The Exchanges

Q Tom Hume at GV, I'd love to hear your thoughts. How do you think about that, the commoditization of the models? And whether there's money to be made investing in the foundation model themselves, or actually in the application layer beneath it.

A My first observation would be the technology is commoditizing incredibly quickly, which worries me a lot. So I think I likened when we talked the other day it to investing a few hundred million into a power station. That's the training time, and then you can turn it on and you've got inference coming out the side. That's your power. Now the problem is, this is an industry where it's going to take you a few months to build your power station, and everyone else is building similar power stations next door with relatively little edge. They're still, they're all using the same GPUs. They're marginal improvements. But you've basically got to depreciate that asset in these foundation models over a few months. I just can't see it happening. And then now, we've got Meta coming into the market. I mean, Zuckerberg's done an amazing job. He'll have 350,000 H-one hundreds by the end of this year. That is 14% of the world's H-one hundreds, and he's going to open source the result. Llama III released last week is already incredible. He's pledged that he's going to invest another hundred billion dollars or so. He's already started to train Llama IV, that team, and they're world class. They're formidable competitors. So to invest now in an asset that you think you're going to have to depreciate, you know, over the space of weeks or months is very difficult to do. Now we have made investments i…

AI assessment note: “we have made investments in Gen AI, but more in infrastructure, more in the application layer”

Q to be made investing in foundation models? When you look at the quantum of capital that is required to go in, you know, there's obviously rumors of Mr. Isle's new funding around. You see the amount of cash that's gone into OpenAI and everyone else. There's a dilution inherent within that. It's just going to be monstrous. Is there money to be made investing in foundation models, do you think?

A So there definitely has been, because if you were to invest in OpenAI in the ten billion dollar round, there's liquidity in the market. You could sell that for a five X now, and you could have done that over a year. So if you've got a momentum strategy and you believe that firm that you're investing in is going to be at the front of the pack and continue to be, I suspect there's money to be made. But if you're investing in fundamentals, it's very difficult to invest in something that actually is going to commoditize that quickly. One of the, one of the, in fact, I'd say the best teacher I ever had was Clay Christensen, just unbelievably smart human being. He wrote The Innovator's Solution. We all know that. And he will talk about, or he did talk about sustaining and disruptive innovations. I think one of the frustrations with Gen AI, as the technology is commoditizing so quickly, is it's a sustaining innovation. It's actually going to get sprinkled across all businesses to lower costs in call centers or to improve the product in personalization. But it's actually not going to create the sort of, it's not going to have a creative destruction effect like the internet did on many industries. And so as an investor, that's frustrating because you want to invest in stuff that persists and completely rebuilds industries from scratch, but I can't really see it. I mean, we found some ta…

AI assessment note: “I suspect there's money to be made. But if you're investing in fundamentals, it's very difficult”

Q You mentioned kind of the doomers of AGI versus those who all think it'll be much further out. What are the observations on those and the capital requirement needs that they have?

A Oh yeah, it's fascinating to me, and I've never seen anyone sort of share this, but I think there is a Correlation between how aggressively people predict AGI is coming and how much capital they need to raise. So you've got Emodi from Anthropic, you've got Altman, OpenAI, Musk, XAI, all of these guys are saying that it's just around the corner and this industry is moving incredibly quickly. Then in contrast, you've got Zuckerberg at Meta, you've got Demis Asabis at DeepMind, and they're a lot more cautious. They don't need to raise money. And so I wonder, my general bias would be to look to the people that don't have to raise money for their point of view, assuming they're all equally smart.

AI assessment note: “Correlation between how aggressively people predict AGI is coming and how much capital they need to raise.”

Q If we go back to you, I love how kind of this has taken winding turns. You mentioned, obviously, you don't know that you're good at investing, especially kind of angel investing, and that was one of the questions you wanted to ask. Results say a lot. What were the results, and how do they look today?

A Yeah, so, I mean, you could take a batch of my angel investments Pre- twenty-fifteen. And I tracked this roughly 27 portfolio companies, about 4.5 X DPI, about 25 X or 24 X TVPI. So the results look pretty good. Superficially, I'd look at that and say, oh, maybe I'm a good investor, but there's two problems to that logic. Number one is my results were pretty good in that period, but I actually think we've had a massive regime change since. That's what you call it in machine learning. At that time, I was investing in startups at four million pre. It's not going to happen now. I mean, a dot AI domain name is more worth more than four million pre. So that's problematic. The second thing, which is incredibly humbling, is if you'd asked me to stack rank that portfolio through that period, say in 2010, I would have got it all wrong. So I can't be that smart about predicting success when I would have actually stack ranked my own portfolio badly.

AI assessment note: “27 portfolio companies, about 4.5 X DPI, about 25 X or 24 X TVPI”

Q is turning into, which is great. That is why I don't believe in reserves models, because exactly to your point there, if you were forced to stack rank, great, meh, and not good, it is not what you would have predicted. Therefore, I think we overestimate our ability to predict our winners. Do you agree? How do you think about what you just said and how that leads to reserves?

A Yeah, so I guess to sticking with the angel, Pat, because I think that's really important. I agree with you. It led me to draw the conclusion as an angel, I shouldn't follow on. So I had examples of companies that would go up sort of 50 X, my pro rata allocation in the next round would be a million dollars plus, and they went to zero. And not only that, you're then competing with VCs. It's a completely different game. So actually, if I'd taken a strategy of doing all my follow on allocations, let's say for the series A after the seed I'd invested in, It would have not wrecked the portfolio, but it would have been far inferior.

AI assessment note: “I agree with you. It led me to draw the conclusion as an angel, I shouldn't follow on.”

Q is turning into, which is great. That is why I don't believe in reserves models, because exactly to your point there, if you were forced to stack rank, great, meh, and not good, it is not what you would have predicted. Therefore, I think we overestimate our ability to predict our winners. Do you agree? How do you think about what you just said and how that leads to reserves?

A Yeah, so I guess to sticking with the angel, Pat, because I think that's really important. I agree with you. It led me to draw the conclusion as an angel, I shouldn't follow on. So I had examples of companies that would go up sort of 50 X, my pro rata allocation in the next round would be a million dollars plus, and they went to zero. And not only that, you're then competing with VCs. It's a completely different game. So actually, if I'd taken a strategy of doing all my follow on allocations, let's say for the series A after the seed I'd invested in, It would have not wrecked the portfolio, but it would have been far inferior.

AI assessment note: “I agree with you. It led me to draw the conclusion as an angel, I shouldn't follow on.”

Q You mentioned kind of the doomers of AGI versus those who all think it'll be much further out. What are the observations on those and the capital requirement needs that they have?

A Oh yeah, it's fascinating to me, and I've never seen anyone sort of share this, but I think there is a Correlation between how aggressively people predict AGI is coming and how much capital they need to raise. So you've got Emodi from Anthropic, you've got Altman, OpenAI, Musk, XAI, all of these guys are saying that it's just around the corner and this industry is moving incredibly quickly. Then in contrast, you've got Zuckerberg at Meta, you've got Demis Asabis at DeepMind, and they're a lot more cautious. They don't need to raise money. And so I wonder, my general bias would be to look to the people that don't have to raise money for their point of view, assuming they're all equally smart.

AI assessment note: “Correlation between how aggressively people predict AGI is coming and how much capital they need”

Q If we go back to you, I love how kind of this has taken winding turns. You mentioned, obviously, you don't know that you're good at investing, especially kind of angel investing, and that was one of the questions you wanted to ask. Results say a lot. What were the results, and how do they look today?

A Yeah, so, I mean, you could take a batch of my angel investments Pre- twenty-fifteen. And I tracked this roughly 27 portfolio companies, about 4.5 X DPI, about 25 X or 24 X TVPI. So the results look pretty good. Superficially, I'd look at that and say, oh, maybe I'm a good investor, but there's two problems to that logic. Number one is my results were pretty good in that period, but I actually think we've had a massive regime change since. That's what you call it in machine learning. At that time, I was investing in startups at four million pre. It's not going to happen now. I mean, a dot AI domain name is more worth more than four million pre. So that's problematic. The second thing, which is incredibly humbling, is if you'd asked me to stack rank that portfolio through that period, say in 2010, I would have got it all wrong. So I can't be that smart about predicting success when I would have actually stack ranked my own portfolio badly.

AI assessment note: “roughly 27 portfolio companies, about 4.5 X DPI, about 25 X or 24 X TVPI.”

Q Okay. So when we look at, as you said, they're kind of 4.6 DPI. Yeah. So when we look at kind of winners and zeros, what are some of the biggest lessons from the winners?

A Yeah. So the, look, the biggest lessons from the winners that generally, if you want to build, one of the paradoxes of our business is we need to invest in people that are doing difficult things. If it's easy, it's going to be commoditized. So if people are doing difficult things, it usually takes a long time to create real value. And so I think my biggest winners are ones that were fundamental investments where they have just continued to grow for the decade since. Look, GoCardless would be an example where the business kind of nine figure ARR, they have just continued growing. It was never a super hot business, but I invested at the point they were at YC. And look, team of three taught themselves to code and hustled for years afterwards, built a great business. You contrast that with others, which, you know, in very quick periods would have looked quite good. So I invested in a company that was acquired by fab.com. So I thought I had a lot of money in equity in fab at one point. I invested in really brilliant team in California called Massive Health that were acquired by Jawbone. I thought that was going to be very valid valuable. Both of those businesses went to zero. So if I look at it, the ones that worked for me were the fundamentals where they just grew great businesses over time. The others were kind of momentum plays, and I wasn't smart enough, nor do I think I had the…

AI assessment note: “my biggest winners are ones that were fundamental investments where they have just continued to grow”

Q am an optimist. You're right. There are a couple of things I just want to talk about on like building companies and, you know, from the founder perspective, we often hear the hailed, ah, it's great to have naivety as a founder. Do you, do you love the outsider to a market who's kind of naively optimistic, or do you like insider to a market who knows the mechanics well?

A Look, that is a great question, and this is going to be a frustrating Answer to someone that probably wants to have a sort of soundbite from it, but it depends. And actually I've got excited about both before. So look, I'll give you an extreme example. We invested in Lemonade, Series A, wonderful founders. One, Shai, had done Fiverr, product and design guy. Daniel, lawyer, had done hardware, and they were going into insurance. Combined insurance experience between those two? Zero. What was their unfair advantage? They understood, like through his, Daniel's legal background, he knew how to manage a business. They understood the tech side, and they would have an incredible clock speed. They were releasing on a daily basis, whereas that industry is every month. So their unfair advantage was speed, and they recruited in depth. They brought in amazing people like Tim, who had actually deep industry insurance experience. So in that case, I was happy to invest in the naive approach. Other end of the spectrum, We invested in CurrencyCloud. Mike Lavin, the founder, I think had 30 years experience in fintech. He was so well placed to understand actually what those buyers wanted and ultimately sold the business to Visa for a billion dollars. Like he was the, not the ultimate insider, but he had real depth. I am absolutely comfortable with both of those approaches, but my question is for t…

AI assessment note: “it depends. And actually I've got excited about both before.”

Q I think of it in like engineering talent, and then I think of it in founding talent. How do you think that differs? Like, as you said there, net net for like deep AI engineers. I think my worry is actually, do we actually have the founder supply that is exceptional that maybe other countries do? And that's the difference that I think about.

A From my perspective, so I completely agree. I think we're rate limited. I think it's the biggest rate limiter actually is supply of founders and supply of operators. The great thing about founders is they'll smash through walls to build stuff. Uh, so you have Melanie at Canva built You know, built that business in Perth, Australia. No right to build a fifty billion dollar business in Perth, but it can be done. If you gave me the choice to have more, uh, Nicholas Zenstrom's or Demis Hassabis's or Stan's, I would absolutely take that. I think, you know, it could only be a good thing. The biggest challenge is for every one good founder, you need five or 10 world-class operators, and I think that's the biggest gap for us. That's the rate limiter. To Stan's point, if I just look at engineering talent, We've got three of the best 10 universities on the planet here. If you look at Oxford, Cambridge, Imperial, they're only graduating between them about 500 computer scientists or roboticists per year. We should five X that number. There's a huge demand. I don't see why we aren't increasing it. And then to Stan's point, we can do a better job of actually making it appealing to come into the UK for the most entrepreneurial talent and maybe retain the talent that does study here. And becomes expert.

AI assessment note: “I think the biggest rate limiter actually is supply of founders and supply of operators.”

Q I think of it in like engineering talent, and then I think of it in founding talent. How do you think that differs? Like, as you said there, net net for like deep AI engineers. I think my worry is actually, do we actually have the founder supply that is exceptional that maybe other countries do? And that's the difference that I think about.

A From my perspective, so I completely agree. I think we're rate limited. I think it's the biggest rate limiter actually is supply of founders and supply of operators. The great thing about founders is they'll smash through walls to build stuff. Uh, so you have Melanie at Canva built You know, built that business in Perth, Australia. No right to build a fifty billion dollar business in Perth, but it can be done. If you gave me the choice to have more, uh, Nicholas Zenstrom's or Demis Hassabis's or Stan's, I would absolutely take that. I think, you know, it could only be a good thing. The biggest challenge is for every one good founder, you need five or 10 world-class operators, and I think that's the biggest gap for us. That's the rate limiter. To Stan's point, if I just look at engineering talent, We've got three of the best 10 universities on the planet here. If you look at Oxford, Cambridge, Imperial, they're only graduating between them about 500 computer scientists or roboticists per year. We should five X that number. There's a huge demand. I don't see why we aren't increasing it. And then to Stan's point, we can do a better job of actually making it appealing to come into the UK for the most entrepreneurial talent and maybe retain the talent that does study here. And becomes expert.

AI assessment note: “I completely agree. I think we're rate limited. I think it's the biggest rate limiter”

Q Sometimes in my head, I think, how many friends do I want to lose in one single show? Um, my question to you, I mean, I, I, I don't agree that prices are better here, honestly. Like, for the best companies, for your word, whereas if they were to stay, they're just super high, they're so inflated.

A I think just a quick thought on, no, no, I, so if I look at where we sit today, some of the best deals are overpriced. I think it's often because they're the ones with the traction, uh, And they're therefore somewhat de-risked. And I think there's two things that make this a really difficult thing to answer. We talked about lagging indicators. The first is we're basically trading against, or we're working against sources of capital that were raised in the past. Like these are not brand new funds often, and often they were raised in ZERP. The cost of capital has gone through the roof, like given the current interest rate environment. I think that's going to get worse, if anything. The fact that a lot of these funds are giving out so many stock grants, you basically need to hit 20% IRR to break even. These numbers are really high. So that's the first thing. I actually think there's probably going to be less money in the market for venture in two years than there is today. It's kind of a question for us. And then the second thing is classic machine learning. I think we're overfitting to history. I don't think we know what the biggest companies look like going forward. And so it's very difficult for me to Just say that actually the sort of returns profile that funds got from investments 10 years ago, they're going to be the ones they looked like before. My belief is that AI is crea…

AI assessment note: “some of the best deals are overpriced. I think it's often because they're the ones with the traction”

Q there money to be made investing in foundation models? When you look at the quantum of capital that is required to go in, there's obviously rumors of Mr. R's new funding around, you see the amount of cash that's gone into OpenAI, and everyone else, there's a dilution inherent within that, it's just going to be monstrous. Is there money to be made investing in foundation models, do you think?

A There definitely has been, because if you were to invest in OpenAI in the ten billion dollar round, there's liquidity in the market. You could sell that for a five X now, and you could have done that over a year. So if you've got a momentum strategy and you believe that firm that you're investing in is going to be at the front of the pack and continue to be, I suspect there's money to be made. But if you're investing in fundamentals, it's very difficult to invest in something that actually is going to commoditize that quickly. In fact, I'd say the best teacher I ever had was Clay Christensen. Just unbelievably smart human being. He wrote The Innovator's Solution. We all know that. And he will talk about, or he did talk about sustaining and disruptive innovations. I think one of the frustrations with Gen AI, as the technology is commoditizing so quickly, is it's a sustaining innovation. It's actually going to get sprinkled across all businesses to lower costs in call centers or to improve the product in personalization. It's not going to Have a creative destruction effect like the internet did on many industries. And so as an investor, that's frustrating because you want to invest in stuff that persists and completely rebuilds industries from scratch. But I can't really see it. I mean, we found some targets and we made quite a few investments, but it's not for me the sort of rad…

AI assessment note: “There definitely has been, because if you were to invest in OpenAI in the ten”

Q Okay. So when we look at, as you said, they're kind of 4.6 DPI. Yeah. So when we look at kind of winners and zeros, what are some of the biggest lessons from the winners?

A Yeah. So the, look, the biggest lessons from the winners that generally, if you want to build, one of the paradoxes of our business is we need to invest in people that are doing difficult things. If it's easy, it's going to be commoditized. So if people are doing difficult things, it usually takes a long time to create real value. And so I think my biggest winners are ones that were fundamental investments where they have just continued to grow for the decade since. Look, GoCardless would be an example where the business kind of nine figure ARR, they have just continued growing. It was never a super hot business, but I invested at the point they were at YC. And look, team of three taught themselves to code and hustled for years afterwards, built a great business. You contrast that with others, which, you know, in very quick periods would have looked quite good. So I invested in a company that was acquired by fab.com. So I thought I had a lot of money in equity in fab at one point. I invested in really brilliant team in California called Massive Health that were acquired by Jawbone. I thought that was going to be very valid valuable. Both of those businesses went to zero. So if I look at it, the ones that worked for me were the fundamentals where they just grew great businesses over time. The others were kind of momentum plays, and I wasn't smart enough, nor do I think I had the…

AI assessment note: “the ones that worked for me were the fundamentals where they just grew great businesses”

Q to be made investing in foundation models? When you look at the quantum of capital that is required to go in, you know, there's obviously rumors of Mr. Isle's new funding around. You see the amount of cash that's gone into OpenAI and everyone else. There's a dilution inherent within that. It's just going to be monstrous. Is there money to be made investing in foundation models, do you think?

A So there definitely has been, because if you were to invest in OpenAI in the ten billion dollar round, there's liquidity in the market. You could sell that for a five X now, and you could have done that over a year. So if you've got a momentum strategy and you believe that firm that you're investing in is going to be at the front of the pack and continue to be, I suspect there's money to be made. But if you're investing in fundamentals, it's very difficult to invest in something that actually is going to commoditize that quickly. One of the, one of the, in fact, I'd say the best teacher I ever had was Clay Christensen, just unbelievably smart human being. He wrote The Innovator's Solution. We all know that. And he will talk about, or he did talk about sustaining and disruptive innovations. I think one of the frustrations with Gen AI, as the technology is commoditizing so quickly, is it's a sustaining innovation. It's actually going to get sprinkled across all businesses to lower costs in call centers or to improve the product in personalization. But it's actually not going to create the sort of, it's not going to have a creative destruction effect like the internet did on many industries. And so as an investor, that's frustrating because you want to invest in stuff that persists and completely rebuilds industries from scratch, but I can't really see it. I mean, we found some ta…

AI assessment note: “if you've got a momentum strategy... I suspect there's money to be made.”

Q Listen, I, I do totally agree with you. So, if we move then to the application layer, how do you determine between, like, sustainable value generation versus, like, I think we see with quite a lot, like, flash in the pan, fast revenue scaling, but not sustainable value generation opportunities?

A So, my colleague we do in London invested in Synthesia, which I think is an interesting business. So, you know, that is a business that creates synthetic video. They do it into learning and development environments, and there wasn't really an incumbent there. So they've concentrated on building a whole go to market business. And I think this to me is what's important in the application layer. You better have something proprietary in terms of data or distribution. In their case, they're just building an end to end enterprise ready solution with security and everything that enables you to spin up the videos. So those are the sorts of things that we're looking for. I like the framework that I think Sam Altman on your pod said, which is the easiest way to look at applications In gen AI or to cut them is to say to yourself, are they happy or devastated if the model's improved by 100 X? I've got to look for businesses that are happy they're going to improve by 100 X. Otherwise, it's just ephemeral in the same way as I think the foundation models are ephemeral.

AI assessment note: “You better have something proprietary in terms of data or distribution.”

Q A lot of cultural debt, I think, happened, especially in COVID when people were fully remote, never actually met their teammates at all. Um, what are your biggest observations when you compare remote versus in-person and the cultures that you have in the portfolio?

A Yeah, I mean, it's an amazing array. So I, I, I'd first carve out companies that are sort of natively remote. So Dave Minicello, who, you know, led our investment in GitLab at the Series A. Phenomenal investment. Never had an office. They also, Sid, the founder is incredibly thoughtful that actually he's very careful that all meetings are recorded asynchronously. Everyone can, you know, contribute to the board meeting document. They get the whole team together. It's not a great cost saving thing. It just works for them. So look, there's native remote businesses that work very well. Then there's everyone else that was sort of forced to be native during COVID and I think executed badly. And generally went through this arc where they said, oh, this is really good. Everyone can work at home. Everyone works really efficiently. Then over time, everyone realized, actually, we're not innovative. The junior people on the team are not learning. People are getting frustrated. Like, you and I went for walks during that period, and I remember us joking that we just became, like, we became very efficient at our jobs by doing 30 Zoom calls a day, but actually we'd become efficient at our jobs by taking all the fun bits out.

AI assessment note: “there's native remote businesses that work very well. Then there's everyone else”

Q to misunderstand that kind of models are variable, and that when you change a certain tax rate, uh, you will see, you know, people leave. Um, We've seen the removal of non-doms. I'm really worried about this. Every single day I have friends saying, hey, I'm leaving, I'm leaving, why are you staying? To what extent is the removal of non-doms a massive problem impacting the future of the UK?

A I, so that I, um, I think this is one of those classic cases of whether you want a sort of principled approach or a pragmatic approach. I'm a pragmatist. If I, uh, I do say, see the brain drain. I recognize it. And I do see that many of the people I know well that have chosen to leave have left. They were also incredible angel investors. They, uh, employed a bunch of people. And so do I think everyone should pay Equal tax. Yes, in principle, but practically speaking, I would rather that talent was in the UK. I mean, I am seeing some exceptions to that. I heard about a billionaire VC who, you know, I think has moved to the UK recently. You do get some movement back in the other direction, but I would take seriously, again, leading and lagging indicators, I would take seriously the leading indicator of some of the non-DOMs leaving.

AI assessment note: “I would take seriously the leading indicator of some of the non-DOMs leaving.”

Q to misunderstand that kind of models are variable, and that when you change a certain tax rate, uh, you will see, you know, people leave. Um, We've seen the removal of non-doms. I'm really worried about this. Every single day I have friends saying, hey, I'm leaving, I'm leaving, why are you staying? To what extent is the removal of non-doms a massive problem impacting the future of the UK?

A I, so that I, um, I think this is one of those classic cases of whether you want a sort of principled approach or a pragmatic approach. I'm a pragmatist. If I, uh, I do say, see the brain drain. I recognize it. And I do see that many of the people I know well that have chosen to leave have left. They were also incredible angel investors. They, uh, employed a bunch of people. And so do I think everyone should pay Equal tax. Yes, in principle, but practically speaking, I would rather that talent was in the UK. I mean, I am seeing some exceptions to that. I heard about a billionaire VC who, you know, I think has moved to the UK recently. You do get some movement back in the other direction, but I would take seriously, again, leading and lagging indicators, I would take seriously the leading indicator of some of the non-DOMs leaving.

AI assessment note: “I would take seriously the leading indicator of some of the non-DOMs leaving.”

Q there money to be made investing in foundation models? When you look at the quantum of capital that is required to go in, there's obviously rumors of Mr. R's new funding around, you see the amount of cash that's gone into OpenAI, and everyone else, there's a dilution inherent within that, it's just going to be monstrous. Is there money to be made investing in foundation models, do you think?

A There definitely has been, because if you were to invest in OpenAI in the ten billion dollar round, there's liquidity in the market. You could sell that for a five X now, and you could have done that over a year. So if you've got a momentum strategy and you believe that firm that you're investing in is going to be at the front of the pack and continue to be, I suspect there's money to be made. But if you're investing in fundamentals, it's very difficult to invest in something that actually is going to commoditize that quickly. In fact, I'd say the best teacher I ever had was Clay Christensen. Just unbelievably smart human being. He wrote The Innovator's Solution. We all know that. And he will talk about, or he did talk about sustaining and disruptive innovations. I think one of the frustrations with Gen AI, as the technology is commoditizing so quickly, is it's a sustaining innovation. It's actually going to get sprinkled across all businesses to lower costs in call centers or to improve the product in personalization. It's not going to Have a creative destruction effect like the internet did on many industries. And so as an investor, that's frustrating because you want to invest in stuff that persists and completely rebuilds industries from scratch. But I can't really see it. I mean, we found some targets and we made quite a few investments, but it's not for me the sort of rad…

AI assessment note: “There definitely has been, because if you were to invest in OpenAI in the ten”

Q Tom Hume, my friend, what about you? Would you invest in OpenAI at a ninety billion dollar valuation?

A I would struggle to make that investment today. And it's not because I don't respect the team. My biggest concern at the moment, but if I observe the emergence of what Meta's doing, if I look at the arms race of what the cloud providers are investing in and the sort of Gemini, et cetera, any advantage is pretty ephemeral and the consumer facing product that doesn't, that drives I don't know, is it 50% of the revenue? Something like that is not sticky. So to invest in a foundation model, what would I want to be true? I would want to believe that they had some unique approach that made them more defensible. So an obvious one is memory. Like actually none of these have cracked memory yet, but if you have a personal assistant, a ChatGPT equivalent, and it, it remembers so that it can actually be applied Probabilities as to what you want going forward, then it's interesting. If it's unique in its ability to take agency, then it might be interesting. There's other orthogonal approaches that might be interesting. But if we're just talking about a foundation model where you're going to throw huge amounts of data, hundreds of millions of dollars of compute at H-one hundreds, like everyone else, it's very difficult to see a return on these investments.

AI assessment note: “I would struggle to make that investment today.”

Q Do you know what? I knew it was a mistake doing this. It's just your feedback, Harry. I thought you were going to say it's just my founders. I was like harsh. No, in terms of the investing side, Tom, I do want to parlay into that. How did you start the angel investing? And what did that entry look like?

A Yeah, I was lucky enough to sell a company. I decided angel investing looked fun. And I thought, right, I'm going to run an experiment. I'm pretty entrepreneurial, so I thought I'm going to try and answer a couple of questions. And the questions I wanted to answer, first question is, what kind of investor am I going to be? And actually, do I enjoy it? Second question was, am I any good at it? So first question, actually, one of my early investors said something I think is really smart to me. They said, there are basically three types of investors. You've got smart investors that know they're smart and they're going to add value. Then you've got passive investors, That are going to stay passive and they're not going to get in the way. Both of those are absolutely fine. You need to avoid investors that are passive or sometimes even dumb, but think they smart and actually get a kind of interfere. And so I set myself, am I going to be kind of smart, smart, or passive, passive, and am I going to enjoy it? Over time, I think I added some value to those founders, particularly on strategy and design, because I would slowly work my way onto their kind of What we'd call in the UK a nine, nine, nine list, a list of people they call if something's going wrong. In the US, it would be a nine, one, one list probably. Uh, and I realized very quickly I absolutely loved that. I loved supporting …

AI assessment note: “I was lucky enough to sell a company. I decided angel investing looked fun.”

Q No, listen, I completely agree with you. Ok, so that's on, you mentioned kind of the, the kind of consistent compounding of great businesses, like your Go Cardless of the world, it's like your Lendables. On the zeros, what are the lessons from the bad investment decisions that you made as an angel?

A Yeah, it's, there's a lot. I think I got sucked into momentum and heat. I think I, one of the best things I ever did was surround myself by other angels, and there were only a handful in London at the time, and learn from them. I think one of the worst things I did is I would occasionally kind of outsource discipline and due diligence to them. That's just too easy to do. I think too many people in our industry, particularly angels who have other jobs, full-time jobs, We'll take the view that probably someone else has done the work. I actually think often it's surprising how people haven't done the work. So that is a massive trap. It's a trap I've fallen into.

AI assessment note: “I would occasionally kind of outsource discipline and due diligence to them.”

Q am an optimist. You're right. There are a couple of things I just want to talk about on like building companies and, you know, from the founder perspective, we often hear the hailed, ah, it's great to have naivety as a founder. Do you, do you love the outsider to a market who's kind of naively optimistic, or do you like insider to a market who knows the mechanics well?

A Look, that is a great question, and this is going to be a frustrating Answer to someone that probably wants to have a sort of soundbite from it, but it depends. And actually I've got excited about both before. So look, I'll give you an extreme example. We invested in Lemonade, Series A, wonderful founders. One, Shai, had done Fiverr, product and design guy. Daniel, lawyer, had done hardware, and they were going into insurance. Combined insurance experience between those two? Zero. What was their unfair advantage? They understood, like through his, Daniel's legal background, he knew how to manage a business. They understood the tech side, and they would have an incredible clock speed. They were releasing on a daily basis, whereas that industry is every month. So their unfair advantage was speed, and they recruited in depth. They brought in amazing people like Tim, who had actually deep industry insurance experience. So in that case, I was happy to invest in the naive approach. Other end of the spectrum, We invested in CurrencyCloud. Mike Lavin, the founder, I think had 30 years experience in fintech. He was so well placed to understand actually what those buyers wanted and ultimately sold the business to Visa for a billion dollars. Like he was the, not the ultimate insider, but he had real depth. I am absolutely comfortable with both of those approaches, but my question is for t…

AI assessment note: “it depends. And actually I've got excited about both before.”

Q Are there any questions you always like to ask to determine the muster of someone? So like one that I always ask is like, how did you first make money? I don't think great, great entrepreneurs first made money from getting a job at Bain after three years at Oxford. Like they did something before.

A Yeah, I love that. I, I think usually you see some trait of entrepreneurship. I think the questions I like to ask actually revolve around their unfair advantage. I'm trying to understand what their unique insight is and why they are uniquely placed to solve it. So I'll ask what the unfair advantage is. I'll also ask the question of why now? Like, it's really interesting that we will sort of, we, we have our recency biases that something may have worked or may not have worked, but we need this difficult thing to happen now. With a founder at the seed stage. So I asked the question of why now, and usually they should have a good point of view. The other one I really love to do is spend some time actually asking them how something might go wrong. What's keeping them up at night? And it's incredible how many founders will actually have nothing to share about their concerns. I mean, only the paranoid survive, as Andy Gray said. Some people have no paranoia.

AI assessment note: “I think the questions I like to ask actually revolve around their unfair advantage.”

Q a really important one, but I do just want to stay on the, ah, kind of outcome scenario planning, because it's such a big part of venture, and it's always predicated around that, oh, we need a fund returner. That's why we do it. Do you agree about the importance of, you have to have fund returners, it's the only thing that matters, and how do you think about that?

A Yeah, no, I, I don't agree with it. I think, ah, a lot of VC strategy is a lagging indicator of what did work in the past, and the test, Or the experiment that worked very well in the past is funds with 25 portfolio companies, power law of returns, and one or two return the whole fund, and then everything else drives decent return and IRR for the LPs. That has absolutely worked, but just because that has worked doesn't mean other approaches can't. And I think we see from different PE models, we even see from debt models, there's other ways to actually be very successful at kind of growth stage. Uh, so I would not want to say that I would only structure a portfolio that can deliver or return or make investments that could, uh, return the whole fund. It doesn't make sense to me. The important thing is have a strategy and stick to it. You made this point in angel investing. It's super important. It's like actually have your strategy and stick to it. Don't fall in love with one company, throw your strategy out of the window and then dump the whole fund into it. Incredibly dangerous.

AI assessment note: “Yeah, no, I, I don't agree with it.”

Q A lot of cultural debt, I think, happened, especially in COVID when people were fully remote, never actually met their teammates at all. Um, what are your biggest observations when you compare remote versus in-person and the cultures that you have in the portfolio?

A Yeah, I mean, it's an amazing array. So I, I, I'd first carve out companies that are sort of natively remote. So Dave Minicello, who, you know, led our investment in GitLab at the Series A. Phenomenal investment. Never had an office. They also, Sid, the founder is incredibly thoughtful that actually he's very careful that all meetings are recorded asynchronously. Everyone can, you know, contribute to the board meeting document. They get the whole team together. It's not a great cost saving thing. It just works for them. So look, there's native remote businesses that work very well. Then there's everyone else that was sort of forced to be native during COVID and I think executed badly. And generally went through this arc where they said, oh, this is really good. Everyone can work at home. Everyone works really efficiently. Then over time, everyone realized, actually, we're not innovative. The junior people on the team are not learning. People are getting frustrated. Like, you and I went for walks during that period, and I remember us joking that we just became, like, we became very efficient at our jobs by doing 30 Zoom calls a day, but actually we'd become efficient at our jobs by taking all the fun bits out.

AI assessment note: “realized, actually, we're not innovative. The junior people on the team are not learning.”

Q So do you outcome scenario plan when you do deals? Cause you see a lot of people say, well, I just, I don't think it can be big enough. I don't think the market's got enough depth. Do you do outcome scenario plans? And how do you think about that given the challenge of seeing the next S curve?

A So we do, uh, and the way we frame it is we ask actually what is the kind of option value in this business? Like great founders understand the value of options. Like an example will be most of the best founders I've ever worked with have collected data without knowing really what it might be used for, but they've, they've instinctively known there's option value in it. They know that it might create some value. And so we try and serve a scenario plan by saying, okay, this is the plan. Do we have confidence in it? And then what's the upside? If this goes right, what opportunities might it unlock? And then obviously we reframe it and say, okay, what are the risks they're in? Premortem, how this might go wrong. And then you get a kind of balanced view of what all the outcomes might be. But then look, you and I have talked before. I don't understand in our industry how anyone can have complete conviction on anything. That makes no sense to me. Like I, I studied physics at university. We would go through a proof and I still didn't have complete conviction that I got it right. And then now we're in a venture capital industry and we see investments and we're supposed to have complete conviction that it's going to work. I can describe a bull and a bear case for every one of my portfolio.

AI assessment note: “So we do, uh, and the way we frame it is”

Q Do you know what? I knew it was a mistake doing this. It's just your feedback, Harry. I thought you were going to say it's just my founders. I was like harsh. No, in terms of the investing side, Tom, I do want to parlay into that. How did you start the angel investing? And what did that entry look like?

A Yeah, I was lucky enough to sell a company. I decided angel investing looked fun. And I thought, right, I'm going to run an experiment. I'm pretty entrepreneurial, so I thought I'm going to try and answer a couple of questions. And the questions I wanted to answer, first question is, what kind of investor am I going to be? And actually, do I enjoy it? Second question was, am I any good at it? So first question, actually, one of my early investors said something I think is really smart to me. They said, there are basically three types of investors. You've got smart investors that know they're smart and they're going to add value. Then you've got passive investors, That are going to stay passive and they're not going to get in the way. Both of those are absolutely fine. You need to avoid investors that are passive or sometimes even dumb, but think they smart and actually get a kind of interfere. And so I set myself, am I going to be kind of smart, smart, or passive, passive, and am I going to enjoy it? Over time, I think I added some value to those founders, particularly on strategy and design, because I would slowly work my way onto their kind of What we'd call in the UK a nine, nine, nine list, a list of people they call if something's going wrong. In the US, it would be a nine, one, one list probably. Uh, and I realized very quickly I absolutely loved that. I loved supporting …

AI assessment note: “I was lucky enough to sell a company. I decided angel investing looked fun.”