. There are like 200 people. Of course, there'll be like five who are exceptional. But
31:49this was like one of those spaces where like six people and like all of them are really good. Um, I
31:54think like that was the like thing that I distinctly remember. And it took me like a while
31:59to, uh, like, get used to that, uh, that feeling. Now, we actually recorded this sort of
32:05reunion episode with, with a few of the folks who, um, who were in your cohort. Um, so, folks, so people
32:12who are listening, if they haven't, they should go check that out and maybe we can link to it. But, um,
32:16I'm curious, uh, specifically as it relates to sort of pulling the thread on workers io
32:23uh, as you sort of went from, okay, I'm pursuing verification. I'm pursuing sort of, uh, code and
32:29engineering as sort of like the, the key area here not moving away from travel industry. What were
32:35some things that you've learned from other people in terms of how to think about your business? Um,
32:40both from your customers, but also from from people in solo founders? Yeah. Um, so
32:47at the very beginning, I was starting to pursue the formal verification idea.
32:55Um, formal verification is like field in mathematics where you model a system in a formal
33:00specification and then mathematically check it. Um. It's much more rigorous than writing a few
33:07tests. There were a few challenges with it. I think one of the key things being you have to maintain
33:13another artifact alongside code. So you need a formal specification along with code. And then
33:18that leaves room for like drift. So your formal specification can be outdated, your code can be
33:24outdated, or you have done a poor job of implementing what you have specified. I think like
33:29while I was working on it, I, I kind of got too deep into it that
33:36I stopped looking at it from my customers lens. Right? Um, I
33:43knew how like, rigorous it was, how well it worked. But I would be blinded by,
33:50like, the novelty and like, uh, the blinded by the technology that I never saw the human side of it.
33:56Who has to use it? And I think Ryan would oftentimes like, like, shake you and, like, tell you,
34:03like you need to. You need to like, stop selling this as, like a is a cool technology that can
34:09magically make everything, like, more, um, uh, like it's like, stop selling it as a tool
34:16and then, uh, do more of the like, look at it from the problems that it's solving with the outcome,
34:22outcome, uh, point of view. And I think like also Javier, like for Javier, I think he does a really
34:28good job at, um, uh, like posting about things and, and, um, like,
34:35like the developer education part is, like, phenomenal. He, he, he basically, if I think about
34:40memory, I don't see of anybody else. Like, nothing nobody else comes to my mind. So he's like very
34:44visible. Um, I think like those are the some of the things that I have to do for my business as well.
34:50You have to make the technology very approachable, um, for the customer. And then at the same time, you
34:55need to do a really good job of, like, um, uh, basically teaching all these developers about
35:01what this is and like, how it works and how they can benefit from it, which his team does like
35:05phenomenally well on on X and like other platforms. Yeah. It feels like, um, you know, founders,
35:10especially solo founders, need to, uh, they might not start off as great at a lot of things, but
35:17they need to get very proficient, if not great in many things. Yeah. Um, and part of that is about
35:23evaluating kind of where your strengths are and, and also where your weaknesses are and whether or
35:27not you need to hire for weaknesses, whether you need to just sort of like, get better at them. And
35:32I feel like, uh, in many ways, looking at other solo founders, seeing where their strengths are, uh, and
35:39realizing, oh, I should try and learn from how they do things. Um, is a great example of sort of
35:45putting things out there. And Nemanja is an example of being like very shameless in a, in a
35:51really optimistic and extremely optimistic. And, uh, and I feel like, you know, he's the type of person
35:57who makes you. And he's the founder of, uh, pump GTM, uh, dot com. Um, which I think, like
36:04everybody in solo founder's program is using now, uh, which is an incredible product for, for GTM, but
36:09the, um, the general idea is like, these people, these people all have different skills, all have
36:15different things you can learn from and be inspired by and take lessons from. And I feel like
36:20in many ways that is a thing that you can then take that and apply it to your business. Um, and
36:26I've seen you like on the content side of things, even though you, you created, you know, a media
36:30company in the past, doing doing content for your startup is a different thing than creating a
36:35media company. And you like it's clearly resonating. I think you told me, um, maybe yesterday
36:41that, you know, a lot of your customers are really coming from all the things you're putting out
36:46there at this point, right? So what have you learned about sort of putting things out for
36:51developers? Um, because I think this is something that, you know is going to be core to your
36:55business, and it's clear that you've taken inspiration from not just obviously people in
37:00Solo Founders program, but from just observing sort of what's happening in the world. Yeah, I
37:05think it's one of the reasons for that is also because every other channel has become extremely
37:10is becoming extremely noisy. So it's very a person in a business, you're like, ICP probably has
37:17some amount of attention that they can give to people outside of their core work that they do,
37:22and they're just like overwhelmed. So the the best way to grab that attention, very little that they
37:28have is to put out quality content. Um, it can be around, um, what
37:35you are building or like the problems that they are facing, but I made the mistake
37:42of delaying it for too long. I think you should just go out as soon as possible. Like
37:49if you don't have a solution, talk about the problem, right? Like, you can you can just do that
37:52and then that will establish you in the space and then eventually get these people
37:59to reach out to you. And that is, I would say, the number one thing that is working for us. I try to
38:05like post every other day. Um, a lot of it is not the best work I do, but, um, it kind of
38:11compounds people. People. I think, like, there's one founder who told me that, um, if people
38:18see you, um, often online, they think you're very successful. Uh, so, so you need to increase
38:25the, the visibility that you have for some of your ISPs, and then eventually they'll think that you
38:31are doing well and then reach out to you and want to be associated with you. For you. You are doing
38:36well, just to be clear. But but I think I think there's something to that. Right? Because, uh, when
38:41we started launching this podcast, for instance, just just to sort of, you know, give another
38:45example of this, you know, people said, oh, are you going to do it like every other week? We're like,
38:50no, no, we have to do it every week. Which is, by the way, really a lot like, you know, it's, uh, there's a
38:56lot of people involved in making this podcast happen. Um, and it's, you know, also, we have to get
39:03great guests all the time, but but, you know, it's, uh, it's one of those things where you're right,
39:08like, if you were on people's feeds, if you were in front of people all the time, um, they see that and
39:13they react accordingly, you know? And, and you want to be known for something I think is really
39:18important. I think the bar for this has, like, significantly increased. Now that you rightly said
39:24you need to do it often, you need to do it well, and a lot of things have to be done really, really
39:29well to really stand out from from the crowd. Because everything, as I said, like it's extremely
39:34noisy. Some people will be maxing out the volume, but like be very bad at the quality. Uh, so you
39:41need to like, make sure you're making the right trade offs and really do it like do more often,
39:46more like better every time, and then eventually that leads to good outcomes. It's it's that whole
39:53idea of like good, fast and cheap. But now, you know, you need to do everything to you. Can't you see
39:57you need all three, right. Yeah. Yeah. I mean, that's very true for like, even our business, I think like,
40:01I was initially thinking that you could maybe just have a great product and then that itself,
40:07people will be tolerant enough to stick with it long enough that they see the value. But now you
40:13need to make sure it's cheap enough, make sure it's also best in class, and then you need to
40:18deliver the value very, very quickly. I think like we need to do all three well to be distinguished
40:23from our competitor. And yeah, like that is the only thing that has worked so far. It's really
40:28fascinating because you said, you know, in hindsight you would have gotten started faster or
40:33sooner with content. Um, but at the same time, it's it's amazing to see that as soon as you turn it
40:37on, if you have something that resonates with people, it resonates in a big way. And quickly. I
40:43mean, it's again, like to draw the parallel between what you're doing with with content and media and
40:48what we're doing. You know, we've only been around for about 6 or 7 months with this, with this
40:54podcast. Right. And, um, we have something like 14,000 subscribers on YouTube. I think we just
41:00broke 12,000 subscribers on Instagram. I never even use Instagram. And and it's and it's
41:05incredible to see that. But I think it's because, you know, if we had just sort of done a generic
41:09podcast that wasn't about solo founders, I don't think we'd have this resonance. You know, I think
41:13we have great guests and we could have great guests and things like that. But you look it,
41:17there's a new podcast launching every day. So it's like, if you can put something out there that's
41:21unique to your audience, that speaks to something that is underserved or under discussed and in
41:26kind of whatever, you know, space that you're in. Um, it is it is remarkable. I'm just shocked by how
41:33quickly this is caught on. Yeah. And, uh, and I think that, you know, now that I've had that happen for
41:38myself, I'm not shocked when that happens for other people. Yeah. Do you think the the like
41:44doubling down on a niche is the part that really helped because I've seen like it worked for one
41:50of the communities that we started about people who are interested in formal methods, and I was
41:56initially thinking that it won't be that active. We got like a few hundred people in it, and it's
42:01the like, I don't even have to contribute to it. Like, people are just so active. They do like they
42:06drive it. And I, I think like the only reason that happened for it was because it's like very
42:12specific about the, the, the purpose of that group. People were dying for specificity. I think
42:18actually, I think that everything is so generalized. Um, you know, people talk about going
42:23on Netflix and they go on Netflix and they maybe have some personalization, but at the end of the
42:28day, you're just scrolling all the time trying to find something that you actually want to watch.
42:32You don't even know. And and I think this is the same thing when it comes to specificity. It's like
42:37the more specific, the more personal, the more it feels like it's a fit for you. You know, also, I, I
42:43think the other question this is like when we're talking about kind of like, uh, you know, there's
42:47two things when you're looking at startups, right? It's like, do you believe the founder could
42:50actually make it happen? And if they actually make it happen, does it even matter? Right. So so that
42:55means, like, if I think we talked about this before we start recording, but if somebody, you know,
42:59started a hot dog stand, well it doesn't really matter as a business unless it's like, you know,
43:03becomes Nathan's or something becomes this massive, you know, chain. So ultimately, um, like,
43:09that's the question for startups is not only can the founder make it happen, but does it actually
43:12mean anything? And in your case or in our case with solo founders, you know, if if we believe that
43:19solo founders will always be this niche thing, well, then maybe it does or doesn't matter. I would
43:23argue it would matter for the individual solo founders, but I think that the answer is solo
43:28founder will be the default way that companies get started, right? I mean, we've already seen that
43:32the amount of solo founders, the percentage of solo founders has doubled in the last five years.
43:36Yeah. And it's not going to stop. It's going to continue to increase. In fact, it's going to
43:40increase faster than ever. So in your case, though, um, is formal verification this interest in that,
43:45is that going to be something that diminishes over time, or is it gonna be something that
43:49becomes more important over time? And I would argue, uh, given the amount of agents that are
43:54going to be hitting any sort of, any sort of code base, um, you know, that's only going to increase
44:00over time and therefore the amount of edge cases and potential error, like room for error is the
44:06surface area for that is going to become much larger. At least that's my perspective as a
44:11layperson. Um, so therefore it seems like something that maybe starts a small and niche could
44:16actually become dominant in the future. Is that is that kind of how you think about it or. But I
44:21think that in the beginning, though, it feels extremely resonant with people if they feel like
44:26they're early to it as well. Yeah. I think there are a couple of things to unpack here. Um,
44:33I think one of the drivers why it's becoming extremely important now is people are losing
44:38touch with what happens in their codebase, and the only way you can still have the guarantees and
44:45the confidence in your system is have some kind of proof of work. And formal
44:51verification or deterministic simulation. Testing gives you that proof of work that whatever you
44:57are putting out, it's not about code. Like nobody's buying code people are buying. If they're buying a
45:01database, a kind of database, they are buying the reliability that it will not lose our data. They
45:07are buying that it will always be up, and all the performance and availability guarantees that they
45:13offer. Um, so if anything like the entire market like software market
45:20that is, is going to be the market for verified software, right? Like people are not buying code.
45:26People are buying codes that can handle their, uh, they can solve their problems and handle their
45:31workloads. And and formal verification is a great way to do that because it not only
45:38does it for like a few cases, but it gives you guarantees of the system overall for like all
45:43possible cases that can happen in a system. Is there a world where we haven't talked about this,
45:48but is there a world? Obviously compliance and things like that. Those are about sort of rubber
45:53stamping something in a way. Right. Like SOC two. Um, and being able to tell people that you're SOC two
45:58compliant, uh, really makes a difference. Is there a world where people are like, we we are very we use,
46:05uh, workers io and this is and this is sort of a sign of reliability and trust, um, to sort of the
46:12external world. Is that is that something that you think will actually happen, or will people look at
46:16that that way? There has to be like, if, if you don't know what, like if you have not reviewed the
46:22code. Because if you think about like every single line of code now being generated, it's mostly
46:27generated by agents, it's being reviewed by agents. The tests are written by agents. If you ask an
46:33average engineer and if you ask them like what's in the test? They don't know what's in the test.
46:36Like, you know, there's a Lego. It's a green check. But. So if you're trying to go in the market and
46:42sell this thing that you have built, have no idea what's going on for the other person to trust it,
46:47there's there has to be a way like they you need to present the evidence of like the correctness
46:52or like that we have, um, the proof of work and formal verification is one of those ways to do
46:59that. Um, I still feel like it's, it's a very it's in the very, very early
47:05stages. Um, you cannot formally model a lot of the properties. So there has to be a lot of research
47:11work that needs to go into this field to basically get all these things that people care
47:16about, be part of these specifications. And like part of the proofs that that we give out, um, which
47:22is why we took like the DST deterministic simulation testing approach, which is like a
47:27little more pragmatic. You get just as good, um, guarantees about a system, but it works on your
47:33code that you have run. Uh, have you have written? All your agents have written. Uh, but but I
47:38definitely see it as a way of, like, um, this is the proof of work that that our system works. Our
47:44system can handle all the failure conditions and all the inputs that the that that that are
47:50possible. Uh, and it's it's a great, great way to give this evidence to your customers and build
47:55that trust. I think that, you know, what I when I work with folks, when I work with founders, and
48:00what do I think about who I want to work with in the solo founders program? You know, I realize the
48:05business is going to change so much over time. And so I don't I rarely think about, um, sort of the
48:11initial market that people are going after. I think about their judgment. And I think that you
48:15just have really incredible judgment. And, um, and let's, let's talk really briefly about this idea
48:21of pulling the thread because you started off, uh, you know, working in something similar to what
48:27you're doing now, but ultimately, you dialed into what you're doing now by pulling the thread and
48:30continually like Finding the connection to the the prior thing, but moving towards like the
48:36stronger pull the the higher value all of that I want I want to tell. Tell a little bit of the
48:43story of pulling the thread and then also kind of ultimately like how big does this get? Like how
48:48much does this matter? Because I think ultimately that's I think that we've pretty clearly
48:54established the case for chai, because chai is like this person who is incredibly diligent about
49:00solving real problems. Um, he he continues to go until he finds the thing that the customers
49:05really value. Um, you've gotten to a place now over a year, by the way, uh, as a solo founder,
49:12where, you know, I think that a lot of people, you know, if they didn't really have their heart in it
49:17or they were dedicated to the solution versus the problem, they would have either given up or they
49:23would have gotten stalled out on something that wasn't actually working, or it was sort of local.
49:28Maxima, I want to understand pulling the thread. And I also understand like how big you think this
49:33ultimately becomes, how important this ultimately becomes when when you when you're inevitably
49:37successful with this. Yeah, I think I had conviction in the problem space.
49:44I at no point I thought like this is a wrong problem to work on, because I firsthand saw in the
49:51previous company, um, that people would do all of these large re factors, right. Like large
49:57systems. But when it came to giving it out to their customers, it was always like the confidence,
50:04the amount of confidence that they have in this system. Um, and they would always try to like, they
50:11would always struggle with like, um, like how like can it work for all the things that we have seen
50:17over the years? So if you have like a database that's running for like 20 years and you go to
50:21them like, oh, there's a new database that you can use, it's better than it's more performant and it
50:27is well tested. And then they'll be like, oh, but my system has been running for 20 years. Have your
50:32system run for 20 years. That's the flaw for me. So how do you give them the confidence that this
50:38system can also survive for 20 years, like all the 20 years of workloads that your production system
50:44has seen? Um, so it was very clear that this problem existed, and that was one of the key
50:49bottlenecks for all the modernization that is yet to happen in the like financial, healthcare, even
50:55travel industry, some of the biggest industries in the world. Exactly. Yeah. Um, but I was still not
51:00sure what the shape of the solution is, which is why I started off with formal verification. Sure,
51:06it gave like, all the guarantees and all the rigor that you would want to, like, uh, to build
51:12confidence in your system. But it was, again, very hard to use. Like most people do not like, if you
51:18give them a formal specification, it looks like math problems and they'll be like, I don't know, I
51:22don't want to read this. So there were always like these few challenges when it came to getting the
51:29customer on board with the solution. Um, I mean, I could have just continued to hammer it.
51:35Like, I don't think that's the right, right approach to it. So then we started, like, experiment
51:41with other ideas and then eventually ended up with where we are at, where it's much more easier
51:48for the customer to get started with. It's it's in the form that they already understand, which looks
51:53like, uh, our code being run instead of like a formal specification being, uh, like used for, say,
51:59model checking. I think like if you are, if you have enough conviction in the problem, you can
52:05always eventually, um, like, you can be very intellectually honest with yourself and like,
52:11figure out the right shape of the solution. Um, instead of like, just giving up on the first thing
52:16that doesn't work. And then ultimately, like when you think about this, this is these are some of
52:20the biggest problems in some of the biggest, most highly regulated or at least, um, uh,
52:27highly risky, uh, industries to be to be messing around with core infrastructure. Right? Uh, this is
52:32why, you know, uh, like, like you said, a lot of the, uh, a lot of these industries are using fairly
52:38antiquated, uh, systems. Right. Because they've been working ish, uh, for a for a long
52:44time. Uh, but ultimately, we're going to start to see, uh, cracks in that infrastructure because of
52:51the scale at which we're going to be throwing changes or usage at them. I think just, you know,
52:55just recently you found all of these bugs in Postgres, which is, you know, it's the widest, most,
53:02most commonly used, uh, you know, uh, database. Uh, can you share a little bit more about kind of
53:09why this, these huge industries, these really important industries, uh, are going to be
53:15experiencing a a moment or a or a time where, you know, the the old way of doing things
53:22is just not going to hold up. They really need to adopt, you know, modern technologies also things
53:27like workers I o to to help them with that. Yeah I think like security is like one of those things
53:32that is now becoming very, very visible. Because earlier for you to like compromise the system at
53:38word. Like the the amount of capital and like like work it would need to compromise
53:45something. It didn't make it worth it. But now with like agents, you can really it's cheap enough to
53:52find like one small issue in a system and like take it hostage. Now for the defenders. Like
53:59or like you need to do a lot more work because you don't just have to patch that one thing. You
54:04need to patch every single thing that exists. Um, but at the same time, you also have this like, um,
54:11this new technology that you can use, um, to, to make your life much easier when you're doing it. I
54:18think, like with the Postgres thing, um, it was very interesting because, like, we were not actually
54:23testing Postgres, it was just part of the stack. Customer was using and we ended up finding issues
54:28in it. None of them were like very serious, but it shows like one of the most well tested systems
54:34out there, like this database that runs in like, probably handles a lot of important workloads in
54:41the world. Uh, has issues. Right. Um, and what does that say about your system? Like, you know, you
54:47probably also have, like, a lot of these, um, like, cracks that you need to, like, fix before some
54:54attacker who is like, uh, just looking for that one thing that they can compromise. So that's one
54:59thing. But I think at the same time, you also have this new world where, uh, the, the frequency of
55:04usage is, like, spiking up, like even a few customers using your product, um, can generate
55:10workloads that would only show up with, like, millions of users using your product. Um, so you
55:16need to think of this problem of, like, engineering from a very different angle. Now, the
55:23the amount of workloads that you will be handling has to be part of this problem that you are
55:29trying to solve. From very early on, you could afford to have like a very crappy solution
55:32earlier where, oh, it only works for ten users, but now those ten users can have like 10,000 agents,
55:38right? So even though you're serving just ten users but like, it still need to like handle all
55:43the workloads that they are going to give you. So a lot of these things are changing. But I think
55:47like one other thing that I often like, try to point out to people is that like you can have
55:53like very safe systems, like for example, you can take the safest database out there. You can take
56:00the safest authentication system out there. You can put it all together. But that doesn't
56:06necessarily mean that your entire system is safe, right? Like, you can have every component safe and
56:12still have like a very unsafe system based on the way you have composed it. Um, and a lot of
56:19these problems exist because so far people were just testing from a very like
56:25legacy, kind of like all these, um, like thinking of all the scenarios where you can run on your
56:31system and then find like failures, but and, or like there would be teams, individual individually,
56:38like testing smaller components and not test the whole thing. Yeah. So you need to like, spend a lot
56:43more time, um, into devising these and like adopting these a lot of these rigorous
56:48technologies to, to, to verify your systems. Do you have been a solo founder this entire time? You
56:54have experience, uh, having a co-founder with your first with your media company? Um, you've worked at
57:00companies that have had co-founders. I'm curious your reflection on co-founder versus solo founder
57:06and then also solo together, because ultimately you have been a solo founder, but also, you know,
57:12there's difference between being a solo founder kind of off on your own and being a solo founder
57:17all around other solo founders. In fact, I think that in terms of the Solo Founders program, I'm
57:22pretty sure that you came up with the idea of solo together or the term solo together, and you
57:26at least came up with the idea of it's like you have co-founders, but you're all working on your
57:30own things, so can you. Can you share a little bit more about your reflections on solo versus
57:34co-founders, sort of why you ended up being a solo founder and what sort of being solo together
57:39means? Yeah, I think it's it's very often that people would. Like whenever I would tell them that
57:44I'm a solo founder, they would look at like, oh, you're solo founder? I was like, but at the same
57:48time, I've never felt that, oh, things would be easier if I had a co-founder. Um, I think that's
57:54also because you can hire a lot of good talent, because the amount of equity you
58:01have, you can you can give out more meaningful percentage to some of these, like, exceptional
58:05engineers out there. And I think surrounding yourself with other solo founders is also
58:11extremely helpful. Um, I was lucky that like SFP kind of had a good mix of people at different
58:18stages. Um, which kind of gives you some visibility into how things are going to be
58:25for you eventually. Like, Ravi was like much further along. Um, I remember, um, Aldo was still
58:31figuring out, um, things. So I think, like, you get like, this very diverse set of insights and,
58:38and kind of like experience people are going through, and that helps a lot. Um, but at the same
58:43time, I think now building anything is extremely easy. Um, if anything, you need to just build less,
58:50uh, things and really figure out if what you're building is the right thing. Um, so at no
58:57point I feel like things would be easier for me if I had a co-founder. Um, I think the engineer
59:02that I hired, Vishwa, he's like an exceptional engineer, and I, the like, has taken off a lot of
59:09the work that I do on just purely engineering side. And I don't think I would have been able to
59:14hire him if I, if I had like two other co-founders with like very little equity that I had on table
59:19for, for a role like his. That's really cool. And what um, when you think about kind of staying the
59:26course and, you know, continuing on with this company, I'm curious kind of, you know, how you
59:32thought about that as a solo founder, because, again, you I think that in tech, it's kind of
59:38glamorized this idea that things just work nowadays. And there's certainly cases where
59:43somebody, you know, comes up with an idea and it just works out of the gate. And, you know, one
59:48company that that I've worked with is MagicSchool, and it really did just go from 100
59:53teachers the first week to a thousand. And now it's millions of teachers. But, you know, at the
59:58same time, that's just very rare. And I think that the better example is these companies where
60:02they've they've gone on for a really long time. Ali from micro1, like he was working on a bunch
60:07of different things before he actually hit on what micro1 is today and now it's ripping. But
60:13that was not sort of an overnight success. You got to a point similar to you where he found this
60:18inflection, and now things are just going crazy for workers. I oh yeah. But you had you had
60:23customers sort of the entire way. And I think this is a, this is a really important thing to to
60:28mention. Actually we didn't cover it before you. It's not like you were trying to get customers
60:33and you couldn't get customers. Yeah. You were consistently building things and getting
60:38customers and then throwing that, not throwing it away entirely, but continually, like saying this is
60:44not actually the stopping point. And Ali was kind of like this too, with with Bitcoin where he had
60:49customers along the way. But he realized that it was ultimately like he needed to keep going. Yeah.
60:54So so what was that from a solo founder perspective? How did you think about that? As I
60:59said, I think if you're if you're intellectually being honest with yourself and if you have like a
61:04few customers, it's very easy to tell if things are like progressing, like if a customer doesn't
61:09use a product every day. Obviously there's something wrong. Sure, they are paying you, but like,
61:13eventually they're going to churn and this business is not going to work out. You can be
61:16blinded by it. And just like put an MRI on it and say like, oh, things are great. Um, but even along
61:23the way, I had good visibility into what was going on with, like, these people who were paying. It was
61:29very obvious that, like, I couldn't continue the way that I was with like some of these solutions.
61:34Um, and I, I guess, like it can be a little, little more challenging if you are the only
61:41person who's building and then also talking to your customers. Um, but to that, I would say it's
61:48mostly just like talking to your customers, which is what you should be focusing on. The building
61:54part is like much easier. You can always make the right trade offs and like pick the pick a narrow
61:59solution initially and validate it. I think everybody is now way more tempted to build a lot
62:04more things. I I'm guilty of doing that. We Overbuilt way more than what we needed. And then
62:10we had to, like, basically abstract it all out. If anything that is going to like, make it super
62:15difficult for your initial customers to understand what you're doing. Um, and that makes
62:19the entire process even, like, harder than it should be. Um, so I think, like some of the things
62:26that you can always do is be very specific about the problem that you're solving and pick the
62:31minimum thing that you can do. And now you can with agents, you can like, do it as a service. Like
62:37I remember the first customer that we had, um, and I was showing him this dashboard, I was like, oh,
62:43you can, we can we can find all these failures in your database and we'll show them here. And then
62:47he was like, can you just give me an Excel sheet? And then and I was like, yeah, no, I'm going to just
62:52give it to this agent and like it'll help me like close them. And I was like, so we spent like a week
62:58building all of it and it was not even needed. Um, and people overestimate, like having a fancy
63:04dashboard or having a product in like some shape is not going to make the sales any easier. If
63:10anything, it'll just make it harder for you to explain things to them. So I guess like being very,
63:16very specific about the problem that you're solving and addressing that part well, and then
63:20having something very, very minimum and minimal and much easier to understand for these customers
63:26to like, once they see get the taste of the value that they are getting. And then then you can
63:30incrementally build from there. And I made like this mistake many times initially when I was, I
63:35was starting well, I mean it's amazing to see sort of the progress and the learnings because clearly
63:41it's working out now. It's it's it's incredible. Congrats on that. I guess like, uh, you know, a
63:45couple a couple final questions here. Um, we do not have a screen here, but imagine we're looking in
63:51between us and up here somewhere. We have we have an image and it's a photo of you and and your
63:57jacket says fixing the internet. Um, can you talk about sort of the problems with the internet and
64:04why? You know, ultimately, like fixing the internet is at a high level, uh, what workers is actually
64:09doing, I think I think this is we have like, collectively being like been
64:16conditioned that software fails. It can be flaky. You can just restart your thing and it works. I
64:22think that's a very like bad way to look at it. Um, I think software is definitely
64:29complex in the sense that, like, it doesn't have any of the physical constraints that other things
64:34have. And and you can, you can it makes it easier for you to like, build them well. But the
64:41amount of people who are frustrated when things go wrong is like being an engineer, like over the
64:48like, whatever. Ten years I've worked at a company, the only memories that I've had were like me, like
64:54fixing, like some outage that is happening at an odd hour and we somehow think that's
65:01okay to do. And your customers are suffering. And I think for the longest time, the panel, it
65:07didn't make financial sense to like, like adopt a lot of these new like ways of verifying
65:14systems. It was extremely expensive. But now that thing is flipping. Now you could you could have
65:20the exact same amount of cost, um, and get just as like high reliability as some of the
65:27most well-tested software out there. So now the only thing is like the intention of like, are you
65:33being intentional and doing it? And I think like that's when and I was talking about some of these
65:38things and Ryan was like, so you're basically doing like you're fixing the internet, right? Like
65:42it's it's all broken and people are okay with it. And then you're going out there and, and cleaning
65:48it up. And I think that's, that's, that's that's something I'm very happy to do. Uh, we find like,
65:54all kinds of interesting failures in our customers and they're like, oh, I didn't think of
65:58this. And it can be very serious. And and it's good that they can close them before like some
66:03customer finds them. Yeah. Shout out to Noemi for always making custom swag for everybody in the
66:10cohort. Yeah, it's pretty amazing. I think the, uh, I think the thing that you touched on there, which
66:15is, uh, which is really important, um, is this idea of kind of why now, which is, you know, these
66:22methods, uh, would have been prohibitively expensive and probably time consuming, too. Um, you
66:28know, uh, back, you know, just even a few years ago. And now it's actually kind of, um, malpractice
66:35or something to not use workers io uh, and to not have these systems in place because ultimately,
66:42um, the business case is there. Uh, the costs have been driven down, and and
66:49ultimately, you're focusing on the result, not the, uh, not sort of a dashboard, you know, uh, which I
66:56think which I think is really important. Don't give people more work. Yeah. It's called workers.
67:00You know, because you're doing the work for them. Right. One of the one of the one of the KPIs that
67:04I look at is like, what's the what's the number of support tickets or what's the number of incidents
67:10that you've had after adopting us? Like, does it dramatically go down then? We are doing a good job.
67:16Um, if not, then there's something that needs to change. Uh, so far that's been the case. And I would
67:22like to continue to, like, keep a very, very close eye to it. Um, because it's not just about, like,
67:28fixing failures, but it's like fixing important failures that cause outages, that cost engineering
67:33time. And it's very easy to measure it with some of these teams. So last two questions. These are
67:40the customary questions that we ask on the Solo Founders podcast. Well first of all, thank you for
67:45being here. It's awesome to reminisce about the first cohort of Solo Founders Program. Uh, it's
67:50been amazing to see you pull the thread. And like now, Workers IO is like the default way that
67:57people, you know do formal verification, ensure that their products, um, can live up to the promise
68:04to their customers. These are the most important products in terms of stability mattering, um, in
68:10finance and databases and in health care. Uh, this is really important work. The last two
68:16questions, the first one is the barricades. Uh, the reason that somebody should consider not being a
68:22solo founder. Um, and then we'll talk about that, and then we'll pivot over to the case for solo
68:27founding their case. Hmm. I think if I like even the bad
68:34case, if you think like a little long term, if the cost of intelligence becomes like the cost of
68:39energy, then I think you could essentially attempt anything. Even the feels that you do not
68:45understand. It can be incredibly lonely. I would say at times like, you don't have anybody to
68:51celebrate your successes with, Um, if anything goes wrong, you are the only one to
68:58blame. And if you do not, like, feel comfortable with those ideas, then I think
69:05it's probably not a good idea to do it because you'll just give up. Um, but I don't see any other
69:11reason. And it should be just default, like, okay, it's default solo. And if you feel a lot of some
69:16of these things, um, then if anything, even if you had a co-founder, you'll be the bad co-founder in
69:22the, in the biz. You're gonna bring everybody else down. Yeah, yeah. So I think I think, yeah,
69:29everything everybody else should just, like, start solo. And you can, you can always like, this is how
69:34I see it. Like, if, like my early team, I'm sure like over ten years, some of them will rise up to
69:41be like, important enough to be like, sort of the co-founders of the company, right? Like, nobody just
69:46starts a company and just is the only thing the co-founders never really, like, stay long enough in
69:51the company to be part of the like the we are the founders. A lot of the other like early team also
69:57does just as much hard work. Um, and I think like this gives us the flexibility to have them give
70:03them more meaningful equity. And then a lot of them will have enough, like skin in the game to
70:08like rise up to that level over time. So you kind of you kind of talked about the case for solo
70:14founding there as well. So maybe the case for the Solo Founders Program then in that case, like
70:19what what what would you say? Um, is sort of the reason that someone should consider the Solo
70:24Founders program, right? I know it's a plug. I know, I know it is to some extent. But at the same time,
70:30I know you give it to people straight, so you're not you're not going to say things that you don't
70:33actually believe. What what would you say are the reasons that maybe somebody should do it or how
70:37would you describe it to people? Yeah, I think it's the place if you are a solo founder, because the
70:43density of very talented, driven solo founders in one
70:50place, um, is really important. In the early days when you have no team, you have no idea. You have
70:56no customers, nothing. Um, I think it helps you, like, see the different journeys other people
71:03are taking and kind of like, uh, learn from their experiences, as I said. Right? Like, it's it's like
71:09having six, uh, six co-founders, uh, in your early journey where. Yeah.
71:15Like, it helped me a lot, right? Like, it helped me a lot. Um, in terms of, uh, like, whenever I
71:22was not being honest with some of the things that I was doing, uh, people would be very quick to call
71:27out, and that helped me. Course. Correct. Um, that's super cool. And also, it's just it's awesome to
71:34have you as, uh, as a neighbor and you live with, uh, two other, uh, SFP founders and a bunch of SAP
71:41founders live nearby. So, um, it's amazing to keep this going and, uh, and to have you as a part of
71:46this. And, uh, I'm just so glad we get to work together. So, Chai, thank you for being on the Solo
71:51Founders Podcast. Thank you for sharing so much of your experiences with folks. Um, this has been this
71:56has been wonderful. Thank you for having me. If you enjoyed this conversation, we would love it. If you
72:01can share solo founders with everyone you know, particularly people who are considering starting
72:06companies solo or who are already doing it, you know, the entire goal of what we do is to
72:10normalize solo founder and encourage more people to solo founded, instead of end up with
72:14co-founders of convenience, if you'd like, and we'd really appreciate it, you can go. Leave a review on
72:19Apple. You can go give us a thumbs up. You can subscribe on the YouTube channel. Anything helps.
72:23And of course, if you are starting a company or considering starting a company, you should go look
72:28at the Solo Founders Program. We work together with you in San Francisco alongside a bunch of
72:33other solo founders. The idea is it's much better to be solo together than to be solo alone. I'll
72:39spend a lot of time helping you, and we also invest $100,000 in your company. So if you're
72:44excited to learn more about that, go check out Solo Founders com Slash program.


