As a general matter, I like people who have a very specific idea. That they want to pursue and they can't stop thinking about it, the proverbial "I can't sleep at night, I can't stop proselytizing.”

• Generally, I don't like those kinds of founders. As a general matter, I like people who have a very specific idea. That they want to pursue and they can't stop thinking about it, you know, the proverbial "I can't sleep at night, I can't stop proselytizing," etc. And they must do this with their lives. And it's just a question of with whom they should do it. Not "Should I do it?"
• There are obviously, like all rules in venture capital, all rules in Silicon Valley are made to be broken. So there's exceptions where that's worked. But typically, I'd prefer people who are very specific. And they may discover some edges that aren't quite polished and they may polish those edges, and it may look a little bit different than when they started. But the direction of the idea was already sort of in their mind, maybe before they left their prior job.

Reid had this idea of a professional network in his mind every day at PayPal and I had a burning ambition for OpenDoor.

• This was true of LinkedIn. Reid had this idea of a professional network in his mind every day at PayPal. He was just waiting to try it. He had two other ideas. One of them was dating, because he had done the early dating side of the 90s. The other one was really bad. Really bad. Okay, so one thing we could do is filter the ideas. I had a burning ambition for OpenDoor. I've been sleeping on it for 10 years, which I wouldn't recommend waiting 10 years either. So those are probably the extremes. But definitely, it was something I've been thinking about for literally a decade before we got off the ground.

I tend to like things that are super ambitious, that almost are ridiculous. If half of my VC friends don't laugh, that means I'm not taking enough risk. And that means the projects aren't ambitious enough.

• So you can definitely see that some of them are easier to vet than others, like consumer products. It's a lot easier to vet enterprise ideas than consumer ones. For consumer ones, you may just have to build the damn thing. You know, get some real users and see what happens and see what they try to do, see what they like versus like your top-down analysis. The general filter I use, truthfully, is the more ambitious the better. I think startups have roughly the same amount of pain to go from zero to one, regardless of whether one is a $1 billion company, $100 million company, a $10 billion company, or a $100 billion company. So you basically have this fixed cost of pain. So you might as well do something important, because the pain is going to be the same, right? So you might as well get the outcome that offsets the pain. So I tend to like things that are super ambitious, that almost are ridiculous. Like I have this sort of saying both at KV now and at Founders Fund, that I want half of my VC friends to laugh at an investment I make. Yeah. And if half don't laugh, that means I'm not taking enough risk. And that means the projects aren't ambitious enough. I don't want all of them to laugh, but I want half.

I like the idea maze. You really want to understand the market and opportunity well enough that you can walk anybody through the trapdoors, the history, what looks alluring, and what isn’t? What’s the real path to the prize? And once you master something so well, then I think you’re ready to do it.

• I like the idea maze. You really want to understand the market and opportunity well enough that you can walk anybody through the trapdoors, the history, what looks alluring, and what isn’t? What’s the real path to the prize? And once you master something so well, then I think you’re ready to do it.
• I still can walk into a meeting at OpenDoor and be pretty, pretty knowledgeable on any topic, despite not being there full time. And despite the fact that there's 1600 people that are there full time, because I spent a decade thinking about it. So there's almost nothing in a decade that doesn't occur to you. It's sort of like you've seen these prison movies. When you're in prison, you're writing in your cell all the time, you think about a lot of things. So if you think about a stupid idea for a decade, you've thought through a lot of permutations and all the corner cases. And you know, when someone says something, it's like I actually have a solution to that, because I was actually thinking about that a year ago or in sleep.

As Tony was explaining his idea, I thought, 'Oh, this could be the 'I'm Hungry' button.' The hard part would be whether they could pull it off: broad selection, affordable, consistent delivery. All those things were really hard. But then also the lights went off, and I was like, 'Oh, this is a no-brainer investment as a seed round.

• So maybe I'll compare it to DoorDash. Tony had been an intern at Square, where I was a manager. When he told me he wanted to do a food delivery thing, I looked at him quizzically. Like, "Are you sure?" was my reaction. Then I thought about a comment that Andrew Mason, the founder of Groupon, had made publicly a while ago. He said that there should only be two buttons on your phone: one for "On Board" and one for "I'm Hungry." As Tony was explaining his idea, I thought, "Oh, this could be the 'I'm Hungry' button." And I was like, "Yeah, that's a pretty big market." The hard part would be whether they could pull it off: broad selection, affordable, consistent delivery. All those things were really hard. But the demand was infinite. I was like, "If you could do that at an affordable price reliably, people will click that button all day long." So then the question was, could Tony and his original team — which included Evan Moore and two other co-founders — were they the people that could figure out whether it was possible to actually do this? In my mind, the answer was yes. They had the right skill sets for what that was, which really was a logistics and math problem. You could break down why that's true. But then also the lights went off, and I was like, "Oh, this is a no-brainer investment as a seed round."

I think the real issue with TAM is that you don't create something that creates enough economic transformation, so you don't capture enough value. It's not that there's not enough people or enough customers, it's that you capture such a small amount because what you do is so marginally valuable, that it's hard to add up to be a massive company.

• I don't do TAM analysis for the most part, or maybe ever. Maybe that's somewhat intuitive, reacting to an idea and kind of triangulating to math, rather than actually thinking about building a spreadsheet or something. But I think the real issue with TAM is that you don't create something that creates enough economic transformation, so you don't capture enough value. It's not that there's not enough people or enough customers, it's that you capture such a small amount because what you do is so marginally valuable, that it's hard to add up to be a massive company. Said another way, the value proposition basically is too weak.

I would basically reframe the problem back to a founder as how do you make the value proposition more compelling? If the value proposition is more compelling, you're going to capture more value.

• So what I would basically reframe the problem back to a founder as how do you make the value proposition more compelling? If the value proposition is more compelling, you're going to capture more value. So basically, you typically capture between 10 and 30% of the value created across all software companies. Somewhere between 10 and 30%. So make the amount you capture a really large amount, and then 10% of that can be a big number that can be very successful. Same thing is true of sales cycles.

As your value proposition becomes more succinct and more powerful, sales velocity increases. Sales cycles that are too long almost always have a poor value proposition or do not have large customers

• As your value proposition becomes more succinct and more powerful, sales velocity increases. Sales cycles that are too long almost always have a poor value proposition or do not have large customers. Large customers can buy something quickly. If you can solve, well, if you can make Walmart's revenue grow by 10%, they will buy your product really fast. They are very capable of buying it in 30 days or less. They just don't usually believe that you're going to move one of the top three business needles for them. Because it's actually quite difficult to move a top three metric for Walmart.

Pivot should involve some synergy between what you’ver learned before, what you’re doing, and where you’re going.

• Pivot should also, for the most part, involve some synergy between what you've learned before, what you're doing, and where you're going. Like, in basketball terminology, you kind of have to keep one foot down. So if you take both feet off the ground, it's not pivoting on anything. So if there's no connection to what you've been doing, you may be better off restarting the company in some other way."

I’m usually more the "The world should be different, and I'm gonna will it into existence" kind of top-down founder. Like, "I just don't like X? Yeah, I'm gonna fix it."

• I'm usually more the "The world should be different, and I'm gonna will it into existence" kind of top-down founder. Like, "I just don't like X? Yeah, I'm gonna fix it." For example, a group of founders, whom I haven't funded quite yet, but hopefully will, are working on a climate change problem. They're just ideologically convinced that they can fix climate change. Great! Tell me what the first product is, and I'll give you some good ideas. I was like, "Great, here's the money." You know, like, "Do you want the money?"

I feel like the co-founders we fund probably have roughly a 50% divorce rate. Startups by definition are going to go through, hopefully, very rapid change. And so predicting what this person or people are going to be like through periods of rapid evolution may or may not be possible.

• I've found that the classic strategy of dating leads to roughly a 50% divorce rate in the US. And at least I feel like the co-founders we fund probably have roughly a 50% divorce rate. So maybe there's a better approach. Part of it is that working together is challenging. In a startup, it's a pretty tough crucible. Things change quickly. Like a relationship, actually, as far as I can tell, it's a lot easier to have a stable relationship when things aren't changing. And it's very difficult on a relationship when one person or the other person goes through a massive transformation in their own lives. So the more predictable environment I think is better for human decision-making. Startups by definition are going to go through, hopefully, very rapid change. And so predicting what this person or people are going to be like through periods of rapid evolution may or may not be possible.

Typically, the way I think about founders is through a different prism: what are the three core risks to your company? What I'd like to see is somebody who's excellent, like a world-class expert, at each of the risks on the founding team.

• So, typically, the way I think about founders is through a different prism: what are the three core risks to your company? Every company has different risks. For example, if you want to compete with Elon Musk and build rockets, it's very different than if you want to be a competitor that is both bold and cute. What I'd like to see is somebody who's excellent, like a world-class expert, at each of the risks on the founding team. So, go to whiteboard and list out the core risks in your company. Then, ideally, plot your current team against those risks and make sure you have world-class talent against each of the core problems. If not, go get the person.

The ability to asymmetrically price a home sight unseen is the core of what we do and without a data scientist, that doesn't make much sense. So I found Ian and introduced him to Eric and JD.

So, for example, when we were founding Opendoor, we obviously needed a world-class data scientist. The ability to asymmetrically price a home sight unseen, is the core of what we do. Without a data scientist, that doesn't make much sense. Eric doesn't know much about data science, but he knows a lot about real estate. And I know a decent amount about data science, but not enough to build a team or really run a team, except at a very high level. Having a good world-class data scientist in my back pocket that neither Eric nor JD was essential. I was like, "Uh huh, I got just the guy for you. You should have coffee with this guy named Ian, and then get your opinion here." I was like, "Hey, what are you doing tomorrow?" And I was like, "Hey, I want you to get together."

Affirm’s first risk was underwriting people who were mispriced by FICO scores better than FICO. The second risk is fraud risk when someone is intentionally abusing with like a fraudulent identity. The core team had both the underwriting and fraud experience. And it would have been disastrous to do the company without both.

• So, back in the early days, Affirm faced a couple of key risks. Really, they mostly revolved around underwriting. The idea was that Affirm would effectively get financing to people who were mispriced by FICO scores, or had very limited FICO scores. And that's a big market. So Affirm had to underwrite better than FICO. That's a question of: can you get your hands on the right data? Can you model it correctly? That's one risk.
• The second risk, a little more subtle, is fraud risk. Which is different than underwriting risk. Underwriting is basically: the person wants to pay you, they just don't have the money. Fraud is someone's intentionally abusing the fact that you can get this product online now, and then they'll figure out how to walk away from it with like a fraudulent identity or something.
• The core team had both the underwriting and fraud experience. And it would have been disastrous to do the company without both. Because even if you got the underwriting right, you would lose too much money on the fraud side.
• Max Levchin is pretty good himself. And it certainly doesn't have to be dangerous. But his original co-founder, this guy Nathan Gettings, who'd run risk and fraud and Pay Pal with me back in the day, and then also co-founded Palantir. So he is probably the single best person on the planet to build the beginning of this that made it immediately fundable.

One of the first 10 employees at Stripe was their general counsel, which is pretty rare. But they needed someone who was a world-class general counsel to streamline the product, navigate partner dependency and regulatory dependency.

• Stripe is an example of this, and it's kind of in the public domain. One of the first 10 employees at Stripe was their general counsel, which is pretty rare. But they needed someone who was a world-class general counsel to streamline the product, navigate partner dependency and regulatory dependency. And had they not found this guy named John, an excellent lawyer, they would have been in deep trouble very fast, and maybe never gotten out of it.
• So not every founder knows how to find a general counsel. Sure, they may know how to assess one, but they may or may not know how to close one. So one thing an investor can do is help you with that. For example, I have a network of general counsels. I certainly know how to assess general counsels and can help close general counsels.
• Same thing with CFOs. CFOs are usually not that strategic early on. But if you're a company like Opendoor, or a firm that uses debt as oxygen, so you use debt to buy properties or to loan money to consumers, you need someone who understands how to do capital markets pretty early, or you just won't be able to grow because basically, you're turning this debt into your source of growth.
• So depending on what the company does, you may need different skills. But a lot of engineer-type founders wouldn't know how to go find a capital markets person or a CFO-type person. You just don't live in the same social networks.

If we write a check, we really care if there's one person who's the CEO that we're backing. We don't vote to fire CEOs and we can talk about that at length.

• Someone needs to assume the mantle of responsibility and be able to handle all that is involved. That's the most important thing. If we write a check, certainly Founders Fund for our brand, we really care if there's one person who's the CEO that we're backing. We don't vote to fire CEOs and we can talk about that at length. But fundamentally, if we don't have conviction around the CEO, we're not going to invest. But how the CEO decides to edit the team over time, that's really the CEO's prerogative. We have feedback. There's a little bit like a baseball metaphor here. You have a starting pitcher who gets called in the first six innings. Typically, for the last 50 years in baseball, other relief pitchers take over the game. And the art is partially knowing when you sort of need the relief pitcher.

Peter was really from the adamant that it had to be differentiated enough that it immediately cut through the clutter. It shouldn't be boring. And he said be so different, that you have a shot of making it work.

I think the most important thing in co-founder selection, other than marrying somebody who can help you solve a core risk, is agreeing on first principles. You cannot be debating first principles all the time in a startup like using open vs. closed systems or disagreeing on the value of various people in company.

• I think the most important thing in co-founder selection, other than marrying somebody who can help you solve a core risk, is agreeing on first principles. Like, always work with someone who agrees on first principles. You cannot be debating first principles all the time in a startup. It is incredibly destructive.
• For example, sometimes people don't understand what I mean by first principles. Let me give you an example. If you happen to be like a closed ecosystem person, think like Apple iOS, you do not want to work with a co-founder who believes in open systems and Android-like products. You're just going to fight over every single decision all the time, and the company is going to get absolutely nothing done while you're fighting. So you need to align on some core principles like that. And it's fine either way. Obviously, you can build open-stack companies that run okay, and you can go closed up, but you want to work with people who generally 90% of the time, we're going to be aligned on the principles, and then it's a question of applying the principles to the specific facts. Same thing with people. This one is harder to test upfront. You cannot build a company where the two most important people in the company are fundamentally disagreeing about the value of various other people in the company. You can have shades of difference, like I can grade you an A-, and this other person would say you're a B+. But it doesn't work if the other person thinks you're a C, and I'm like, he's an A. That just is incredibly painful. When you see that, and most people in the company, whether the founders or executives disagree. And so that is very difficult to figure out in advance, if you haven't worked with someone before. So that is a source of, you know, sort of unavoidable friction. But when it works, and when you generally agree on how people are performing their job, they can be extremely healthy.

Every founder has great strengths, especially that list of all world-class founders, and some of them in multiple dimensions at the same time but everyone needs a compliment. One of the things I've been able to do is pair with people and compliment them.

• Every founder is a little different. They have great strengths, especially that list of all world-class founders, and some of them in multiple dimensions at the same time. But does anyone need a compliment? Yes. Everyone needs a compliment. It's like basketball. So if you have a great point guard, which we're looking for, it's a little different than if you start with a center and say you're trying to get from there. One of the things I've been able to do is pair with people and compliment them. And the reason why is typically I don't have my own vision. In other words, I've been able to convince very strong-willed and opinionated people that a fact of life is I can implement their vision, versus I try to change it.

People who are experts tend to know what you can't do very well. They've mastered the rules. They typically don't ask enough "why" questions. Why not? Why can't this be done this way? I'd like people who kind of don't know what they're tackling a bit. But they're very fast learners.

• Typically, I don't like domain expertise. You know, we had this talking point at PayPal: out of 254 people in the Bay Area, only three had any financial services expertise. That's it. To me, that's about the right ratio. Square is roughly the same. Even including me, and I definitely wasn't an expert. People who are experts tend to know what you can't do very well. They've mastered the rules. They typically don't ask enough "why" questions. Why not? Why can't this be done this way?
• So I'd like people who kind of don't know what they're tackling a bit. But they're very fast learners. They find people who have the history and experience, and they can extract the key information out of them. So they're not blind, they're just very good at figuring out what they need to learn really fast.