7:27
This wave is more about are you living at the edge of the technology and not as much about do you understand the right go-to-market right business model etc. This is about do you understand this technology 10x better than everyone else and I think nobody does that better than in India. Wow, what a day. The energy in this room has been incredible. The energy here around startups and technology and AI is just without parallel anywhere else in the world. Thank you all so much for coming out today and for being such an incredible audience.
So we wanted to wrap up today by sharing some of our takeaways from all of the talks that you just heard the incredible founders and all the advice that they gave and for giving you some very tangible advice that you can use to take away when you go home today and you start thinking about what you want to do in your journey to becoming a founder. To do this, we decided to invite a couple of friends a couple of dear friends of ours Puneet and Arnav both worked at Y Combinator at one point and so they were our colleagues. We got to know them really well. They've gone [snorts] on to have incredible careers of their own after YC but they've remained close to us and they have been two of the most instrumental people in
helping Y Combinator expand into India and are just some of the keenest observers in the Y in the Indian startup ecosystem. So thank you guys so much for being here with us. Thank you. Thank you for coming here. Look at the energy. Do you guys want to start by just introducing yourselves? Yeah, sure. So I've been coding since I was 10 years old. When do I do Bombay? My last startup was SuperDaily which was a grocery delivery company. We went to YC in winter 17. Jared was actually my partner. We scaled that company to around $100 million in annual revenue.
Uh Thank you. Uh exited to Swiggy, uh which is now a public company. Uh left in September '21. Uh wanted to partner with entrepreneurs on the other side of the table, so joined YC, and then came back in India and uh was a venture partner at Nexus. Uh but the itch to build never went away, so I'm back to building. Thank you. First of all, thank you for YC organizing. This is like a rock concert. It's quite amazing. Uh I'm Arnav. I would used to work at YC for many years, uh working with a lot of people there, and especially with uh Indian companies and developer companies. I'm with the venture firm now called Peak XV. Done a bunch of things
in my life. Had a startup that didn't fully work out. Uh worked at a company in the developer space, and now mostly trying to work with uh you know, all the young builders building in AI and investing in them. And it's uh just amazing to be here. Okay, so my first question is for Puneet. So, Puneet, you built SuperDaily, which is this incredible company, like part of the heyday of like grocery delivery exploding in India. Built it to a 100 million ARR. Incidentally, how many engineers did you have when you sold the company to Swiggy? This is a fun fact.
Yeah. Um so, we used to say that we have two engineers, but like I was one of them. So, we had just like really had me plus one. So, you had a 100 million ARR and one engineer. And this is before AI. Uh anyway, more recently, you've been an investor at one of the top VCs in India. Um and so, you have a like a really like front-row seat as like what's happening now in India. Um what is your advice on like what ideas folks in this room should be working on? Yeah. I think I'm very excited about um for the first time we can build global companies out of India and this follows what Mukund was saying from Emergent as well that um, now is the time I think over the last 10-15 years we have seen couple of
waves of startups in India and uh, we now know how to design well, how to do good products, how to do technically deep products. Uh, we and the shift with AI is not local. So, what happened with the last wave was that with mobile phones uh, labor got tokenized. So, anyone anywhere could say that I have 1 hour free, I can deliver for Swiggy, I can deliver for SuperDaily, I can deliver for Zepto and that happened because of the mobile revolution. Uh, the AI revolution is not local. Uh, that revolution led to hyper local network effects and so there were lots of companies in India and Latin America and US etc. I think the AI change is global
and uh, this is our time to shine. This is the time like Mukund said to look at the ceiling and uh, for the first time build global companies in India. Uh, and I'm very excited. I think some of the largest companies in the world will come out of India because we have the best technical talent and this is this wave is more about are you living at the edge of the technology and not as much about do you understand the right go-to-market, uh, right business model etc. This is about do you understand this technology 10x better than everyone else and I think nobody does that better than in India.
Okay, so as a follow-up to that I can imagine that there are some people in the audience who are like, "Cool Puneet, I got it. I should build for the global market." But they maybe don't understand the US market at all. Maybe they never even been to the US and that might sound very intimidating to them. Um, what would your advice be for them for like how they can actually do that. I think you are right that this was very hard to do before. Uh when you were for example building a SaaS company, you had never been to US, you didn't know anyone, you had to find a warm connect, or you had to maybe live in the SF
for like a few years to build like the network. But what has happened with the AI is that everyone simultaneously across the world just understands AI is important. And so uh we just saw that Giga and Emergent both, they never lived in SF before starting their companies. Uh and because as I said that this time it's more about are you building at the edge of technology uh than are people open to it. Everyone is open to it. Uh in the last YC batch there was a company from India which was which cold outreach to you insurance companies in the US and sold to them. This was unbelievable to me. I like it's hard to sell to insurance companies, but selling to insurance company in the US by a cold email when you are a Indian
student in IIT in third year is just unimaginable. But like what a time to be alive. And so I think it's just we have to get rid of our preconceived notions that we need warm connects. I think you can just reach out if you have a great product, people are open to it. Uh people want meritocratic solutions, solutions which drive outcomes. Uh this is not about where the budget goes, this is about whose solution is better and can drive better outcomes. And companies are just like fully open. And it's another leaky cup founders who we heard from today are maybe like a great example of that. They were very young founders with no connections in the US who like just made it work by just building a better product.
Like obviously apply to YC. And YC is the best way to go to US. And so uh like it has been the right way for Giga and Emergent both. And uh I think even for us when we went uh through Super Daily, just going to YC and breathing the air in SF just raises your ambition 10x. And so like YC is also like the great conduit to uh go global. So, Arnav, um you shared with us some really insightful thoughts earlier about the Indian educational system cuz you grew up here um and what it means for startup founders and what sort of personality traits people in this audience should think about incubating in themselves if they want to transition from being students that came out of the Indian educational system to being great
founders like the ones that we here saw here today and I was hoping that you could share those thoughts here, too. Yeah, totally. So, look a few thoughts. It is very likely that the future of the India AI ecosystem is going to be defined by all the people in this room and not by people who came from a prior generation. The traditional advice of most educational systems and the non-AI native people is irrelevant now. If you think about the traditional educational advice people get is, you know, go get a very high-paying, prestigious, safe job.
Go be a banker, consultant, an engineer, a doctor. And it is possible that a lot of those jobs won't exist 10 years from now or at least they will change dramatically and historically that safe path might actually now be the risky path. If you believe in the end goal of where all of this is going, it is also possible that the people that are most insulated from this are owners and entrepreneurs of businesses. Now, I will caveat that it's very it is very hard for an average young Indian to take the types of risks that someone who grew up in Silicon Valley can. Just given the social safety nets, I think the Indian startup ecosystem is booming but maybe not as deep as the US. And for a lot of people in this room,
right, who grew up in India, who came from humble backgrounds, if you get a super high-paying job that's safe, that's stable, that's very prestigious, you are already in the top 10% of this country and you've killed it and I hope many of you go and do that. But, you know, I'm willing to bet there are a lot of people in this room that disagree with that cookie-cutter advice and disagree with the advice that the traditional education systems go give you. And so, hopefully, I think with AI as people can build their own tools, people who are super high agency can now release stuff in the wild, can at least form their own convictions on what to do with their lives versus being told to do
what the Indian education system tells you or any other education system tells you. If you look at you know, a lot of the companies are getting born in the AI era, it is really benefiting young technical founders who live on the cutting edge. And it's likely that you won't learn a bunch of that in school today. And so, it does benefit the people who are super high agency, who want to totally go for it. And also, the reality is we're only year three or four in this whole AI boom. All the people in this room have easily 30, 40 years left to go build. You guys will be the ones are going to be the experts that should be telling the rest of us on like what AI tools we should be using. And for the people who at least have their own independent
point of view and high agency, I think this era favors those types of people who really want to go for it. And I guess if I could jump in and ask you there, how do people develop an independent point of view? Like, what does it take to actually develop that? Yeah, I can only speak from my personal experience and some of this is my own battle scars. You know, it's easy to say go develop be high agency, go develop your own point of view, but it's hard to do it unless you're not surrounding yourself by people who go do that. I think there's a reason why certain universities or even
YC, I mean, when I think about why YC is so magical, you know, I think founders come into the batch, they're already very, very good, but I feel like after the batch is, you know, PG used to say, they become the 007 versions of themselves, they become absolute beasts. And some of it is like just truly people network effects. And so, the fact that a lot of you even showed up here to come today on a weekend, I think speaks volume that you guys are interested in meeting other people like you. I you know, I think the other thing about kind of the Indian education system or maybe all education systems is it's like it's not cool to be ambitious or say that you want to be super ambitious. But
if you come to a place like YC, it is expected that you're going to be ambitious. And so in many ways like surrounding yourself with people like that just allows you to at least form your own opinion because otherwise you constantly get the cookie cutter advice from people who have done it in the past. And in AI actually it's very dangerous to follow the advice of people who are not AI native because they're just not in the game with you. And so I think in this era just surrounding yourself with people who are at the cutting edge I think is a very conscious deliberate choice you can make in your life which I think will just compound as you grow through your career. That's amazing advice. Thank you.
That might segue nicely into a topic I was hoping that you could talk about John which is the phenomenon of YC founders getting younger and younger. There's a lot of young people in this audience. I met some people here who are like 15 years old. Um what can you tell us about like what's going on there? Yeah, I think first of all it's certainly true that if you look at the last sort of 10 20 batches the average age of the founders have gotten younger. Um it's not that we intentionally try to fund older or younger founders. I think what we're seeing couple things that some of the kind of the stories that you've heard it is now far more true that AI has completely leveled the
playing field for young founders, right? And you know, it was always true that you didn't have to be an expert and some of the folks that talked today you didn't have to be an expert to really get deep into the field. But AI has completely changed that to the next level which is you are no longer limited by your ability to build, you're limited by the pace with which you can learn. And for young founders that's a huge advantage because you can build very quickly and you can get insights very quickly. And that's what we're seeing the best young founders are doing. The other thing is the best young founders that we're seeing are tinkering more, right? And you know, to the point earlier, what do I mean by
tinkering? They're following their curiosities. And they're working on the edge of what might be possible today. And you know, some those of you that kind of walked in through the front, you saw this model that's live in the future and find what's missing, right? And this is actually how to live in the future and find what's missing, which is to tinker and to follow your curiosity and to work on the things that are barely good enough for the models to do. And in so doing you're going to beat benchmarks, you're going to establish new heights, you're going to figure out where the bottlenecks are. And those bottlenecks are some really good ideas. And this is not just me saying, you know, what you
should do, but this is actually what we're seeing the young founders coming into YC are doing. I think like one very empirical thing that I suspect is part of what's driving it too, is that if you kind of think back to all the stories we heard today, I think almost every founder had the experience of their big idea was not their first idea. There's some exceptions, but for almost all of them, they went through some number of pivots. And I think it's often because the a good startup idea is not obvious, right? It's like you have to kind of be in the field building something to notice the really good idea. Like I work on a whole bunch of side projects now, and like every time I work on a side project, I
discover like five good startup ideas. And I have to decide what I want to do with those or not. Um I'm sure that's how you started to get back into building too. You started tinkering and you were like, "Oh my god, like I can just like make one of these into a company." And I think what hap what's happening now is that with coding agents, you can get something off the ground so fast that like as a college student just kind of messing around on nights and weekends or in between classes or skipping classes or whatever, you can just very quickly find extremely good ideas that would not have been obvious if you were just on a whiteboard thinking, but are very obvious once you start building something and you realize what's missing.
Ankit, I want to riff on that point. Um it's certainly something that struck me from the talks today was how many of the ideas were not the first ideas. The founder went on like a pivot journey. And another thing that struck me was how many of the ideas that worked were not the first mover in their space. They were like this, you know, the third, fourth, or like 42nd like version of the idea to launch. Do you want to talk about second mover advantage and maybe like more concretely even like what that would mean for like people as they think about what startup ideas to work on?
I mean, I think this is one of these crazy things we see over and over right now. I mean, I think back to the example that Varun from Giga shared of when he closed DoorDash, he was competing against several companies that had several hundred employees and he had eight. Uh and he won the contract. And I think it speaks to what a really strong technical team that can build a superior product can pull off. Um and I think what's happened with coding agents is that it means that people that have great product clarity and vision can make their ideas become real extremely fast. Um and often with even more, you know, precision and quality than you would expect. Like I think there's this um
sort of dated worldview of like AI slop or something like that. Like people love talking about like, "Oh, yeah, if I use coding agents, it's going to produce slop or whatever." It's like, "No, that means you're actually not using them well." Like the people who use them well produce incredibly sophisticated code um at a pace that was much faster than before. And I think this causes the second mover effect cuz you can kind of like I would I'm not exactly saying this is my advice, but like a kind of reasonable way to compete right now is like find something that's kind of working and then do it better than them and then beat them. And unless that first thing had incredible network effects, which few things do, you
actually can just beat them from a better product. And we see this with Emergent also as a great example of a company that's um pulling this off and now scaling really fast. Um speaking of AI coding, could you maybe expand people's Overton windows of like just how much you can like crank up AI coding if you're like really go like all the way in on it? Yeah, this has been um my last 4-month rabbit hole basically. And if any of you guys follow Gary on Twitter, it is very much his last 4-month rabbit hole. I think that I was also one of these people that didn't fully realize how good things had
gotten until I'll be totally honest until like December of this last year. It was until this Christmas break when um my Twitter wouldn't shut up about how good Opus 4.5 had gotten. And I decided, "Okay, it was time for me to actually try to build something end-to-end with these models." And I think what I learned in doing that and I finally paid for the $200 a month max plan. And this is the Anthropic max plan um or the Codex one, kind of similar. And what I didn't realize was that unless you are paying for at least that level of usage, you actually are not anywhere close to the frontier right now. Like there's a meaningful unlock you get from just letting the tokens rip. Um and there's
even more extreme versions of this. Um Gary Tan, for example, spends thousands of dollars per day on token usage. Uh and [clears throat] obviously, I'm not claiming that everyone can spend thousands of dollars per day. That's an incredible position of privilege he has and that we all get to have. But as a result, we just get to peer into the future, right? Like I think it's very safe to assume cost of compute per um level of intelligence is going to go down a lot. And I think what you see when you're willing to spend tons of effort undoing computing well. For example, not just writing the 20 unit tests that a human would write, but write 10,000 unit tests because why not?
Um not just writing you know, a small number of docs, but writing tons of them. Um not just having a few uh corner cases covered, but all of them. Um you can actually get incredibly effective and powerful code really fast. Um Incidentally, you also run into good startup ideas all the time. Like uh I'm currently working on this uh email client side project you guys may have seen on my Twitter and in building that, which was a project I did basically like quasi just for fun and because I found my existing email client annoying, I discovered several new ideas in building it just from basically pushing these models to the limits of what they could do and realizing that you can get
a much better output if you basically just don't worry about the uh the cost of inference, right? It's like uh Gmail's auto reply thing sucks, but like if you're willing to spend like $5 per email, it's actually really good. And it really shows you like how far you can push these things if you're not kind of bandwidth constrained by intelligence. Um so I suspect what we're going to see I maybe just like the lesson for founders in [snorts] this room, I think a big part of why we wanted to hook you guys all up with credits is not just cuz like oh yeah, it would be cool, but it's like we actually want you to feel a taste of what it's like to not be capital constrained, to like actually just like let it rip. Um
and in fact, maybe I'll go further is like whoever in this room manages to like get a product off the ground that uses those credits the fastest, like one of the larger credit packages there, should email me. Uh and I would like love to hear what people built with that cuz I think um it's like it's quite incredible what you can do if you just let it rip. Totally. Let it rip is good advice when it comes to spending tokens. Let's talk about a topic. about people who cannot afford, what can they do? How can we help them?
Uh let the tokens rip. Well, um John raised the topic of open source models. Um do you guys have some thoughts on open source models and what that can mean for bringing down the cost of what Gary Tan is able to do to make it accessible to people who aren't Gary Tan? I mean, Arnav and I were just talking about this uh earlier. Arnav, I'm curious what you're seeing also, but like I don't know, like I've been playing around with the minimax models and a bunch of others and it's like they're pretty good. They're really cheap and they're pretty good and um I think I'm I'm of two minds about this where on the one hand it does seem like being at the very frontier seems
extremely useful for a subset of tasks, especially coding. But for a lot of tasks, um not necessarily. Like I was quite inspired by um Vidit from Meesho saying that he wants to use voice AI to bring the next billion people online with shopping. And I think to do that it sort of requires the models to work at a price point that the next billion people can access and I imagine he's going to be building on open source models to do that. Yeah, I would also highly recommend there's this YC company called Open Code which is built on top of the open source models and they're very good. Um but at the same time, look, I think there are so many companies that are thinking about this in the right way
where they want to give unlimited budgets to their employees. And so for whatever reason, if you're budget constrained, go try and work for the companies that want you to token max and just be at the frontier to see what the power is. I you know, I think we should safely assume that the cost of tokens will keep coming down. Open source models will keep getting better and better. And so being early in the game as fast as possible as you can early in your career will just give you such a huge advantage on how the tools work. And you know, if you just back to the earlier point, if you try and build for the model 6 months ahead, a year ahead, or just think through what the models can be capable of
6 months or a year ahead, again, you'll just be far ahead than some of your competition in terms of the thinking and the road map and the types of things that you guys can capitalize on. Let's talk about a topic that I imagine is on many people's minds. Um I'll bet many people in the audience are thinking about applying to Y Combinator at some point, and they might be wondering what does Y Combinator look for? How should they think about like having a great Y Combinator application? John, do you guys want to maybe clear up some misconceptions and talk about like the real talk about what we really look for?
I think there's a common misconception that you want to write things in your application that is like the most impressive or the things that's the most complicated. Um clarity is most important. If we can't understand what people are actually building, um that's not going to work, right? And so clarity above all else. Um we talk a lot about taste and agency in founders, right? Um first of all, I think everybody sort of understands we've talked about this a lot, which is we're not investing in ideas. And as many of you have heard today, the first ideas might not be the ones. And they might not work. Um the first ideas well after the batch that have gotten to millions of ARR could also be subject to pivots. You might be pivot pivoting
several different times. So it's not about the idea. Fundamentally, it's about you, the founder. And we talk a lot about taste and agency, and I'll I'll offer kind of what that means to me. Taste, I think the common misconception is that you build products that look, you know, good. You can exercise kind of your taste. But it's fundamentally more about intention. Are you building products and showing demos that are intentional? Are there insights from the customers that you've talked to that back your design choices? That, you know, that and the pace with which you can actually get to those insights and develop the product that encapsulates that intention. That's taste. Um agency,
Kunal mentioned earlier, I think you will become a more formidable version of yourself and I think there's a there's an um an old essay that is very like many PGAs that I would recommend everyone in this room read. It's titled Relentlessly Resourceful, right? And you know, it's the notion that do you let the conditions of the world happen to you or do you exact what you want to do on the world? And do you will things to happen? And so, those are some really strong founder qualities. And of course, in the end, your rate of learning matters a lot. And I think the reason why we care about those things is not just from some like philosophical belief that they're right, but it's because if you kind of just
reflect on the six founder stories we heard today, they're like perfect examples of why that matters, right? Like essentially all of those companies had to do all of those founders had to lead their companies through completely insane times to get to where they are. And being relentlessly resourceful and having taste as expressed through customer obsession is kind of this is consistent theme through all of them. It's actually really interesting to listen to them and say like wow, like we didn't prompt all these people to talk about their customer so much, but like they all like very intuitively are thinking about their customer in a very
centric way. And I think that like speaks to how important that is. Like to make it, you have to be so obsessed with that. And I guess a related thing I think about is like what people should do before they're ready to start signing companies. Like we mentioned tinkering earlier. And maybe I'll take that a little bit further and say like people should work on projects. And I'll give a very specific definition of project. Um a project is when two people build something that was not assigned to them and get someone to use it. And what's fascinating about that definition of project is you can have an entire computer science education and job and be successful and never do a project.
Either you did it solo or it was assigned to you from an internship or a class or whatever. Or you never got anyone to use it or you never built it. It was just like a whiteboard exercise. And it turns out if you just do projects, especially when you're young and in college, um and if you use, you know, the tools that we're giving you through that um starter pack. Uh you're just like guaranteed to find startup ideas. And it's it's projects kind of lead people to have those exact traits that John was discussing. One thing I'll say though is Jerry, you've been doing this for a while now. You joined YC 10 years ago and have been working in your own startup in this for 20 years.
Are there things What's Has anything changed? Or like what has changed? Weirdly, surprisingly little has changed. Um and I think it's because these like fundamental truths about like what makes for great founders go way back. I mean, you could go back to like Thomas Edison. And like if we could interview Thomas Edison, I'll bet he would sound just like the founders from today. I'll bet he'd be obsessed with customers, obsessed with tinkering, obsessed with building at the edge of what's possible with technology. And so like if it hasn't changed since Thomas Edison, it's not very likely to change in 20 years. Um what I mean what has changed is that AI has given the best builders so much more leverage and it has democratized
the ability to build great stuff to everyone, which is how we're able to fund people like the founders of Giga or audit the founder of Zepto at like age 18 and like they can go on and build epic companies really, really quickly. I think the speed part seems like the most tangible immediate thing too where it's like, "Oh, you can just build you can just ship that like tomorrow or like next week, not like a year from now." More like tonight, not tomorrow. Yeah, more like tonight, yeah. I wanted to wrap up with a few announcements. Um first, all six companies that you heard from today are all very actively hiring engineers. One of the reasons that they came here today
is because they are very interested in hiring the people in this audience. So, if anyone is interested in working for them, you should go apply on their website for a job um or email the founders. Um I will say that if you want to become a really exceptional founder, one of the best ways to get there is to go work at a really exceptional company like the ones that presented today where you can learn from the best in uh business. Um second announcement is um there's an incredible roomful of people here and it would be a shame if everybody went home and didn't keep in touch with each other. Like Ankit was saying, projects are really best done with two or more people. And so, I would
really encourage you all after we finish here to hang out. We're going to have food and photos on the lawn. We're going to hand out um swag bags um to like just like hang out, meet as many people as possible, exchange numbers, start WhatsApp groups, keep in touch, meet up after this, and like start working on projects with people who you met here. You might even meet your co-founder right here in this crowd here today. Okay, and we saved the most important announcement for last. Ankit, do you want to wrap us up by talking about the credits and what people should do with it?
This story two things I want to say. Well, one is that um my introduction to startups was when in 2013 as a freshman in college, I went to Startup School in San Francisco. It was basically the exact same event as this. Um and we had some amazing speakers there and I realized that the well-trodden path I was on was not necessarily the one I wanted to stay on. And I really hope that some of what you all um heard [snorts] today inspires you all to consider that as well. And I'm really excited to be able to give you guys all the resources to help starting to do that. Um and so what I would really love to see happen um over the next few days is to see teams form from here or from your best friends back home, wherever you're
coming from, and for you guys to use those credits to start building amazing things. Um you'll all get an email from us um tonight or tomorrow giving you access to um all of those credits. You'll be able to start redeeming them right away and get them within a few days. And we're really excited to see all of the incredible stuff that you guys built. That's it from us, folks. Thank you all so much for being here with us today. We can't wait to see what you build. Thank you.
