The Exchanges

Q Tom Blomfield at Y Combinator, you have this incredible perspective. You see so many thousands of companies that apply for YC every year. How do you think about the wrappers versus non-wrappers in AI, and then also bluntly where the excitement is, whether it's in application layer versus infrastructure layer?

A There are clearly some wrappers. You know, if you can build it in a weekend at a hackathon and make a bunch of money, probably not defensible. For most Businesses building on top of these models. You know, you can describe the last generation of startups as like MySQL wrappers or AWS wrappers or something like, you know, it's the same kind of logic applies. I think where the sustaining value lies is identifying an industry, deeply understanding the regulation in that industry, the tooling, the language, the, all of the training, how people sort of work and behave and act and tailor your software To fit into that industry in a way that's extremely deeply embedded. Most people building application layer stuff in an AI say it's 80 to 90% traditional software with 10% AI. It's working in construction, in construction, fingering out how Procore works or, you know, how the Salesforce CRM works or some Oracle database. And like, I don't think OpenAI is going to come and steamroll the construction company, AI companies, because they're not going to deeply integrate into the processes and software

AI assessment note: “Most people building application layer stuff in an AI say it's 80 to 90%”

Q What specifically was it? It enabled you to see something of the world that you didn't know?

A We were play acting before then. We were three guys in London who'd come out of consulting, and we were play acting at being startup founders. You know, we'd hired a bunch of interns. We were just running around doing dumb stuff. We had no role models. We had no one who was smart or successful around us to model ourselves afterwards, after. And, um, being in YC just put us amongst this Group of, like, high ambition, high achieving technical founders who thought they could accomplish something. And then every week, a new founder would come in. Max Levchin would come and tell us about the early days of Facebook, the early days of PayPal. And then I remember, um, the next week, this sort of timid, short guy walked up and said, hi, I'm, you know, I'm Mark Zuckerberg to give a talk on the early days of Facebook. And it was kind of incredible being surrounded by those People. Whereas in London we just didn't have those role models, and I think we would have, the company would have like slowly, slowly sort of died in London. We couldn't raise a single penny of investment in 2011. So YC really gave us that break and, and raised the bar for us.

AI assessment note: “being in YC just put us amongst this Group of, like, high ambition”

Q And they're not allowed to start before then?

A We strongly encourage them not to, and we'll be very, very disappointed if they do, because it's not in their interest, because running a competitive process with lots of people bidding is going to get them better terms. And what happens is nervous founders will fundraise early because they're like, well, what if demo days aren't going well, so I'll just talk to some early. And then what happens, either the VC looks at you without much progress and writes you off because you've not made enough progress, or they like you so much, they give you a preemptive offer. But the valuation is not as good as you would get by demo day, because that's why they're doing this, right? So they give you an offer of two million on, pick a valuation, I don't know, X, right?

AI assessment note: “We strongly encourage them not to, and we'll be very, very disappointed if they do”

Q Do you find the advice that you're giving these startups is different to the advice that you gave in a completely pre-AI world?

A Big companies are now more willing to spend money on this stuff than they ever were. So the previous advice of, like, go and sell to other startups, I think is actually not good for very, and maybe a dev tool startup, go and sell to other startups, but in general, I push companies, B to B SaaS companies, To start mid-market and go up as quickly as they can. Every manager of every big company in the world is being asked by their boss, what is AI going to do to our company? And many of them do not have a good answer, and they're looking around for startups to help them. And for sure, it's innovation budgets and proof of concepts and all this stuff. Um, but where it works, they convert to real contracts. I've seen companies go from zero to like half a million or a million of revenue within the YC batch. With really, really big companies, like Fortune, 500 companies, and turn them into recurring contracts. This is real. This is saving people so much time. It's astonishing.

AI assessment note: “the previous advice of, like, go and sell to other startups, I think is actually not good”

Q Did you fall in and out of love with what you did over that time? I think people always say, like, ah, founders, you're always so in love with what you do, and I think that's a bit of a lie.

A I loved the product all the way through, but by the end I hated the company because it, um, got so big and so over-regulated in my opinion, and It wasn't as simple as identifying something customers wanted and building an amazing product that they would pay for. That was not enough. There was this other whole, like, crazy language that we didn't speak called financial regulation. You had to keep these regulators happy as well. And I was not good at that, really, honestly. And I think TS, the new CEO, and Sajata, the COO, are spectacularly good at that, as well as maintaining the culture and the products. And they've, they've just done such a good job taking it over the, you know, when I left it over the next four years that I couldn't have done.

AI assessment note: “I loved the product all the way through, but by the end I hated the company”

Q So this is my trouble. I see 10% the whole time, and I'm just like, how on earth do you expect a great ambassador to really, really be a partner with you when they're going to get six? Because you also want to have these angels in. It's not going to happen.

A Yeah, so I think the Reality, like the, the more nuanced truth is not YC is telling founders to only raise 10 or 12%. They're, we are telling founders that over diluting at seed is not good for them, and that retaining control would be good for them, and they can make progress towards their series A milestones with probably a million and a half or two million, and the valuations they might get at demo day are around 15 to twenty million, which roughly ends up at 10% dilution. But if a great partner comes in, if a top tier fund or someone they, Has a specialism they really, really like, and they'd want 15% ownership or even 20% ownership. Great, absolutely consider it. I wouldn't write anything, I wouldn't rule anything out. And I think this is where it gets misunderstood that YC's rule is only 10%. It's absolutely not true. We've had Sequoia and Andres and Accel and Founders Fund and Google Ventures and all of these funds leading rounds in companies in the last batch. And it's just about having a reasonable conversation with the founders to say, where's the flexibility and where can we meet in the middle?

AI assessment note: “I think this is where it gets misunderstood that YC's rule is only 10%.”

Q What did you learn by being a visiting partner?

A I learned so much different stuff from different people. So, um, I did a batch with, my first batch was with Dalton, and the way he can, um, help young founders find new ideas, really deeply technical young founders without a clue what they want to work on, Dalton can come in and sort of Unpick their background and history and find some thread to pull on that they're super excited about and turns into a great startup idea, and it feels like it's their idea, but he somehow, like, conjures that up out of their background. Michael Seibel I did, um, a batch with just before I got partnership, and he has this incredible way of giving the most brutally hard feedback you can imagine. The founder absolutely needs to hear, and then finishing off with his This, like, heartwarming chortle, this, like, laugh that, that kind of reassures you. Yeah, makes you realize he loves you deeply. He might think you're a total pile of shit, but he really cares about you, wants you to succeed. So being able to, and that's something I'm trying to learn, I think I can do the harsh feedback quite easily, that comes naturally, but the sort of, the warmth and the empathy alongside that, I think I, I really want to learn from Michael.

AI assessment note: “I learned so much different stuff from different people.”

Q a brilliant video of Putin and Xi who kind of both like cheers with the glass and then put it down. And you're like, ah, ok. Um, ok, so we have the 25. I'm sorry for us, but I actually, I listen to a lot of startup podcasts. I don't hear this. Like, what does the time look like then for you with the 25? One hourly meeting per week?

A So, um, By default, um, every week there is a, so YC spit into four groups, so it's not a whole YC, it's four separate mini YCs, it's sharded, and, uh, you'll get a very different experience in each one of them. The partners are specific, and then we have our own speakers, so, um, one night a week will be your group event, and you'll get someone like Brian Chesky, or Paul Graham, or we had Kevin's sister on the Instagram founder come and talk to our founders, um, So you've got your Tuesday dinner, um, you've got pre-scheduled office hours, sort of one-on-one or one company with one partner every two weeks, and then group office hours every two, the alternating two weeks, which is a really great mechanism, I think was introduced in 2011, where you have seven to nine companies who all come for a sort of one and a half to two hour session, and you might bring, and they're organized thematically, so we had a fintech section and a, uh, biotech or whatever it might be, Ideally all at roughly the same stage of, of company life as well. And then we, basically the founders are sort of problem solving with each other. You know, how do you find a banking partner in fintech? Or how do you deal with government regulation? Or how do you find your first, how do you sign your first enterprise deal as a B to B SaaS company? So these group office has a really good way to share knowledge between …

AI assessment note: “one night a week will be your group event, and you'll get someone like”

Q If you could change anything about the YC process in that three to four months, what would you change?

A The thing I am trying to change, um, I ran an experiment last batch where I randomly matched groups of eight to, um, six to eight founders with, across groups, uh, for dinner, and just sent them out for dinner. Um, it turned out to be Unbelievably difficult to get a group of eight founders to organize anything for themselves. It was astonishing. And so this time round, we are doing way more of the curation where we're like, we'll literally tell you, turn up at six PM at this restaurant and you will meet seven random YC founders and you can chat about your startups. So the thing that I found most valuable as a founder in London was meeting groups of peers that I would then share experiences with. So we're trying to curate those experiences for people so they get more of a community. If it works for the batch, We'll start doing it for alumni as well. So you can say I'm in London and we're going to do 20 YC dinners next week of all the YC alums so they get to meet each other and just make that network of YC even more powerful.

AI assessment note: “The thing I am trying to change”

Q just going, I get it, but the trouble is you don't have distribution, and what I worry about is distribution trumps product in this world, which is where Microsoft Office just bake it in, And because of Microsoft's distributing capabilities into all, uh, was it oncologists or dentists or whatever we want to use as that famous dentist example. Fucking dentists. Poor dentists. But do you see what I mean?

A I do, but I don't agree because I think Microsoft Office, like, by definition has to be a general product. It will be the best word processor for sure, but it's not going to be the tool that the dentist is actually using to write their clinical notes after a meeting. Um, it's gonna be, that is a specialist piece of dentistry software, or the three D modeling software that some, um, architect is using to, to design a skyscraper. Microsoft Office is, it will benefit for sure Google Docs and Microsoft Office as a general document writing thing, but, um, one of our companies was called, uh, Solve Intelligence, which is a tool for patent writing. So taking scientific inventions and papers and turning into really, really high quality Patent submissions are very likely to be accepted. I don't think you can tell me Microsoft Office is going to help write those patents.

AI assessment note: “I do, but I don't agree because I think Microsoft Office”

Q Final, final one, I promise. Do you ever challenge in terms of loss of identity? Like, I don't go on holiday because you're detached from work, and you mentioned me starting since I've never ever done anything else but this. So it's really uncomfortable actually to have 10 days or seven days with nothing.

A Yeah, this was a big problem for me when I, when I was thinking about leaving, the sort of year or so before I actually left. I couldn't imagine myself and the company not being the same thing. And who I would be, or what I would be, or whether I would be relevant anymore, or I was worried the company wouldn't survive without me, which was a very arrogant and untrue thing to think. It thrived without me. Um, and I had to rebuild my, you know, my sense of identity, my ego. And honestly, going to America where no one knew who I was, and joining YC as basically an intern, was incredibly humbling and really positive, because I had to, like, rebuild my entire self-image. At Monzo, anytime I said anything, A whole team would, like, spring into action, and even if I didn't intend them to do it, I'd be very careful what I said, because literally I'd come back two weeks later and find 30 people had gone off in crazy tangent. And I joined YC, and I say things, and people, like, ignored me. And not in a nasty way, but I was just, like, just another person, right? I was a visiting partner, I was a teaching assistant. My ideas didn't matter that much, and that was so humbling. Um, it was great. Honestly, really, really great.

AI assessment note: “I couldn't imagine myself and the company not being the same thing.”

Q Ok, so we move then to, like, the application layer that sits on top of foundation models, which is where I presume you're excited. Yeah. Why, why is that? Because most, uh, it's a wrapper, it's non-sustainable. Why are you excited?

A There are clearly some wrappers. You know, if you can build it in a weekend at a hackathon and make a bunch of money, probably not defensible. For most businesses building on top of these models, I see it, you know, you can describe the last generation of startups as like, MySQL wrappers or AWS wrappers or something like, you know, it's the same kind of logic applies. I think where the sustaining value lies is identifying an industry, deeply understanding the regulation in that industry, the tooling, the language, the, um, all of the training, how people sort of work and behave and act, and tailor your software to fit into that industry in a way that's Extremely deeply embedded. Most people working in AI now, most people building application layer stuff in an AI, say it's 80 to 90% traditional software with 10% AI. It's working in construction, in construction, figuring out how Procore works, or, um, you know, how the Salesforce CRM works, or some Oracle database. And like, I don't think OpenAI is going to come and steamroll, like, the construction company, AI companies, because they're not going to deeply integrate into the processes and software that exists in each of those industries. So I really believe everyone who works with a computer will have an AI co-pilot assistant thing in the next two or three years, whether you're like an oncologist or a law professor.

AI assessment note: “I think where the sustaining value lies is identifying an industry, deeply understanding the regulation”

Q Did you fall in and out of love with what you did over that time? I think people always say, like, ah, founders, you're always so in love with what you do, and I think that's a bit of a lie.

A I loved the product all the way through, but by the end I hated the company because it, um, got so big and so over-regulated in my opinion, and It wasn't as simple as identifying something customers wanted and building an amazing product that they would pay for. That was not enough. There was this other whole, like, crazy language that we didn't speak called financial regulation. You had to keep these regulators happy as well. And I was not good at that, really, honestly. And I think TS, the new CEO, and Sajata, the COO, are spectacularly good at that, as well as maintaining the culture and the products. And they've, they've just done such a good job taking it over the, you know, when I left it over the next four years that I couldn't have done.

AI assessment note: “I loved the product all the way through, but by the end I hated the company”

Q So this is my trouble. I see 10% the whole time, and I'm just like, how on earth do you expect a great ambassador to really, really be a partner with you when they're going to get six? Because you also want to have these angels in. It's not going to happen.

A Yeah, so I think the Reality, like the, the more nuanced truth is not YC is telling founders to only raise 10 or 12%. They're, we are telling founders that over diluting at seed is not good for them, and that retaining control would be good for them, and they can make progress towards their series A milestones with probably a million and a half or two million, and the valuations they might get at demo day are around 15 to twenty million, which roughly ends up at 10% dilution. But if a great partner comes in, if a top tier fund or someone they, Has a specialism they really, really like, and they'd want 15% ownership or even 20% ownership. Great, absolutely consider it. I wouldn't write anything, I wouldn't rule anything out. And I think this is where it gets misunderstood that YC's rule is only 10%. It's absolutely not true. We've had Sequoia and Andres and Accel and Founders Fund and Google Ventures and all of these funds leading rounds in companies in the last batch. And it's just about having a reasonable conversation with the founders to say, where's the flexibility and where can we meet in the middle?

AI assessment note: “if a great partner comes in... and they'd want 15% ownership... absolutely consider it”

Q When you see him as CEO of YC and what he's done, the excitement that he brings, especially around Demo Day, when you see him taking photos with the old school camera and just the joy that he, I didn't mean it horribly, has brought back it then to the YC community. What do you think makes him so good as a CEO of YC?

A I think YC has changed a lot in the last two years for the better, and I put it, a lot of it down to Gary. Um, He was there in the early days. Um, when I was there in 20 11, he was a, I think, a part-time partner. He helped us design our pitch deck for GoCardus in 20 11, and he was taking the photographs back then. Um, so I think he, like, experienced YC at its purest, at its core. And then really, honestly, under Sam Altman, I think it expanded in a bunch of different weird areas, and, um, it just lacked strong, and Sam was distracted. He was doing OpenAI from very, very early on. Um, and so to have a CEO now who is, whose full focus is the batch. It's a core YC. It's not some growth program. It's not something international program. It's not, you know, creating AGI. It's the core YC batch that Gary has experienced and loves and cares about. I think is, has made it great.

AI assessment note: “to have a CEO now who is, whose full focus is the batch”

Q What specifically was it? It enabled you to see something of the world that you didn't know?

A We were play acting before then. We were three guys in London who'd come out of consulting, and we were play acting at being startup founders. You know, we'd hired a bunch of interns. We were just running around doing dumb stuff. We had no role models. We had no one who was smart or successful around us to model ourselves afterwards, after. And, um, being in YC just put us amongst this Group of, like, high ambition, high achieving technical founders who thought they could accomplish something. And then every week, a new founder would come in. Max Levchin would come and tell us about the early days of Facebook, the early days of PayPal. And then I remember, um, the next week, this sort of timid, short guy walked up and said, hi, I'm, you know, I'm Mark Zuckerberg to give a talk on the early days of Facebook. And it was kind of incredible being surrounded by those People. Whereas in London we just didn't have those role models, and I think we would have, the company would have like slowly, slowly sort of died in London. We couldn't raise a single penny of investment in 2011. So YC really gave us that break and, and raised the bar for us.

AI assessment note: “being in YC just put us amongst this Group of, like, high ambition”

Q Um, okay, so we have that. Then what?

A So, then I spent all my money. This money I was supposed to deploy over three or four years, I, uh, went through in nine or 10 months, and I was finding a lot of work, you know, signing that many documents, wiring, simply making the wire transfers for that number of investments is Very, I spent a lot of time on the phone with my, with my banker and it was just painful and it was lonely. So I, I spoke with Matt Robinson a lot. I spoke with Carlos. He's gone to index who are very, very good angels. I was trying to learn my, learn the craft, but I found it a lonely process. And so YC serendipitously like came along at just the right time and said, would you like to join us? You can invest our money. We've got a finance and ops and legal team that'll do all the admin for you. And you'll have a team of partners around you that you can learn from. Which is all of the things I wanted. Um, and they said the only catch is you have to move to Silicon Valley.

AI assessment note: “So, then I spent all my money. This money I was supposed to deploy”

Q What did you learn by being a visiting partner?

A I learned so much different stuff from different people. So, um, I did a batch with, my first batch was with Dalton, and the way he can, um, help young founders find new ideas, really deeply technical young founders without a clue what they want to work on, Dalton can come in and sort of Unpick their background and history and find some thread to pull on that they're super excited about and turns into a great startup idea, and it feels like it's their idea, but he somehow, like, conjures that up out of their background. Michael Seibel I did, um, a batch with just before I got partnership, and he has this incredible way of giving the most brutally hard feedback you can imagine. The founder absolutely needs to hear, and then finishing off with his This, like, heartwarming chortle, this, like, laugh that, that kind of reassures you. Yeah, makes you realize he loves you deeply. He might think you're a total pile of shit, but he really cares about you, wants you to succeed. So being able to, and that's something I'm trying to learn, I think I can do the harsh feedback quite easily, that comes naturally, but the sort of, the warmth and the empathy alongside that, I think I, I really want to learn from Michael.

AI assessment note: “I learned so much different stuff from different people. So, um, I did a batch”

Q Um, My question to you is, when you speak to the other partners, especially I guess going through COVID where it was completely remote, do they have any lessons, observations from the complete all remote YC to the all in person?

A Everyone's a lot happier now with the all in person. There are elements that we have retained of not very many of the remote batches, so we had a few that were fully remote. Um, doing demo day with remote presentations, like record or live but On Zoom presentations with an investor reception in the evening seems like a really good balance of the two before, uh, of the, you know, all on Zoom versus all in person. Sitting through two days of presentations in a stuffy lecture theater where the AC doesn't work well is not a pleasant experience. So that's one thing. Uh, remote interviews. We used to fly everyone to San Francisco for interviews, which is, seems insane now. Um, I don't know when that changed actually, whether that was pre or post COVID. Um, But overall, the in-person experience is dramatically better. Having in-person office hours, group office hours, even cooking for people, like the, the act of cooking dinner for a group of people you've invested in, and then serving them with your own hands. It's just like a, an act of care, in a way, that, that creates a, just an emotional bond. Totally. I trust you not to give me food, boys. That when it's just, When it's just a 20 minute zoom every week, it's so transactional and empty. I really feel like I know these founders now after four or five months.

AI assessment note: “Everyone's a lot happier now with the all in person.”

Q a brilliant video of Putin and Xi who kind of both like cheers with the glass and then put it down. And you're like, ah, ok. Um, ok, so we have the 25. I'm sorry for us, but I actually, I listen to a lot of startup podcasts. I don't hear this. Like, what does the time look like then for you with the 25? One hourly meeting per week?

A So, um, By default, um, every week there is a, so YC spit into four groups, so it's not a whole YC, it's four separate mini YCs, it's sharded, and, uh, you'll get a very different experience in each one of them. The partners are specific, and then we have our own speakers, so, um, one night a week will be your group event, and you'll get someone like Brian Chesky, or Paul Graham, or we had Kevin's sister on the Instagram founder come and talk to our founders, um, So you've got your Tuesday dinner, um, you've got pre-scheduled office hours, sort of one-on-one or one company with one partner every two weeks, and then group office hours every two, the alternating two weeks, which is a really great mechanism, I think was introduced in 2011, where you have seven to nine companies who all come for a sort of one and a half to two hour session, and you might bring, and they're organized thematically, so we had a fintech section and a, uh, biotech or whatever it might be, Ideally all at roughly the same stage of, of company life as well. And then we, basically the founders are sort of problem solving with each other. You know, how do you find a banking partner in fintech? Or how do you deal with government regulation? Or how do you find your first, how do you sign your first enterprise deal as a B to B SaaS company? So these group office has a really good way to share knowledge between …

AI assessment note: “one night a week will be your group event, and you've got pre-scheduled office hours”

Q If you could change anything about the YC process in that three to four months, what would you change?

A The thing I am trying to change, um, I ran an experiment last batch where I randomly matched groups of eight to, um, six to eight founders with, across groups, uh, for dinner, and just sent them out for dinner. Um, it turned out to be Unbelievably difficult to get a group of eight founders to organize anything for themselves. It was astonishing. And so this time round, we are doing way more of the curation where we're like, we'll literally tell you, turn up at six PM at this restaurant and you will meet seven random YC founders and you can chat about your startups. So the thing that I found most valuable as a founder in London was meeting groups of peers that I would then share experiences with. So we're trying to curate those experiences for people so they get more of a community. If it works for the batch, We'll start doing it for alumni as well. So you can say I'm in London and we're going to do 20 YC dinners next week of all the YC alums so they get to meet each other and just make that network of YC even more powerful.

AI assessment note: “The thing I am trying to change, um, I ran an experiment last batch”

Q Ok, so we move then to, like, the application layer that sits on top of foundation models, which is where I presume you're excited. Yeah. Why, why is that? Because most, uh, it's a wrapper, it's non-sustainable. Why are you excited?

A There are clearly some wrappers. You know, if you can build it in a weekend at a hackathon and make a bunch of money, probably not defensible. For most businesses building on top of these models, I see it, you know, you can describe the last generation of startups as like, MySQL wrappers or AWS wrappers or something like, you know, it's the same kind of logic applies. I think where the sustaining value lies is identifying an industry, deeply understanding the regulation in that industry, the tooling, the language, the, um, all of the training, how people sort of work and behave and act, and tailor your software to fit into that industry in a way that's Extremely deeply embedded. Most people working in AI now, most people building application layer stuff in an AI, say it's 80 to 90% traditional software with 10% AI. It's working in construction, in construction, figuring out how Procore works, or, um, you know, how the Salesforce CRM works, or some Oracle database. And like, I don't think OpenAI is going to come and steamroll, like, the construction company, AI companies, because they're not going to deeply integrate into the processes and software that exists in each of those industries. So I really believe everyone who works with a computer will have an AI co-pilot assistant thing in the next two or three years, whether you're like an oncologist or a law professor.

AI assessment note: “where the sustaining value lies is identifying an industry, deeply understanding the regulation”

Q just going, I get it, but the trouble is you don't have distribution, and what I worry about is distribution trumps product in this world, which is where Microsoft Office just bake it in, And because of Microsoft's distributing capabilities into all, uh, was it oncologists or dentists or whatever we want to use as that famous dentist example. Fucking dentists. Poor dentists. But do you see what I mean?

A I do, but I don't agree because I think Microsoft Office, like, by definition has to be a general product. It will be the best word processor for sure, but it's not going to be the tool that the dentist is actually using to write their clinical notes after a meeting. Um, it's gonna be, that is a specialist piece of dentistry software, or the three D modeling software that some, um, architect is using to, to design a skyscraper. Microsoft Office is, it will benefit for sure Google Docs and Microsoft Office as a general document writing thing, but, um, one of our companies was called, uh, Solve Intelligence, which is a tool for patent writing. So taking scientific inventions and papers and turning into really, really high quality Patent submissions are very likely to be accepted. I don't think you can tell me Microsoft Office is going to help write those patents.

AI assessment note: “I do, but I don't agree because I think Microsoft Office, like, by definition”

Q Do you find the advice that you're giving these startups is different to the advice that you gave in a completely pre-AI world?

A Big companies are now more willing to spend money on this stuff than they ever were. So the previous advice of, like, go and sell to other startups, I think is actually not good for very, and maybe a dev tool startup, go and sell to other startups, but in general, I push companies, B to B SaaS companies, To start mid-market and go up as quickly as they can. Every manager of every big company in the world is being asked by their boss, what is AI going to do to our company? And many of them do not have a good answer, and they're looking around for startups to help them. And for sure, it's innovation budgets and proof of concepts and all this stuff. Um, but where it works, they convert to real contracts. I've seen companies go from zero to like half a million or a million of revenue within the YC batch. With really, really big companies, like Fortune, 500 companies, and turn them into recurring contracts. This is real. This is saving people so much time. It's astonishing.

AI assessment note: “So the previous advice of, like, go and sell to other startups, I think”

Q Final, final one, I promise. Do you ever challenge in terms of loss of identity? Like, I don't go on holiday because you're detached from work, and you mentioned me starting since I've never ever done anything else but this. So it's really uncomfortable actually to have 10 days or seven days with nothing.

A Yeah, this was a big problem for me when I, when I was thinking about leaving, the sort of year or so before I actually left. I couldn't imagine myself and the company not being the same thing. And who I would be, or what I would be, or whether I would be relevant anymore, or I was worried the company wouldn't survive without me, which was a very arrogant and untrue thing to think. It thrived without me. Um, and I had to rebuild my, you know, my sense of identity, my ego. And honestly, going to America where no one knew who I was, and joining YC as basically an intern, was incredibly humbling and really positive, because I had to, like, rebuild my entire self-image. At Monzo, anytime I said anything, A whole team would, like, spring into action, and even if I didn't intend them to do it, I'd be very careful what I said, because literally I'd come back two weeks later and find 30 people had gone off in crazy tangent. And I joined YC, and I say things, and people, like, ignored me. And not in a nasty way, but I was just, like, just another person, right? I was a visiting partner, I was a teaching assistant. My ideas didn't matter that much, and that was so humbling. Um, it was great. Honestly, really, really great.

AI assessment note: “this was a big problem for me when I, when I was thinking about leaving”

Q I mean, speaking of kind of justifying the ROI on those campaigns, I'm always super intrigued by a lot of people saying about the consumer downturn with kind of incumbency advances, meaning they're just bidding up CAC on these traditional platforms. How do you think about maybe a lack of free and open distribution today with regards to Unitecon customer acquisition at scale?

A I think you've got Got to, got to play a different game. If you try and compete head on in the games, the big banks are playing, which is basically rewards for opening an account, rewards for switching your salary. Yeah. You're going to get killed. Typically you're seeing CACs of 152 102 150 pounds. And those are just not figures currently we can compete with. We burn through all our cash. We spend really all of our attention on product growth. So we have teams looking at the pirate metrics funnel, right? It's awareness, Acquisition, activation, retention, referral, and then eventually revenue. So those steps need to put product teams on each of them and really sort of eke out one, two, three percent incremental grain gains every month, but then they compound and really, really add up. So our growth rate in the last three or four months has doubled based on product growth alone, which is free zero cap. And I think we've got, we can double that again. That's the stuff that banks are really, really bad at. I spoke to one bank executive, um, about their signup flow. And I recently tried to sign up for an account and remarked on how difficult it It was to open an account with this person's bank. And he sort of smiled and said, yes, we, we only really want the customers who are, who are really bought in, who really go through that, you know, really want the account. It's like, you w…

AI assessment note: “We spend really all of our attention on product growth.”

Q Tom Blomfield at Y Combinator, you have this incredible perspective. You see so many thousands of companies that apply for YC every year. How do you think about the wrappers versus non-wrappers in AI, and then also bluntly where the excitement is, whether it's in application layer versus infrastructure layer?

A There are clearly some wrappers. You know, if you can build it in a weekend at a hackathon and make a bunch of money, probably not defensible. For most Businesses building on top of these models. You know, you can describe the last generation of startups as like MySQL wrappers or AWS wrappers or something like, you know, it's the same kind of logic applies. I think where the sustaining value lies is identifying an industry, deeply understanding the regulation in that industry, the tooling, the language, the, all of the training, how people sort of work and behave and act and tailor your software To fit into that industry in a way that's extremely deeply embedded. Most people building application layer stuff in an AI say it's 80 to 90% traditional software with 10% AI. It's working in construction, in construction, fingering out how Procore works or, you know, how the Salesforce CRM works or some Oracle database. And like, I don't think OpenAI is going to come and steamroll the construction company, AI companies, because they're not going to deeply integrate into the processes and software

AI assessment note: “where the sustaining value lies is identifying an industry, deeply understanding the regulation”

Q When you see him as CEO of YC and what he's done, the excitement that he brings, especially around Demo Day, when you see him taking photos with the old school camera and just the joy that he, I didn't mean it horribly, has brought back it then to the YC community. What do you think makes him so good as a CEO of YC?

A I think YC has changed a lot in the last two years for the better, and I put it, a lot of it down to Gary. Um, He was there in the early days. Um, when I was there in 20 11, he was a, I think, a part-time partner. He helped us design our pitch deck for GoCardus in 20 11, and he was taking the photographs back then. Um, so I think he, like, experienced YC at its purest, at its core. And then really, honestly, under Sam Altman, I think it expanded in a bunch of different weird areas, and, um, it just lacked strong, and Sam was distracted. He was doing OpenAI from very, very early on. Um, and so to have a CEO now who is, whose full focus is the batch. It's a core YC. It's not some growth program. It's not something international program. It's not, you know, creating AGI. It's the core YC batch that Gary has experienced and loves and cares about. I think is, has made it great.

AI assessment note: “to have a CEO now who is, whose full focus is the batch”

Q Um, okay, so we have that. Then what?

A So, then I spent all my money. This money I was supposed to deploy over three or four years, I, uh, went through in nine or 10 months, and I was finding a lot of work, you know, signing that many documents, wiring, simply making the wire transfers for that number of investments is Very, I spent a lot of time on the phone with my, with my banker and it was just painful and it was lonely. So I, I spoke with Matt Robinson a lot. I spoke with Carlos. He's gone to index who are very, very good angels. I was trying to learn my, learn the craft, but I found it a lonely process. And so YC serendipitously like came along at just the right time and said, would you like to join us? You can invest our money. We've got a finance and ops and legal team that'll do all the admin for you. And you'll have a team of partners around you that you can learn from. Which is all of the things I wanted. Um, and they said the only catch is you have to move to Silicon Valley.

AI assessment note: “So, then I spent all my money. This money I was supposed to deploy”

Q A lot of founders worry, yeah, but I'm going to have a super specialized product then, and I can't sell that big vision to VCs. What do you say to them?

A So this is a key skill of a founder, which is holding these two realities in your head simultaneously without the cognitive dissonance or driving yourself crazy. One is the big vision of 10 years. Say the, the one percent best outcome, if this really, really works, what could this become? And holding that in your head. And for Monzo it was, we're building a bank for a billion people around the world. That's the big vision that you have to hold in your head. And then you have to hold the, what is my top priority today and this week and this month, which is very, very different from this billion people around the world, right? And you have to execute on that and get your team to focus on it. But you have to have both. If you only have the big vision and you think that's today, you're a bullshitter. You're just full of hot air. None of it's real. Whereas if you're too execution focused and you continue that for several years, you build a small business. You build a very successful, or rather a medium-sized small business that's very profitable, it's never going to get big. So you have to have both simultaneously.

AI assessment note: “holding these two realities in your head simultaneously without the cognitive dissonance”