Jake Saper, GP @ Emergence Capital: "We Sold Salesforce Early and Lost Out on Billions"

At Emergence Capital, we’ve returned a little over $8 billion in cash. We did this analysis on how our deals have fared in terms of certain graduation metrics relative to the market: 9 out of 10 of our deals have gone on to raise successful follow-on rounds, 1 out of 5 have gone on to raise rounds north of $1 billion, and 1 out of 10 of our early-stage investments have gone public.

Jake, I am so excited for this, dude. I loved our walk around London last night. I actually walked around and thought, “You know what? James Corden does Carpool Karaoke. I should almost try and do this around London.” It was great.

This is a PSA for anyone who visits London: you give the most enchanting walking tours of this place. Some might say romantic—just an incredibly beautiful appreciation for the history of this thing. Thank you.

That was when I took you down the prettiest street, and I was just like, “This is getting a little bit much.” But we were able to talk about my wife in that moment, which made it feel better.

1. The Zoom Investment Story

Exactly. I loved it because I also got so much context that I wouldn’t normally get, even in prep calls, which I think are generally shit, to be honest. But I wanted to start with Zoom. This was your first deal at Emergence, and I wanted to start there: talk to me about Zoom. How did it come to be?

We aspired to be a thesis-driven firm. Before I even joined Emergence in 2014, in 2013, the firm had developed a thesis around the fact that there was an opportunity to replace WebEx. WebEx was a tired product that wasn’t very good.

In fact, when I was interviewing at Emergence, the case they gave me was for a company likely called Fuze, an early competitor to Zoom. I was supposed to diligence that case and make a recommendation on whether we should invest. They were going to hire me based upon that. I did a bunch of work and ultimately concluded that we shouldn’t invest. Fortunately, I made the right call; that was also the decision they made, and they hired me.

A few months later, I joined the firm, and the very first deal we were pursuing where I was tapped to lead diligence was Zoom. The good news was that we had a prepared mind around the space. We also saw incredible early product-led growth. The company was around $2 million to $3 million in revenue and was growing very quickly, but obviously it was very, very early.

We believed in Eric (likely Yuan). Eric was the VP of engineering at WebEx, so he knew a lot about the space. He had rebuilt the core technology, called the codec, and it worked really well. My partner Santi, who ultimately led the deal, is from Argentina, and he used the product to call his parents—his family—back in Argentina. He discovered, “Hey, this thing works way better than everything else in the market. This codec is real. We should take this really seriously.” We paired that with the growth and thought, “Let’s dive in.”

However, the deal was not straightforward. The reason it was not straightforward is that it was going to be the largest check we’d ever written, at one of the highest prices we’d ever paid, and the company was incredibly early. For context, it was going to be a $20 million check out of a $250 million fund, which is a lot of concentration.

That’s a lot of concentration. For people, that’s 7% or 8%.

It was huge. It was going to be priced at a $200 million post-money valuation. Remember, the company was at around $2 million in revenue.

Revenue in 2014, right? This wasn’t 2021 peaks. This was pre-AI and all the rest of it.

But we really believed in the thesis. We believed in Eric, and the product-led growth was really strong. Even though it was early, it was super strong.

I was tapped to lead diligence, but the numbers were really going to matter because of the scale of this decision. I flew out to Boston because I was visiting my then-girlfriend, Danny Herzberg, who was an employee at HubSpot at the time. I was sitting in the HubSpot cafeteria, frantically trying to build the financial model for this company to justify the investment.

Brian (likely Halligan), who was then the CEO of HubSpot, walked over and gave me a hard time the first time I met him. That added to the stress of the whole experience. I didn’t have a background as a banker; I was an operator and a consultant before, so I didn’t really know much about how to build a financial model.

I was stressed out trying to build this thing, trying to make the numbers tie, and some numbers just wouldn’t tie—specifically, the churn numbers. Eric had given us a bunch of churn data on his business, and I was trying to figure out how that ticked and tied with the numbers I was calculating. It was just off.

Fortunately, the way our firm operates is fairly collaborative. I reached out to my teammate Joe, who has a private-equity background and was a Math Olympiad kid, and I was like, “Help me.” Joe dove in and figured out that the numbers Eric had given us around churn for the business were wrong.

Eric himself was miscalculating churn. Specifically, he was counting upgrades from a low tier to a high tier as churn, and he was counting pauses as churn. He was unnecessarily burdening the business with churn. Eric thought the business was worse than it was.

This is the only time in my venture career where that has been the case—when you dive in and discover that the founder actually doesn’t know how good his business is. We went to Eric and said, “Look, we found this. The business is actually better than you think.”

This was an interesting choice because we did it before we had finalized negotiations, so we gave him more leverage. But we thought it was the high-integrity thing to do. Eric realized that and said, “I want to work with you.” So we signed the deal. We were the first institutional investor.

Life-changing. Was it very competitive?

It was competitive. It was competitive both because there was early growth and because the company was profitable, so we didn’t have to raise capital. That’s part of the reason why the price got as high as it did.

Ultimately, I think the reason why he chose us is that we committed to helping him build out a proper enterprise-sales motion on top of his product-led-growth motion. I think that’s a really important point: as exciting as bottoms-up PLG is, all PLG companies eventually need to layer on enterprise-software motions to be sustainable and enduring.

Eric was great technologically, but he hadn’t had to build that kind of motion before. We committed to helping him do it. We hired a guy named Dave (likely Berman) from RingCentral after we made the investment, and we built out a proper enterprise-sales motion. The company was off to the races.

There are so many things I have to unpack off the back of that. The first is that you mentioned Eric’s background as VP of engineering—a phenomenal technical mind. When we actually look at that, I often think: what is better? Is it someone with intense domain experience, like Eric, or is it a 22-year-old founder who’s building with a naive but fresh mind?

The honest truth is that both work, and that’s what makes our job so hard. We’ve made a lot of money on domain experts. Peter Gassner from Veeva was also a domain expert when he made his money. But we’re also doing pretty well with some of these early-stage, smart young people, whether it’s Eric at Bolt or Austin at Unify.

There’s some naivety that comes into it, which helps them break rules. On another point you mentioned, Eric really chose us because we helped with enterprise go-to-market. Honestly, I don’t want a founder to need me.

I often think about [likely Keith Rabois]. The best founders don’t need you. They’re made better, but they don’t need you. How do you feel about that?

I think the reason why a founder chooses a VC is because they believe you’ll help them bend the odds of success on the journey. That can be in all sorts of ways. It could be helping them with something like go-to-market, bringing them people to hire, giving them advice, or being a great therapist to the founder.

There are lots of ways where you can help bend those odds, but I think ultimately that’s why someone chooses you. You need a reason—you need a reason to be chosen. I agree that, in general, if the founder doesn’t need to rely on you a lot, then that’s great. But the reality is that the odds of success for these things are so low that if you can even bend the odds of success incrementally, it matters a lot.

Was Eric good at sales?

Eric was not an experienced salesperson when I made the investment.

Wow. And you’re a politician. We’re learning a lot about each other.

2. Founder, Market, Traction: Rank Them

Another thing you said was about having a prepared mind around the space. I honestly think that’s kind of shit from large, multistage firms that like to present these packages. They can actually lead you to the wrong conclusion. How do you feel about the benefits of firms having prepared minds versus just packaging up a venture product?

The reality is that I don’t think any venture firm makes all their money on prepared minds. I just think that’s bullshit. I do think, because I’ve lived it, that you can have a prepared mind and it can help you get there sooner and faster than others.

In the Zoom case, it’s a great example. They literally gave you your interview case on a competitor. You must have just been like, “Boom, I’ve done this before.”

When we got the real thing, I was like, “Oh, great. I’ve already studied for this.” They basically gave me the test before the test. So it is true that we’ve had prepared minds, but sometimes it’s not always obvious.

In the case of Chorus, for example, Chorus.ai was an early competitor to Gong. It was acquired by ZoomInfo for a little less than $500 million. We led the seed round; we were the early investors there.

It was sort of the first voice-AI company. We had some prepared mind around the idea that voice would matter. My partner Gordon really thought voice would matter in the enterprise, but we didn’t have a full conception of what was going on. We just liked the young founder and thought there might be something there.

We made the investment, and ultimately it grew nicely and was acquired a little early, but it was acquired for a nice outcome. That helped inform a broader thesis that we’ve since developed around voice AI. We made a bunch of investments in Bland, Regal, and Assemble, as well as a bunch of vertical-specific ones.

I personally think we more broadly develop prepared minds from the portfolio-company investments we make. What is it about what’s happening at Chorus that informs a broader thesis beyond the sales-specific use case that Chorus was focused on, and that could help inform how we invest more broadly?

I think that’s real because I’ve lived it.

Can I be honest? Do you make much money from Chorus? You do the seed, it gets bought for half a billion, which is great, but I’m like, is that in stock or in cash? So many of these deals actually, behind the scenes, you kind of make 3x, and it’s not as good as it looks.

We made—I forgot the specific multiple. It was definitely more than 3x in that one. I don’t remember the specific number, but I think it was north of 5x.

The challenge is that when you have a fund of a certain size, even if it’s a 10x, it’s still not necessarily going to move the needle for that fund. I believe that fund was Fund III, which is the same fund that has Zoom. It’s the same fund that has some other really large outcomes.

That fund is already at—I think it’s a 16x DPI. So even with a 10x on the Chorus investment, it’s not necessarily going to move the needle.

3. Lessons from the 16x DPI Zoom Fund

It’s unfair, because it’s kind of in the details, but how much of the 16x DPI is from Zoom and the other major outcomes?

By far the most of it is from Zoom. What’s interesting is that the fund also had Salesloft, which was the biggest outcome. It was the highest multiple ever paid by private equity for a software company. The company was acquired for $2.3 billion. [likely Vista] paid $2.3 billion for it, and we were the Series A investors there.

That was a very good outcome, but that one returned the fund a few times. I can’t remember how many times it returned it, but Zoom returned it more than 10x.

Wow. Fuck that. I mean, okay. I’m trying to understand, in the power-law dynamics, just how much of fund returns are in those.

The cool thing about that fund, though, is that even if you took Zoom out, it would still be a top-tier fund because of Salesloft, Chorus, and a bunch of others. There are still companies that are just dating—companies like DroneDeploy and others in that fund.

We also just had a big crypto outcome in that fund called Xapo that distributed. I think there’s a large, multibillion-dollar distribution coming.

Can I ask? You mentioned Chorus, and there’s Outreach and there’s also Gong. When I enter markets, I often like to think about distribution dynamics within the market. Is it winner-take-all, or is it relatively evenly distributed?

How does this market play out in terms of Uber and Lyft, or actually multiple providers all distributed evenly? Do you think about that? I don’t want to be in a market where everyone gets a little bit, like Chorus, Gong, and Outreach. No offense, but I don’t think it’s a great market.

I think you can make money both ways. Let’s take the winner-take-all market. Network-effect businesses generally obviously win-take-all.

In our experience, we invested in a company called Doximity, which started as LinkedIn for doctors. Are you familiar with this one?

Yeah, it’s a fucking beast, and no one knows about it.

It’s crazy. We led the Series A there. My partner Kevin led that Series A, and Kevin’s still on the board there. That company is absolutely hitting, and it continues to hit in the public markets. Read up on it; it’s amazing.

That’s a winner-take-all business because it’s LinkedIn for doctors, and having a number two there doesn’t make any sense. We definitely underwrote that investment to that dynamic, and it worked.

But let’s take the Bolt-Lovable-like dynamic that we were talking about last night. For folks who aren’t aware, these are companies that help you build web apps with no coding experience. You can go to Bolt...

New, and you can write, “Build me a website for the new 20VC product, have it blue, put these buttons here, and make it do this, this, and this,” and then it just happens. It’s astounding how easily it happens, and it raises a lot of questions about the knock-on effects in our economy.

Those businesses will clearly disrupt Squarespace, Webflow, and so on. What’s fascinating about that space is that there are many very large incumbents, which supports the fact that there could be multiple large players in this next generation as well.

We’re jumping around, but I love a conversation like this. Help me understand what happens to those players. Do they consolidate? Do we have private equity providers come in? What happens to Webflow, Wix, Weebly, and Squarespace, just to name a few? Then there are the unbundled providers that are much more vertical-specific. What happens?

I think there will be consolidation, there will be private equity, and there will be a few that reinvent themselves. There will be a few that are nimble enough to figure out how to adapt to the new era, but I think most will get consolidated.

Do you think we overestimate private equity as a savior?

That’s a really good question. “Savior” is too strong of a word, because most private equity outcomes don’t generate incredible returns for the VC. Savior is not the right word. It may be a savior in the sense that it would get your money back.

The Salesloft outcome was an awesome one, but that was an exception, where private equity was willing to pay whatever—20 times ARR. In most cases, they’re paying 3 to 5 times, so it’s a totally different thing.

I totally get that. Thank God. I love that. I look at your Anaplans and your Coupas, and I think, “How the fuck are they going to get that money back?”

Again, some of these companies catch a second wave. We talked last night about the even smaller ones, the ones that are at $50M or $100M in ARR.

This is what really worries me: the ones that were funded in the last 5 to 7 years. What happens to those?

I think all those companies are in this existential moment where they have to figure out, “What do I do? How much do I invest in agents? How real is it going to be? Are my customers going to adapt to it? How complementary is that to my core product, or am I just shouting into the wind?”

The honest truth is that most of them are going to make it, unfortunately.

Do you think they’re structured for embracing the next wave of AI?

Some of them are. I work with one called Guru, which is a knowledge management software company.

Rick—I spoke to him before this about you.

Rick is God.

He said you’re a dick.

Rick is one of the most amazing human beings I’ve ever met.

Yeah, he looks like Jesus. He looks like a savior, actually, if you look at a picture of him.

We led Rick’s Series A probably in 2016, so I’ve been on the board there for almost a decade. It’s a knowledge management product that grew really quickly and then, during COVID, started to stall a bit.

To Rick’s credit, and to his executive team’s credit, they realized that as AI started to emerge, pairing a knowledge management tool with generative AI and building generative-AI-enabled search could be a very compelling offering. They could offer both knowledge management and an AI-enabled search product together.

They’ve done that, and now growth has reaccelerated. Through that journey, he did have to restructure the company. He cut the company down—I think we now have around 60 employees or something—but he’s doing that profitably and growing quickly again.

Can I be a dick here? Is that actually interesting from a venture-outcome perspective, in terms of the impact on funds?

It’s great, but Glean is at $100M in ARR, [likely Sana] is doing incredibly well and growing very quickly, and here we’ve got a company that’s rearchitected itself and pivoted into a market with players that are already growing fast.

What those other fast-growing players represent is market pull. The buyer wants this. The burden on a company like Guru is to figure out how to offer something that’s different from what those players have while still tapping into that market pull.

In Guru’s case, it’s knowledge management and AI enterprise search. In Glean’s case, it’s AI enterprise search only. There will be a bunch of people who just want that, and there’s a bunch of people who just want this.

That makes total sense. How quickly do you know your winners, going back to this, when you have them?

It’s not always obvious. There’s a lot of humility that’s important in this industry for lots of reasons, but that’s one of them.

Bill.com is an incredible software company that’s been around for a while. We were one of the earliest investors in that company as well, and that was not a straight line to the right company. It grew nicely before the financial crisis, and then the financial crisis happened and the business started to stall a bit.

We helped them figure out their channel-partnership strategy, partnering with banks. I think Bank of America was specifically the first one that unlocked really accelerated growth for that business. We figured out, “We can sell through these banks, where low-ACV products can be expensive to sell through traditional go-to-market channels. If we find a channel partner, the whole thing can work.”

That business absolutely took off and has been an amazing winner since then. It’s a great example of a company that wasn’t necessarily just this. It was kind of this, then this, then this. That is very possible, and that company made Fund I for us. That’s part of the reason why that fund is so good.

Another example that comes to mind around this humility point is a conversation I had in 2015 with a peer investor at another firm. I remember he came to my office and said, “I just made my career-defining investment.”

I said, “Please tell me what it is.”

He said, “Zenefits.” This was when Zenefits was on an absolute tear. I remember this. He poured a ton of money into it, and he was my peer—I think he was a principal or something at the time—so he was really putting his neck on the line. He said, “This is going to be it.”

To be fair, Parker was and is amazing. Parker’s a monster. There wasn’t a bad bet. Obviously, things happened to that company that made it an unsuccessful outcome.

We had just invested in Gusto, which was a competitor that was growing not quite as quickly as Zenefits. The reality is that Gusto has endured and become a massive company, and Zenefits didn’t. Parker went on to build his own business, which is doing quite well now, as we all know.

The broader point is that just because something is a breakout right after you invest, or early on, doesn’t guarantee that it’s going to win. The breakout can indicate market pull, which is the most important thing of all, but it doesn’t necessarily indicate an enduring company.

Can you talk to me about market pull being the most important thing of all for people listening—founders or investors? How do you think about that?

You want people to be desperate for your product. That’s something that’s so overlooked when someone is starting a company, particularly when they’re starting a company because they want to start a company, rather than because they’re trying to serve a specific need.

You want people who have tried desperately to solve this problem themselves. It’s not a desperate problem if your buyer hasn’t tried to hack together something on their own to solve it, if they haven’t bought an inferior product to solve it, or if they’re not spending countless hours themselves dealing with it. Otherwise, it’s a nice-to-have.

You need something where people say, “Oh my God, this is a massive problem. I need to solve it.”

How do you try to unpick that when you’re doing diligence on a B2B company?

When you talk to users or prospective users, the things I want to hear are phrases like, “If my boss stopped paying for this, I’d quit,” or, “If my boss stopped paying for this, I’d pay for it out of pocket.”

When you hear phrases like that, you know, “Holy shit, this has changed someone’s day-to-day life.” There’s real market pull for this.

I totally agree. I instantly think of a product likely called Linear, actually—the product management tool. It’s an amazing tool, but I always hear that with it.

How do you think about defensibility in a wave of AI? You could have market pull but also not be defensible.

This is where the founder comes in. If you were to ask me to rank founder versus market versus traction, hypothetically—

Yeah.

Market, founder, traction. That’s a good question. It’s a brilliant question.

The first one is market pull. That’s the most important thing when you’re evaluating a potential investment or starting a company. The second is the founder, because it’s the founder’s job to figure out how to build something defensible in that world of market pull.

It’s not enough to just build something that’s tapping into the zeitgeist. That’s particularly true with a lot of the voice-AI companies coming out right now. There’s a lot of market pull for those companies, but their job is to figure out how they can parlay that landing wedge into something more durable.

Different companies have different strategies for this. To go back to the Bolt example, the thesis we underwrote to was that Eric and his team had built a technology called Web container that allows you to host a web app dynamically in a really robust way. We thought that was going to provide some defensibility to the business.

It’s still very early, so we have no idea how that’s going to play out. But when you’re making an investment, you have to have a hypothesis about how the founder is going to build defensibility if they haven’t yet done it.

Can I ask when you thought there was market pull that actually wasn’t?

I’ve got a good one. We made an investment in a company that helped exercise instructors go out on their own and build their own business outside of their existing setup.

You’re laughing because I’m going to be honest: I could have told you that was a bad one.

I should reveal something here: my mother is a [likely Jazzercise] instructor. Do you know what that is?

I take it it’s jazz as a sport?

It’s kind of like jazz as a sport. It’s jazz and dance. It was a huge craze in the ’80s in the US: dancing and aerobics combined. I remember Miss Motivator. It’s kind of like that.

My mom has been teaching this since before I was born, and she’s still teaching it. The woman is almost 70. It’s amazing, so I have a soft spot in my heart for aerobics instructors in general.

The thesis was that during COVID, everyone was working out from home, and these people needed a business in a box to help them run their own show so they could do it on Zoom and so on. The reality is that there was market pull briefly for this product, and then when people went back to more normal life post-COVID, the gym teachers often went back to the gym. There was less market pull.

How much did you put in?

It was a Series A. I think we put in around $9M. We ended up getting most of the money back because the founder, to her credit, realized when the market pull declined and shut down the business and returned the money.

My question to you, on the back of that, is this: if I was your partner, I would say upside maximization. It may not be a bad business, but I don’t think it was ever a good business, so there would have been a hard block on that one. Jake, sorry.

Even if it works, the upside is highly questionable. Maybe it’s a $500M business. We say that while poo-pooing it, but it’s not at all bad—it’s amazing. But for a fund the size of Emergence, it’s not going to make a dent.

How can you justify to me a $5B business from that business?

If you think about this more broadly as a creator-economy tool, if you think there are going to be more Harry Stebbingses of the world who are 16, sitting in their rooms, figuring out, “I want to start my own business,” and you provide a business in a box for them, that’s potentially interesting.

They will not. There will be no more Harry Stebbingses. I will crush them. I’m worried for the world if there are more Harry Stebbingses.

If there are far more of them, we’re going to provide that business in a box for them. That was the thesis.

Remember, at the peak of COVID, you were making assumptions around how the economy would look going forward. The reality is that it’s hard to forecast. There was a world where people would be remote indefinitely, and we wouldn’t be back in more concentrated areas.

We made some bets that focused on that future, and we made some bets that focused more on an office future. That’s our job: to call the future.

I do want to go back to the element you mentioned about Zoom, which was freaking nuts. It was at 100 times revenue 10 years ago. I mean, 100 times revenue today is more normal. I still think it’s crazy, and a lot of people still think it’s crazy, but back then it was completely unheard of.

Have the best always been expensive?

They’re not always expensive, but they are often expensive. If I look back at our portfolio, Gusto was expensive, Zoom was expensive, Yammer was expensive, and a lot of the good ones—Ironclad was expensive.

But some of them weren’t. Veeva wasn’t expensive because that was non-consensus at the time. Salesloft was also non-consensus at the time and wasn’t expensive.

More recently, my partner Lyla led an investment in a company called [likely Federato]. It’s AI software to help insurers underwrite better. She made that investment before the zeitgeist, before people were saying, “This is obviously going to happen.”

To her credit, there were a lot of questions, and she pushed through and got that deal done at a pretty good price. Then the zeitgeist hit, and the company did a Series B at a much higher price.

I do think it’s still possible in this world to be non-consensus, be right, and get a good price. But it’s also true that there are increasingly more consensus deals, and you want to be in both.

Are the best always competitive?

No, I don’t think so. Veeva wasn’t competitive. There’s still a world where you have a unique insight that other people don’t believe in, or you get to the person first, where you can have a better deal.

But most of them are competitive.

You’ve mentioned many names. One was Gusto, and another was Veeva. Gusto is a brilliant business, but it’s raised a lot of money. Veeva, likewise, is a brilliant business, and it’s raised next to no money.

Do you think about dilution potential down the line before investing, and what that does to the outcome?

Yeah, we do. The framework we use internally to figure out whether we should do an investment is called “What You Have to Believe.” I was a consultant, so I really like frameworks.

The framework basically means you try to identify the 3 to 5 things specific to that deal that you have to believe for the investment to return the fund. If you unpack that, there are things like dilution—how much additional capital will they have to raise, and will the founder be able to raise that capital?

There are obviously questions around defensibility, market, competition, and team. All those questions depend on the company, so when we do the analysis, they’re always unique to the investment opportunity and to the fund we’re investing out of.

When we’re doing diligence, what we’re trying to do is identify the 3 to 5 “What You Have to Believe” points that are specific to the company.

Do you set those pre-diligence goals after you have the pitches? After you’ve spent some time with the founder and looked at the materials, you have some hypotheses as to what those could be?

Then, as you’re doing the diligence calls and doing the actual work, you’re refining them. Importantly, you’re gathering the data to support or negate each “What You Have to Believe.”

Every diligence exercise results in a chart. The chart is: What are the 3 to 5 “What You Have to Believe” points? What is the data supporting each one, and what is the data negating it?

Then we can all stare at this and say, “Hmm, on balance, do we believe it?”

Fascinating. I was debating whether to bring this in, but I actually want to, because you mentioned reference calls there. When we chatted last night, you said that everyone is on at least one reference call.

That’s supremely strange, bluntly. Normally, there’s an owner of a deal who does all the work and brings it back, and then you have a discussion as a partnership. How is that? Bluntly, it’s not a very efficient way to do it.

It’s super efficient. It’s important to understand the context of Emergence and how we operate.

We’re a focused firm, and we’re focused in 3 ways. The first is what we invest in. The second is how we invest, which relates to this question. The third is how we grow people.

On the “what” we invest in, all we do is B2B software. It’s all we’ve ever done, and it’s all we ever will do. Our first investment 20 years ago was Salesforce. Then the vertical SaaS thing happened and we did Veeva. Now the AI thing is happening, and we’ve done Together, Bolt, Bland, Unify, and a bunch of others.

We’re very focused thematically, and all of us do that work. We all focus on B2B, which means we’re uniquely able to invest collectively as a team and have everyone do the work.

On the “how we invest” side of things, every partner makes, on average, 1 investment per year. We are super focused in terms of the amount of investing we do, which is obviously very high risk, but it’s high conviction. If it wins and returns the firm many times over, which we’ve been lucky to do with a number of funds, that approach allows us to work collaboratively as a team when we’re only doing a relatively small number of deals.

The reality is that this has largely worked, but you’ve got to be really good pickers. One a year—I mean, no offense, I would not like to be you.

It’s a hard job. It’s a hard job, and it’s an average. Sometimes it’s 2, sometimes it’s 0.

Does it ever set the bar too high?

Potentially. The good news is that we’ve expanded our partnership, so we now have 7 partners. We have more shots on goal. When you think about it less as me as an individual and more as a firm, we’re taking a few more shots, which helps us with the averages.

You’ve never lost a deal, which we chatted about last night, and I would say that means you’re not taking enough risk.

No, what you mean is that none of my deals have ever gone to zero.

That’s right. None of your deals have gone to zero.

That’s correct. It does mean I haven’t taken enough risk, and that’s something I talk about with my partners a lot. Some of them may still go to zero, but you’re right.

I think part of it is that we’re investing in B2B software businesses that have recurring-revenue models. In general, there’s relative downside protection, given the entry point and the way we invest, assuming I’ve done the work and we’ve done the diligence well.

4. Why Does Every Partner Do Reference Calls on Every Deal?

I also genuinely believe this, although some other VCs will be more skeptical: our job, by making few investments and having everyone involved in the investment, means all of us help after we make the investment. I’m not the world’s worst investor, because I like to take everyone off on a different tangent.

Go back to why everyone is involved: they’re all in on the diligence process.

We just raised our new fund, and as part of that we analyzed how our deals have fared in terms of certain graduation metrics relative to the market. Nine out of 10 of our deals—our early-stage, seed and Series A investments—have gone on to raise successful follow-on rounds.

One out of 5 have gone on to raise rounds at north of $1B. One out of 10 of our early-stage investments have gone public.

We’re good at picking, and I also think we’re good at bending the odds of success. Part of it comes back to this model: we do the work together as a team.

The way we do the work together is that we have the founder come in and present to our partnership relatively early in the process, so all the partners get to know the person and the founder. Then we have a discussion afterward: is this something we’re excited enough about to make a priority deal?

Those are holy words at Emergence. When you say “priority deal,” everyone’s calendar gets blown up. Whatever you thought you were doing that week, you’re now doing something different.

It basically means everyone is focused on doing diligence for that deal. For every investment we make, every partner does reference calls, calls customers, calls management references, and does back-channel references.

Many partners also do on-site visits, where we go and spend time with a team and see what’s happening in the kitchen. We pick up on all the less-structured data points.

By having multiple partners make that trip, you’re collecting a bunch of first-party data. When it comes time to make the decision and you’re staring at that “What You Have to Believe” sheet, it’s not just 2 people in the firm who made it. Literally everyone has contributed to it.

One person can say, “I heard this customer say this,” and the next person can say, “Yes, but this customer sounded a little more wishy-washy in their voice.”

There’s this process of seeking truth, in contrast to most firms I’ve worked in, where it tends to be an associate and a partner doing a bunch of work and then defending their investment against an onslaught of questions and doubters. If you survive that onslaught, you get to do the deal.

Our process truly is a process of seeking truth collectively, and I think that allows us to pick better and, hopefully, once we make the investment, help the company better.

I think one of the biggest mistakes firms make is getting associates to do all the reference calls. I’m saying nothing against associates, but there’s so much in the tonality, the pause, and the facial expression. It’s a human business, completely.

One thing that really struck me there is time. We don’t always have time. Jake, I would love to do multiple on-site visits and do everything, but time compression is real in deals. How do you do an engaged and diligent process when you have real time compression and the end of the week to make your decision?

There are ways in which our process hurts us there, and there are ways in which it helps. It hurts because you have to coordinate lots and lots of schedules, which can be a pain.

It helps because we can do 7 diligence calls in the same slot. If you just have 1 person or 2 people doing diligence, then their calendar is booked out. But if I have 7 of my partners, plus principals and senior associates, all working on this, you can do an incredible amount of work in a day.

I’ve been amazed at how much work we’ve been able to do in a day when you have 7 or 12 people doing it, and you have 1 quarterback who’s the diligence lead. That was the role I played at Zoom, pulling it all together.

What it requires is trust. It means that if I bring in a founder and you meet the founder for an hour, but you don’t know anything else about the person or the deal, you have to trust me that there’s enough there for you to be willing to blow up your schedule for the week, even though it’s not your sponsored deal.

That culture of trust is what the founders build at Emergence, and I think it’s really special.

How do you do knowledge management across the partnership? What we do, for example, is record calls, so then I can listen to the call Jake had with a customer and feel like I’m in the room. How do you create that shared knowledge across the diligence process?

We record the calls, and we also send out really detailed notes from every conversation. Every night, we send out an email with a summary of what’s going on: “Here’s what we learned today. Jake did this call. Harry did this call. We talked to this customer. Here are the outstanding questions. Here’s what everyone needs to dive in on. We need help with this.”

It’s a constant stream of information bookended by these nightly emails.

The other thing we do is have a lot of calls at night. We realized that if we have these diligence calls, particularly the internal calls, during the day, they get compressed because we have 30 minutes. We’re just getting into the meat of it in minute 27, and then we have to go do something else.

If you do the call at night, after the kids go to bed, after you’ve had dinner with someone, theoretically you have an unlimited amount of time on the back end. The reality is that we do a lot of late-night calls discussing what we’ve learned and trying to synthesize the day.

You mentioned the on-site, the in-person. I love that. COVID meant that was impossible. Do you think the quality of investing went down dramatically during COVID?

That’s a good question. We still managed to do some on-sites during COVID.

I know—you broke the law.

I know. I’ll tell you a silly, crazy story.

There was a Series A in a company called Regal. It was very hot. I think it had 6 term sheets from top venture firms. The company had grown from $0 to $1.5M in a year, which, at the time, was very, very good. The founders were really credible. It was just a great company.

It was a consensus deal. We got to know the founder relatively late in the process, but we really clicked with both of them. We felt like we needed to spend time in person to really get there on both sides, but it was the height of COVID, so it was difficult to find a way to do it.

To make things worse, it wasn’t as if they were in San Francisco and I was in San Francisco, where we could go to a park and walk around. The CEO was in Steamboat Springs, Colorado, and I was in San Francisco. It wasn’t an easy place to get to.

What Alex Levin and I agreed to do was fly to Denver Airport, meet in a field outside the airport, and go for a long walk.

Was it raining?

It wasn’t raining. That would have been really cinematic.

The thing that made it worse was that, as you can see from how fair my skin is, I didn’t put on sunscreen. I hadn’t been outside in a year. I didn’t remember the sun.

We were in this field, and we did a 4-hour hike through literally a field outside the Denver Airport. I got completely, horribly red. It was terrible.

The good thing was that I really got to know Alex’s vision and understand him as a person, and I think vice versa. We made the investment.

The other thing that really helped, given that my partners weren’t able to take that same sunburn walk with me, was that Alex and Rebecca, his co-founder, did one-on-one Zoom calls with every one of my partners.

Do you ever get founders who say, “No, I’m sorry, we’re not going to do that. That’s too intense, and we don’t need the time”?

There are certain founders who aren’t looking for the product we sell. We sell a low-volume, high-touch product. There are other founders who are looking for a high-volume, low-touch product to get us out of their hair.

I think the nice thing about our diligence process is that it selects for the right founders.

Bluntly, I go back and forth on this. We have a similar product, not necessarily in terms of how we approach it. We have a product, and the founder can opt in or opt out.

5. Where Will Value Accrue in a World of AI?

The truth is, I don’t fucking want to miss Revolut, period. If they’re not that type of person, I don’t care. I want to be their person. You worry about that too?

I think there are situations where that can happen. The reality is that we’ve won a lot of deals doing this, even with those young B2B AI founders.

Austin Hughes at Unify is a great example. Austin came from Ramp, and his co-founder came from Scale AI. They’re prototypical young, hot AI founders. We led their founding round, co-led their founding round, and then led their Series A.

I think part of it is that Austin really believed in the product we were selling.

Why are you laughing? “Founding round,” whatever. What do you want to call it?

The first one—the inception, a pre-seed round. The reason I call it a founding round is that we agreed on terms before they had incorporated the company. For legal reasons, we couldn’t send them any materials until they incorporated the company.

Oh, so that’s a founding round. I don’t know.

No, totally. I’ve done founding rounds. Can I put “founder” on my—

Totally. That’s amazing. I love it when someone says, “I’m the founding designer in residence in London,” and I’m like, “What the fuck does that even mean?”

Yeah, I got a load of shit there when I put “social.”

You take away all our titles.

Do people like you?

Yeah, less so over time.

You mentioned $0 to $1.5M and it being, like, wow—100%. I was brought up in the SaaStr Jason Lemkin school of what great companies are: $0 to $10M in 2 years, I think it was.

Have we misled a generation of founders with triple-triple-double-double being great, when now that’s not enough with Bolt and Lovable?

The market has changed. Back to market pull: there’s more market pull in general right now than there was in the pre-LLM era. That’s true both because the LLMs themselves are doing amazing things and because everyone’s boss is saying, “Go buy AI.”

Many of these products have tapped into huge areas of market pull. But they still have to figure out defensibility on the back end, which we touched on before.

What that means is, yes, it is possible to grow faster. We’re seeing examples like Bolt, which went from $0 to $20M in 2 months. We’ve seen a bunch of others. Together AI is a good one. We were in the Series A 15 months ago, when they were at $2M in revenue. The business is now north of $100M in revenue.

These things can grow insanely quickly because there’s so much market pull.

They have shit marketing.

Exactly. It’s insane. It’s really good.

So, $2M to $100M in 15 months. That’s absolutely insane.

They really have terrible marketing.

You should tell them they have brilliant marketing. It’s so brilliantly articulated. That’s insane.

So what’s changed?

Market pull has changed. I think you’re right that if we tell founders the top decile is triple-triple-double-double, it’s just not true anymore.

I think it’s more like the great companies are quadrupling year over year. But the thing that we have yet to really see come home to roost—I don’t know if that’s the right way to say it—is retention.

We still don’t know, for many of these businesses, because they haven’t had 2 or 3 years of retention data, how that’s going to look. If I were to posit a replacement for the triple-triple-double-double phrase, maybe it’s something like “quadruple-120.”

What I mean by that is that, yes, you should be growing very quickly—perhaps quadrupling year over year, triple-quadrupling, and so on—but you should also have net dollar retention of 120% or above.

For those who don’t know, net dollar retention of 120%—can you explain that?

There are different ways to calculate it, but in general, the way to think about it is this: if I had $1 from a cohort of customers that I sold to last year, when they renew this year, they’re at $1.20.

The customers that churn from that cohort are outweighed by the customers that upsell. The net growth would be $0.20, or 20%.

Totally get that. So we want quadruple and NDR of 120%.

I think that if these companies prove to have net dollar retention of 120% or above and maintain this growth, they’re generational companies.

A couple of things there. One is margin. In a lot of cases, these are funnels for OpenAI and Anthropic. How do you think about margin improvement over time versus margin maintenance, given that they are funnels for LLMs today?

I don’t know if this is a commonly held belief, but in general, we’re not super concerned about the margins that OpenAI and the closed-source models are commanding, for 2 reasons.

One is that there’s already a lot of competition among the closed-source models, and you’ve already seen pricing decline a lot. Most of our application-layer companies that are providing applications on top of these products are seeing gross margin increase over time because of that competitive dynamic.

The second reason I’m not that concerned is that open-source LLMs are really, really good and getting better. The reality is that if you’re an application provider and, for whatever reason, OpenAI comes to you and says, “You know what? It’s 10 times the price,” and that eats into your gross margin, you now have a credible ability to spin up an open-source model and have almost no gross margin impact—basically, 100% gross margin.

That’s actually what Together AI…

And so part of the reason why Together AI has grown so quickly is that companies are saying, “You know what? We’ll spin this up ourselves. We’ll have complete security and data privacy, and we’ll control our own margins.”

You said earlier that you have to figure out what you have to believe for an investment to be a good investment. For Together AI, what did you have to believe?

The clearest thing you had to believe for that investment was that open-source LLMs would be a dominant part of the market over time; that enterprises and businesses were going to want to buy and use open-source models, and not just Anthropic, OpenAI, and the closed-source ecosystem.

We made the bet, and that was trending positively. I think it’s trended more positively, but it’s still frankly a little TBD for me. Obviously, it will happen, but it’s just a question of the extent to which it will happen. Is it an 80/20 market, or is it a 60/40 market? That might be the right way to put it.

You need it to be a dominant part of the market. It doesn’t necessarily mean it’s the majority, but if it’s 1% of the market and not 20% of the market, then the outcome looks totally different.

I would argue, as a partner of yours in this case, that even if it was 90/10—and I think most enterprises are not as intelligent and adventurous as we think, and will stick to the core providers—even if it’s 10% of the market, the world of companies is enormous. Ten percent is still really interesting.

It’s also a function of how good Together AI gets at helping companies spin this up. How painless will it be? If it becomes really painless, then that 10% could be the whole world.

You’re right. Do you prefer market creation or market expansion?

I borrow this from Mike Maples, our mutual friend. I read his book, Pattern Breakers, which I highly recommend. I actually wrote out a synthesis of it and presented it to my team by hand. I wrote it myself; I didn’t use AI to do it. That’s actually a learning from this: AI is amazing, but I retained more by doing the synthesis of the book myself than by asking an AI to do it.

One of his core insights is that you should look for a business that itself has a unique insight on an inflection that’s happening. An inflection could be a technological inflection. For example, open-source LLMs are a thing, so what is Together AI’s unique insight on how to deploy that? I won’t go into the specifics, but theirs is really about how you maximize inference within that context.

In Zoom’s case, obviously there was an increase in the use of video conferencing, and there was distribution of mobile. You could actually run the application on the computer itself or on the phone itself. Eric had a unique insight into how to distribute this stuff with a codec on the product.

It’s a long way of saying that, in both replacement markets, which Zoom was, and new markets, which Together AI is, you can find a situation where the founder is playing off some sort of inflection happening outside of their business and has a unique insight on how to take advantage of it. I would use Maples’ framework, and I care a little bit less about whether it’s a replacement or a new market.

I use it on every call I have, which is, “What do you believe that the world around you doesn’t agree with?” It’s a similar way of getting to the unique insight, so I totally agree with you there.

You mentioned that the thing we haven’t really figured out is retention, and a lot of these companies just don’t have the data yet. To be fair to them, they’re too young. Do you think the retention cohorts will be worse, better, or as expected on average?

I think they’ll disappoint, and I think there will be outliers that are better than we expect.

What will drive the outliers?

I think there will be businesses that find some sticky wedge that enables them to endure. This is why we did Ironclad: it’s deeply entrenched in in-house lawyers’ workflows and is very core to workflows across teams.

This is interesting, and it’s why adding AI into SaaS has differences. There are a lot of learnings from the previous SaaS eras that we have to take into this next era. One of them is that workflow is sticky. All those lawyers, the majority of whom will exist at least in the medium term, spend their day in this piece of software. It’s really hard to out-Salesforce Salesforce.

Salesforce isn’t the best CRM now, and I say that with a lot of love, as it was our first investment, and with love for Marc Benioff. The reason Salesforce is dominant is that tens of millions of people work in that thing on a weekly basis.

When you think about where there is sustaining value in a world of AI, how do you answer that question? Part of it is, if you can build something that people use every single day—if it becomes part of the way they go about their lives—I think that’s part of the reason why OpenAI is powerful. They’ve built a situation, a little bit more on the consumer side of things, where you build the muscle memory to open that app instead of Google when you’re searching.

I totally agree with you. I think brand is the Trojan horse that everyone is forgetting. Everyone has a consumer front end. Anthropic has a consumer front end, but how many people go to Claude? That’s part of Anthropic’s challenge going forward.

They’ve got an incredible coding machine. If I’m Anthropic now, I’m saying, “For fuck’s sake, just appreciate that you have Cursor, [likely Codeium], and an unbelievable coding machine. That’s a huge business.” Do you think we’ll see specialization of LLMs?

I can’t speak for Anthropic, and it seems like Dario has focused more on the long game of, “How do I do this AGI thing safely?” My guess is that his ambitions are focused there.

But I do think you’re going to see a lot of specialized LLMs, and I think that a lot of them will come from open source. Back to the earlier point, you’re going to see people who say, “I’m trying to solve a problem in the mortgage world, and I’m going to build on top of an open-source LLM a tool that helps me analyze and make recommendations on how to write the best mortgages in a very specific way.”

The cool thing about that—and this ties into a thesis we had back in 2017 that my partner Gordon started called coaching networks—was that it was a poorly branded but, I think, correct insight that the way AI would take place in business software is as a coach that shows up and says, “I see that you’re about to write this mortgage. Here’s all the data you should actually be using, and here are some suggestions on how to do it.”

It learns from what actually happens. You write the mortgage: does the person take it or not? Do they pay their loans or not? Based on those outcomes, it makes better recommendations to anyone else in that situation and in the network. We called that coaching networks.

The reality is that “copilot” is a term that took off, but what’s cool about that is that it’s domain-specific. If you build domain-specific large language models using that data, you’re going to have insights that even OpenAI won’t be able to have.

You mentioned copilot, that wonderfully hot word for usage in AI. We see people like [likely Clay] saying, “We’re replacing all of our SaaS tools and building them ourselves. AI allows us to build all of these tools ourselves.” People are genuinely asking the question, especially with vertical SaaS: are all of these tools dead? Will we be able to have very custom applications that we build ourselves?

I have a strong take on this, and I realize it’s a self-serving take, in that I’m an investor who invests in B2B software vendors. I obviously hope that B2B software vendors continue to exist in this world. I think it’s probably also helped you as well.

But I believe that B2B software—and I believe software vendors—have an important role in the future, even if the Bolts of the world, the Cursors of the world, make coding cheap, easy, and in some cases free. There are 3 reasons why I think that’s the case.

The first is that when you’re buying software from a vendor, you’re not just buying the code. You’re buying an opinionated perspective on how to solve a problem. That’s a really important point. Ultimately, if there’s a software vendor that has dedicated its life to figuring out the best way to solve a problem across a bunch of different use cases, they’re going to have a lot more insight on how to solve it, and they’re going to have that proprietary data, like I mentioned in the mortgage use case, that an open-source or a closed-source model is not going to have. You can’t just get that off the shelf.

That’s the first reason: you’re buying an opinionated perspective. The second is that the very factors making this software easier to build yourself also make it harder to maintain. You could spin up something in Bolt or Lovable or with Cursor yourself, and that thing becomes out of date in 6 months, or even faster.

Unless you have someone and some process to constantly keep it up to date, the software is immediately out of date. That’s often why enterprises start with build and then go back to buy when they realize, “We built it, but we can’t maintain it.”

The third and most important reason why I think software vendors still have a role in the future is that the buyer wants a throat to choke. Ultimately, when you buy something from a vendor, you’re getting a guarantee that the vendor will serve you well, that the software will not have downtime, that they’ll be there when you have questions around support, and perhaps that they’ll guarantee some outcome. This starts to move into a world where software pricing is evolving and could look more like outcomes-based pricing over time.

I want to get to pricing, but I think you forgot creation. Fundamentally, the majority of enterprises, especially in Europe, do not know what Slack is, let alone what Notion is. The fact that they’re going to create their own verticalized AI tools is absolutely fucking moronic. It’s really stupid.

Maintenance is another point. The average company has 172 tools. Are you seriously saying you’re going to maintain 172 tools? Absolutely moronic again. There’s also the accountability element: we need someone to blame.

That’s why we have consultants.

Sorry, former consultant, but we need to blame you. “It’s not my fault; McKinsey told me to do it.”

[likely McKinsey] told you to do it.

Fine. I totally agree with those points.

You mentioned pricing. Everyone’s saying that, on the back of this, we’re going to see a huge shift. I interview a lot of people. I interviewed [likely Anu from likely Lilac] the other day, and he said he didn’t know which company he would short, but he would short a company that has archaic pricing and doesn’t adjust on a pricing-model basis. What do you think is the future of B2B pricing in an AI-first world?

I think there’s a spectrum of pricing. You have the classic per-seat model, then you have usage models, which look like all sorts of things, and then you’ve got true outcomes-based pricing.

I think we’re in a world right now where most of the forward-leaning AI providers are experimenting with usage-based pricing. That could obviously be based on how many tokens you’re using, but it could also be based on something else. We work with a company called Assembled in the support AI space, and they charge based on how many interactions the support bot is having with your customers.

Then over time, you think that—

Sorry, I’m interrupting, but compare that to [likely Fin], where you actually do it on an outcomes basis. It’s resolution-based.

The direction this world moves over time is toward resolution-based pricing. I’ve spent a bunch of time learning about the Fin approach, and it’s hard for now. There are a few reasons why.

One is, back to the accountability part, that it’s hard to establish causality. Many support tickets, particularly higher-level support tickets, have multiple touches. A bot touches it, then maybe a human weighs in a little bit over here. How do you establish who was the winner?

You don’t want to create an antagonistic relationship with your buyer where you say, “I did all this,” and they say, “No, no, you only did some of this. I’m only going to pay you for this.” All of a sudden, instead of having an easy monthly bill, you’re negotiating every month with the customer. That sucks.

I think over time we’ll start to figure out some of those hiccups and bumps, but we’re still in early land on outcomes. It almost doesn’t work when there’s a human in the loop.

The easiest form of outcomes-based pricing today in AI is AI-enabled services. This is a business that takes on the whole delivery of a product. They don’t say, “I’m going to sell you an AI tool to help you do support.” They say, “I’m just going to do all your support. I’ll do the people; I’ll do everything else.”

We invest in a company called Mechanical Orchard that moves mainframes into the cloud using AI. They built a tool that’s basically Cursor for mainframes, but what they don’t do is sell that tool to Bank of America and say, “Use this tool to move all your stuff into the cloud.”

What they instead do is sell a service. They say, “We’re going to use this really cool AI tool we built, and we’re going to move your product into the cloud. It’ll take 50% as long, and we’ll charge you only 80% as much as the incumbent. If it doesn’t work, you don’t pay.”

That’s outcomes-based pricing. If you’re moving in that direction, it’s easier to establish outcomes-based pricing because there are no questions about whether you did it or not.

6. Three Reasons Why AI Will Not Replace Vertical SaaS

Do you have margin degradation on AI-enabled services, given the fact that you own the full vertical and have to ingest that risk yourself?

It depends on pricing. This is a really interesting question.

If you’re pricing on a labor basis, which is generally how most services are priced today, you’re taking the risk upfront in some ways. You’re saying, “If it’s going to take me this long, I’ll charge you this.” But if your AI doesn’t work, you could be in a world where your margins are really degraded upfront.

If the AI does work, then you actually capture way more margin. You have to be really thoughtful about how you price. You’re basically taking a bet on yourself: how good is my AI?

When we look at this distribution in an AI wave, who benefits most? Is it incumbents with incredible distribution advantages, or is it startups with none of the technical debt and the ability to move fast and integrate quickly?

There’s a third category, and I don’t really know how to describe it. We talked about it a bit earlier: these growth-stage companies, like the Notions of the world and the Ironclads—the companies that are above $100 million in ARR, growing nicely, and still dynamic and young enough to make changes, but not startups anymore.

I segment the world into those 3 buckets, which is an oversimplification. The biggest thing I’ve changed my mind around in the past 12 months relates to this question.

I was fearful when the power of LLMs came out that most of the value would accrue to incumbents because of their data and distribution advantages. What I underappreciated—which is just the recurring lesson of startups—is the value of focus.

It doesn’t matter how much distribution Salesforce has or how much data it has. If you’re a startup focused narrowly on solving a very specific problem, if you’re Unify, helping with the go-to-market stack in a much narrower way than Salesforce is, you’re going to run much, much faster, and customers are going to want your product more.

We’re seeing that play out. The thing I’ve changed my mind on is that I’m less fearful that incumbents will be able to accrue most of the value. It’s still early days in this game, and things could change, but thus far the focused startups are outpacing the incumbents.

With the growth companies, it’s more of a mixed bag for lots of reasons. There are companies that are a little more AI-focused and aren’t necessarily taking advantage of the stuff, and there are others that are still young and dynamic enough to pivot and take advantage of it.

I’ve been phenomenally impressed with the speed of incumbent shifts. When you look at the Adobes of the world, I think we always said, “They’re so slow; they’re so slow.” Actually, incumbents are shipping faster than ever. I’ve been very impressed by that.

How do you think about the “Google could just build this” argument? We’ve been investing for years, and everyone’s like, “Email autocomplete? Google could build that.” Whatever it is, how do you feel when you hear that?

This is where I go back to solving a narrow problem. Start by solving a narrow problem, because Google isn’t going to solve that narrow problem as well as you will, and you can expand from there.

Veeva is a great example. When we made the investment in Veeva, a CRM for pharmaceutical companies, the entire market for that was $400 million globally. That’s not big enough to build a multibillion-dollar business.

Veeva today is a $35 billion market-cap company. They obviously found a way to expand. They started narrowly, and then they became the board-level vendor to the world’s largest pharmaceutical companies.

That’s a really important phrase. I think most startups don’t think about how they can become so important to their customer that they’re discussed at the board level. If you achieve that, your ability to upsell is obviously much higher.

Veeva has now upsold all sorts of products to these massive pharmaceutical companies and has a $35 billion market cap as a result.

The same is true in the AI era. If you’re building an autocomplete tool in Gmail, if it’s a horizontal tool, you’re probably going to have your lunch eaten. But if it’s for a very specific use case and it works really well, then theoretically you earn the right to expand and sell that same buyer something else.

You can use it as a landing wedge and then expand to sell something that perhaps is more defensible.

I also find that you continuously underestimate how big the landing wedge is.

That’s right, particularly if you’re really good at solving a problem and the pain point is really big. If the pain point is really big, you don’t just have customer demand for it; people are willing to pay a lot of money for it.

The other thing that’s true is that, in the AI era, a lot of these businesses are able to capture some labor spend in addition to software spend. That narrow wedge, while it may be narrow from a software-spend perspective, may not be narrow from a total-spend perspective.

Sarah Tavel has written before about paying for the work, not just for the software. Do you buy that transition?

I think a lot of buyers will find it difficult in their minds to justify paying for labor when it is software.

What I’ve seen thus far—and it’s still early days—is that most buyers of this stuff aren’t firing people. What they’re doing is not hiring new people and trying to be more efficient with whatever they currently have.

I just invested in a voice AI company in healthcare, and we talked to a bunch of their customers. The customers said, “We love this thing. It’s amazing.” We asked, “How many headcount did you reduce?” They said, “None.”

I said, “Wait, why do you love this thing?” They said, “I love it because I’ve grown my business 3 times with the same headcount.”

I think right now—and part of it is emotional; people understandably don’t want to fire their employees—businesses are able to grow more efficiently than they were in the past because of this stuff. Therefore, these software vendors should be able to capture some of that labor spend.

We’ve mentioned Salesforce multiple times. In the next wave of AI, respectfully, everyone on the show has said that Salesforce would be one of their biggest shorts. It’s not mine. Why? What would the bull case be for Salesforce?

I think the incumbency advantage of Salesforce that we talked about before is very real. The fact that millions and millions of people use that product every single day cannot be underestimated for such core workflows.

That’s not the one I’d short. If I were to short a stock publicly, it would be IBM.

Why?

IBM still makes so much money from selling these refrigerator-sized mainframes. I think the thing people don’t realize is that 75% of the Fortune 500 still run their core applications on refrigerators in their closets, and those applications are written in a language called COBOL that no one writes anymore.

IBM makes billions of dollars every year selling maintenance and new servers to support these massive companies and their legacy code. That code has been trapped on these machines, and AI is, I believe, the critical enabler to get this spaghetti code into the cloud and hosted in a much more efficient manner.

If companies like Mechanical Orchard succeed in doing that, the IBMs of the world would be in trouble.

7. Why is Jake Worried About AI’s FTX Moment?

Given the dominance of these companies still running their software on those refrigerators, do you think we overestimate AI adoption in the short term?

My guess is that there’ll be a little bit of a trough of disillusionment, just like there always is in technology adoption. A lot of people are trying everything right now.

The good news is that there’s a lot of movement from experimental budgets into real budgets in these enterprises. The bad news is that a lot of these companies that aren’t actually delivering enduring value are going to get cut. There’ll be some buyers who say, “This thing didn’t work as well as I wanted,” so I’m a little disillusioned.

The other thing that could happen is there could be an FTX moment in B2B as these agents come out. What I mean by that is that these agents are incredibly powerful and do things for you: they send emails, buy things, and take action.

That’s very powerful, but with great power comes great responsibility. It’s very possible—in fact, likely—that some big enterprise is going to deploy an agent, and the agent is going to do something really bad. It might send a bunch of emails to customers or prospects that it shouldn’t, or buy a bunch of things that it shouldn’t.

It’s not hard to imagine what could happen. There could be a bit of a backlash where people say, “Wait, this isn’t good. We shouldn’t do it.” The reality is that we do need to figure out the guardrails for these products so they’re deployed safely.

You mentioned disillusionment. We’ve seen a huge amount of disillusioned talent within venture firms leaving.

What a transition. I’m pretty good at this. I’ve done a couple of these shows.

Impressive, honestly.

It’s like 3,000 in, and I can do a transition. I’m glad you noticed that one.

There’s a huge amount of disillusioned partners. We’ve seen it with new firms and with departures. They’re fundraising more than ever, and they’re hugely undifferentiated. They’re just starting another firm.

My question to you is: you’ve never lost a partner, which is nuts. What do you do that no one else has been able to do?

We grow them from within.

Why is that important?

Most venture firms hire 2 profiles when they’re looking to hire investors. They hire either seasoned, often ex-CEOs who presumably have a great network and a great brand, or they hire an army of junior people, give them a checkbook, and say, “You’ve got 2 years to prove to me that you’re good.”

The incentive that creates is for those people to write as many checks as they can. At the end of the 2 years, one of 2 things generally happens. It’s almost always never enough time to really see if these investments are good or bad, so the person often leaves.

They either leave because the firm says, “These investments aren’t trending, so you’re out,” or their investments are good and the person looks up and says, “You know what? If I build my career here, there’s no chance I’ll ever be an equal partner.”

That’s how most firms work. Most firms don’t have an equal partnership. They have founders who retain carry after they depart. As a result, if you’re really good and pour your entire career into something and make great investments, you don’t ever get to capture a portion that’s fair.

What you’re seeing is a merry-go-round in venture capital. You have a bunch of people who leave and go from firm to firm or start their own firms because of these dynamics. It’s really bad for our founders, because what happens is these founders become orphaned deals.

No one talks about this, and it’s a big deal. Why is it a big deal, and what should founders know?

Unfortunately, I’ve been part of a lot of boards now with orphaned deals, where the original investor who made the investment leaves the firm, and the company doesn’t have support within that firm.

When it comes time for a new round, the odds that you’re going to get that support are lower. All of a sudden, the founder has to scramble and figure out what to do. Or a new person comes onto the board who may be less constructive than the person chosen by the founder, and that could have negative implications.

There are so many ways in which founders can feel tricked, although that’s not quite the right word. They signed up for something that’s not what they bought. They’re getting something that’s not what they bought.

When you buy an investor, you’re really hoping to buy that firm, but also that person. These journeys last a decade. You’re hoping to sit across the table from that person for a decade or more, and you kind of know that. That’s why the vetting process is so important. That’s why I flew to get sunburned in Denver with Alex and others.

If there’s a firm where people are leaving, it strands these founders without the person they originally wanted.

Ten thousand percent, it’s signaling a real risk. Do you agree with the multistage funds that say it’s not really a signal and is just used by seed investors as an instrument to keep their jobs?

From a seed perspective or in general?

In general.

I think when multistage firms make seed investments and then don’t double down, it is absolutely a signal. It comes down to the topic we discussed before: focus.

Most multistage firms treat their seed programs as an option program. They write small checks, often through their junior people, and use it as a way to track the company and see if it breaks out. If it does, they try to pour in and get proper ownership.

That’s the way most seed programs are run in multistage firms, and it’s not great. It does provide signaling risk, and it also doesn’t help the founder as much because they don’t get much love from the firm.

The way we try to do it is the same way we try to do everything else: with focus. When we make a seed investment, we treat it like a core bet. It’s a partner doing the deal, and we’re generally owning double-digit percentages so that we really care.

It also allows the founder to raise a round from someone else if they need to, because we already have our ownership.

How ownership-sensitive are you?

It depends on the context. We try to be pretty ownership-sensitive.

Would you do a deal at 8%?

Sure. We did Zoom at 10%.

How do you differentiate between, “It’s a stretch, but we’ll do it,” and, “I’m sorry, that’s too low”?

It comes back to the “what you have to believe” framework. If you have to believe that the investment will return the fund, and ultimately it’s a low-ownership investment but we think this company is going to be Salesforce, won’t be super-dilutive going forward, doesn’t have to raise a ton of capital, and can return the fund, then we’ll do it.

This is a game of outliers. You have to convince yourself that this one is an outlier. You have to convince yourself that all of them are outliers, but the lower ownership you have, the more you have to believe that it can return the fund.

Do you buy that we have such a fixation around a venture investment returning the fund when, actually, part of me thinks, “A, if it returns half the fund, that is still very good, and B, we always underestimate the size of our winners”?

We’ve been around for 20 years. We’ve deployed a little less than $2 billion in capital, and we’ve returned a little over $8 billion in cash. That doesn’t include all the private holdings, which are obviously worth a lot more.

The reality is that the bulk of that $8 billion has come from a handful of companies. The outliers really have driven those returns.

What’s interesting, as I said before, is that even if you remove some of those outliers from some of our funds, they would still be top-decile funds because of the Choruses of the world. They get bought for $500 million, but you invested at the seed stage, so it’s still good money.

What was the best fund?

The best fund was our Zoom fund, which was Fund III.

Even over the Salesforce fund?

We sold Salesforce too early.

8. What Losing Billions on Salesforce Taught Us About Selling

That’s something else that doesn’t get talked about enough in venture: exit matters. I had this written down when you mentioned it earlier. Public selling—let’s talk about public selling.

When do you sell?

It’s a really hard question. We sell very early in some cases, shortly after a company goes public.

When did you sell Salesforce?

Very early, shortly after it went public.

That was bad.

I wasn’t there for that.

You were young.

Exactly. I was young. I was there for a lot of the more recent decisions: when do we sell BILL, when do we sell Zoom, when do we sell Veeva, and so on.

Do you have a formal framework?

We do. Every quarter that there are earnings, we do an analysis led by the sponsor. In almost all of these cases, we’re still on the company’s board, so we have inside information. That’s the reason we hold the position. If we didn’t have inside information, we’d have no defensible reason to hold it.

As long as we’re on the board, we have that information. The sponsor takes that inside information and updates the partnership, saying, “Here’s what’s going on.” Then we make a decision. Because we’re on the board, we can only make that decision within a specific window after the announcement, for legal reasons, obviously, about whether we hold or sell.

In some cases, we buy. In the case of Doximity, we were huge believers. We invested in the company very early, then actually bought more at the IPO, and that’s been a great investment. We’ve already returned 3x from where that was.

It’s a hard decision because every quarter you look at it and say, “I know what’s going on with the company, but I don’t know what’s going to happen in the macro.” There’s all sorts of uncertainty in the macro, so that weighs into it. You do your best.

Do you feel pressure from investors in less-liquid times when you could liquidate positions?

For sure. I think the nice thing is that the vast majority of our investors are big charities, charity foundations, and endowments. These folks aren’t as concerned with needing cash right now. They really want us to maximize the size of the outcome.

I’m very grateful for those LPs. We’ve returned, as I said, over $8 billion on $2 billion deployed to them, and they generally trust us.

We did an analysis recently on how good we’ve been at managing public companies.

What did that show?

The analysis was, first, what if we had sold all of our shares in all of our public companies at lockup, meaning once the lockup expired, we sold immediately? The second scenario was what we actually did: how much have we returned from those deals?

The last thing we looked at was the value if we had sold at the peak price of those stocks, which was often in 2021.

You’ve got a fourth option, which is just holding in perpetuity.

We could hold in perpetuity. We’ve done that to some degree. In the case of Veeva, we’ve distributed 90% of our position, but we still have a meaningful stake in it because we owned more than 30% early on.

Selling at the peak price is obviously going to be the optimal return.

For sure.

What’s the swing between what you’ve actually done and the peak price, versus what you’ve actually done and selling at lockup? If we had sold them all at lockup, you’d have massively fucked yourself.

Yeah.

You’d have killed yourself on Salesforce alone.

If we had sold all of our shares at lockup, we would have returned $2 billion less to our LPs.

Okay.

What’s interesting—and this is totally coincidental—is that if we had sold all of our shares at the peak price, whenever the peak price was for that stock, we would have made $2 billion more for our LPs.

No, but I mean this in the nicest way: that’s not true. Salesforce would have been another $200 million.

Yeah, yeah.

To be clear, we excluded Salesforce from the analysis because it was so long ago, and we made such a bad decision. The analysis is on the stocks we’re managing now, the stuff that’s currently public, including Zoom.

Wow.

Exactly. What we’re trying to say to our LPs is, “This is how good we currently are at doing this.” Over the past 5 years or so, we’ve been managing our public stocks, and the answer is that we made them $2 billion more than if we’d just given them all the money right now.

Do you agree with [likely Roelof Botha’s] thesis that venture managers with inside information are best placed to manage companies even in the public markets?

Obviously, to some degree, I do, because we’re still managing a lot of those positions.

If you believe that, by staying on the board and being super active, you have more insight, then you do have more insight. The downside, honestly, is just time, because you’re taking time away from new investments.

There’s a huge benefit to the firm in having that connectivity to those incredible companies. One other thing we’ve started to do—I mentioned our work, which has been pretty cool—is fund really early-stage AI companies, particularly within specific verticals, and try to pair them up with these giants.

If we pair a really big company with great distribution with a small company, they can do some sort of deal that may involve equity, where the big company gets to buy a little bit of the small company, but the small company gets the incredible distribution advantage of these massive companies that we’re already a part of.

There’s this beautiful symbiosis that has been playing out for both sides so far.

We mentioned signaling earlier. I don’t think we talk about reserve investments enough. How do you think about reserves, reserve allocations, and the decision-making attached to them?

It matters. Some of the secondary purchases, both actual secondaries and the second- and third-check investments we’ve made in some of these winners, have been hugely important from a returns perspective.

In retrospect, it’s always hard to know, because we’ve definitely made some really bad third-check investments in companies that we thought were trending but didn’t end up trending.

What’s the story of that, and what did you get wrong?

One of the big examples, when we look back at failures in our portfolio over time, has been when we thought there was product-market fit but there wasn’t.

The term I’m using now is “mirage product-market fit.” A company has grown really quickly, and you can fool yourself into thinking it has incredible product-market fit. But there are a couple of different downside cases where you think you have it and you don’t.

In traditional SaaS, one case is that you’re selling a product to a very diverse audience, and everyone is using it for different things. You think, “I’ve got product-market fit,” but the reality is that one customer is using your product for one thing, and another, completely different type of customer is using it for something else.

How do you figure out your go-to-market motion? How do you figure out your product-development motion when all these people want completely different things? That can cause companies to blow up.

There’s a second dynamic happening with these AI-enabled services companies. Let’s say you go out and say, “I’m going to be an AI-enabled accounting firm. I’m going to charge you 15% less than the incumbents, and I’m using AI, so it’s going to be even better—higher quality, faster, better, cheaper.”

Of course customers are going to buy that because it’s cheaper, faster, and better, so you’re going to grow really quickly. But that doesn’t tell you whether you’ve used AI to provide a high-margin service.

You can sell a lot of something, but if you haven’t figured out a good business model around it, then it’s not really product-market fit. This happens in consumer businesses as well, where people sell a dollar for $1.50 and are underwater from a gross-margin perspective.

The same thing can be true in AI-enabled services. All of that can lead to a situation where you, as an investor, or you, as a founder, think, “Oh my God, I’ve got product-market fit.” You pour in more cash, but you don’t.

You have unanimous decision-making on the initial check, which personally I think is strange, but it clearly works. On reserves, how does that look? Do you worry that there’s bias because the deal sponsor likes the founder, it’s their name on it, and they want to keep it alive?

In a lot of cases, yes. I think when it’s a “keep it alive” scenario, we’re a lot more thoughtful and will often get someone else involved.

Every reserve situation is different. If it’s a reserve for a pro rata on a Series B, where we did the A and the company is performing well, it’s a little more straightforward.

The hard thing comes into play when it’s a Series C and you need to put an inside round together to figure out what to do. In those cases, we actually have the founder come back in and present to our full partnership. Then we’re all up to speed on what’s going on, and we can check the founder.

You’re right: there’s a lot of emotional investment, and there is bias involved. But if the founder comes back to us and gives us the story, we’re able to poke and prod a little bit.

I remember Mike Maples, our mutual friend, saying on the show that 99% of the time, bridge rounds are a bridge to nowhere. Do you agree with that in your experience?

There’s a business that was bought by a British company called Sage, called Intacct. Intacct was the number-two cloud ERP player. We were early investors there.

ERP is a tough thing because it’s the most mission-critical system of all, so it’s really hard to rip out someone’s ERP. But once you get in, it’s really sticky, and you can stay in that business.

That business grew slowly, then had some cash problems, and we decided to bridge the company. That saved it. We figured out the business and also figured out a channel-partnership motion through accounting firms. The business took off and was bought by Sage, I think for $1 billion or something.

We made a ton of money on the whole thing, including on that bridge, because the bridge was obviously on favorable terms given the condition the company was in. That’s a cherry-picked example, but there are examples where it’s not a bridge to nowhere.

Have you ever been part of a pay-to-play?

I personally have not yet been part of a pay-to-play. My firm certainly has, but it sucks. It feels bad on all accounts. Sometimes it’s necessary to get the company funded.

When do you think IPOs will return?

I tend to think next year. I think there’s enough uncertainty in the market right now, particularly in the macro and political situations, that a lot of people are nervous. My best guess would be the beginning of next year.

9. Quickfire Round

I’ve peppered you with so many different questions. I do want to do a quick-fire round.

Let’s do quick fire.

Are you ready?

Hit me.

What do you believe that most people around you disbelieve?

You can teach anyone to sing.

Huh?

I’m a singer. I’m a classically trained singer, and part of the way I made money growing up was by giving voice lessons. In college I gave voice lessons, and even in graduate school.

I was taking a really hard electrical-engineering class. The teaching assistant spent a bunch of extra time with me to teach me electrical engineering because I was taking a master’s-level EE class with no background in it. I was doing a second degree in addition to my MBA, which was really hard.

He told me that he wanted to propose to his girlfriend and wanted to do it in song, but he had never sung before. I spent a bunch of time teaching him how to sing, and he did it. They’re married.

Wow. That’s a really romantic story. I like that.

I think I’m inspired by the walk. I wasn’t expecting that.

What can I say?

I’m here for romance and tour guides. I’ve actually got a tour-guide business on the side. It’s why we do the podcast: so entrepreneurial.

You can buy and hold 1 public stock for the next 10 years. Which one and why?

I think Microsoft. The reason is that I’m obviously long B2B software, and Microsoft is probably the best index for that.

If you believe they’ll continue their dominant position, then it’s an index on that growth. You’ll get a 30%, 40%, or 50% price increase, but you’re asking me to hold 1 stock. I’m putting my entire portfolio into something safer. You don’t want some super-risky GameStop shit.

It’s not my style. Jesus, Jake, come on.

You’ve got a seed fund, a Series A fund, and a growth fund. Which do you invest in for each? You can’t say Emergence.

I generally like stage-specific firms. On the late stage, I like the Meritech folks and the Greenoaks folks—anyone who is a student of that craft.

On Series A, similarly, I like the people who tend to be more thematic and focused. Mavon is great on the consumer side of things. I have a ton of respect for USV; they tend to be deeply focused on crypto and other areas.

On the seed front, it’s similar. Our friend Rick Zullo runs a fund called Equal Ventures, which is a seed fund that is super thesis-driven. I respect it. He makes contrarian bets super early.

What deal have you lost, and who did you lose to?

I think the first deal I lost was [likely Ironclad], and I lost it to Jess Lee at [likely Sequoia]. It was super painful.

The context is that I had gotten to know Jason, the CEO, before he started the company. He was at a coffee shop and met my wife because she was wearing, I guess, a HubSpot shirt. He was curious to learn about sales—that’s what they tell me.

They talked, and Danny, my wife, said, “This guy’s amazing. You should meet him.” I met him, and I really liked him. He started the company, but at the time we weren’t doing much seed investing, so I didn’t look at the seed round.

At the Series A, we had invested in a company called SimpleLegal, which was billing software for lawyers. On their roadmap, they had contract management. They hadn’t built it yet, but it was on their roadmap.

We made the investment, and then I called Jason and said, “Listen, man, I really love spending time with you, but I think I shouldn’t be part of the Series A because of this.” He agreed, so we didn’t participate in the Series A at all.

Then the Series B came very quickly. To Jess’s credit, she ran fast, got in front of him first, put in the term sheet, and won it. The happy story is that I ended up investing in the next round. I think I was the only external investor in that round.

What was the price of that?

Somewhere in the $300 millions, maybe $400 million. I don’t remember specifically.

The company is now north of $100 million in revenue.

It’s doing quite well.

If we actually pick a multiple on them, it’s like a $3–$4 billion company, if we want to be really generous.

I think it could be much larger. I’m obviously biased, but DocuSign is a comp. That business got very large and was literally just signatures. We now do that as well, in addition to contract management, which is a much bigger thing.

We also have an AI product we’ve built called Jurist, which is a Harvey competitor. It’s pretty cool and growing really quickly. I do think there’s real upside to it, but we’ll see.

What’s your biggest win, and what did you learn?

Obviously, being involved with Zoom very early was huge. The 2 learnings I had there were market pull—incredible market pull—and founder insight.

As I said before, if you can find founder insight around creating something differentiated, it’s a recipe for success.

What’s been your worst deal, and what have you learned thus far?

I haven’t had any zeros. I’m sure I will.

The worst exit I’ve had was a company called Comfy, which was building energy-efficiency software. It gave employees within an office the ability to change the lighting and temperature from their phones wherever they were. It would follow them around, remember their preferences, and change the building accordingly.

It was actually quite cool. The business grew really quickly from a bookings perspective. They had these big 7-figure contracts from Salesforce and others, but the people in charge of deploying the product didn’t care. There was a huge incentive issue between the buyer and the implementer in that business.

We had huge bookings, but we didn’t have great deployed ARR, and that bit us in the ass. We ended up selling the business to [likely Siemens]. We made a little bit of money on the deal, and Andrew, the CEO, and I stayed close.

He actually bought me a gift certificate to The French Laundry, which I still haven’t been able to use because the reservations are so hard to get. I’m grateful to him for helping navigate through the outcome. He ended up making a good amount of money.

Anthropic at $60 billion, [likely Grok] at $50 billion, and OpenAI at $300 billion: which do you buy, and which do you sell?

I sell [likely Grok]. I don’t yet know what niche they’ve carved out in the market.

OpenAI feels expensive, but they have a strong consumer brand. I probably buy Anthropic, assuming they can figure out the app stuff more. The underlying things happening with the model seem quite promising.

What’s the craziest thing you’ve done to win a deal?

Assembled, which I mentioned before, is AI for support teams. We led the Series A there.

It was a very consensus deal. The 3 founders came from Stripe, where they had built the tool to serve Stripe. They realized it was bigger than just one company and that they could spin it out and start a company. Stripe did the seed round; I think it was the very first deal Stripe did as a seed investment.

As you can imagine, the Series A was very frothy because everyone and their sister wanted to invest in these hot Stripe founders. I’d gotten to know the founders for a while and had tried to push Brian, the then-CEO, to let us invest. He wanted to run a process.

I got a text message from him, I think as I was coming back from my honeymoon. I was blissed out, and he said, “Hey man, the process is live. Do you have time?” I was not in a place where I was running after deals, and I was pissed because I’d been in front of that one for a while and was really excited about the company.

But I flew back, worked really hard, and my partner YZ, who was then an associate, did an insane amount of work to get us up to speed. We got into the top 3 of the bidding process, and then they went silent. I thought, “That’s not good.”

I got a phone call at 9:00 p.m. from the CEO. He said, “The good news is you’ve made it into the top 3. The way we’re going to decide this is with a mock board meeting. In 1 hour, I’m going to send you a bunch of materials as if this is a board meeting. Then we’re going to hold a board meeting—the 3 of us and you—and see how you perform.”

That’s exactly what happened. At 11:00 p.m., he sent me the materials. I prepared, and then we had a board meeting. He saw how I showed up as a board member and ultimately chose us.

I’m very passionate about board service and the right way to show up as a board member.

Do you know who you beat?

I think Index.

When you hear that they’re on a deal, do you think, “Oh fuck, I need to get my game on”?

The reality is that there isn’t a specific person or brand where I think, “Oh fuck, I’m fucked.” It’s much more deal-specific.

If the person has an existing relationship with the founder, that’s hard. If they’ve worked together for 10 years, that’s really hard. It should be hard in some ways because you want to choose someone you really trust and think can add value.

That’s part of why my job is to get to know people early. In our last fund, Fund V, we did this analysis and found that we knew the founders, on average, 13 months before we made the investment. We really try, because we know it matters.

Penultimate one. Very often, older partners hog carry pools. How does the carry distribution look in the partnership?

I’m so grateful for this. As I mentioned, we grow partners from within, and we think it’s one of the things that makes us different.

Part of the reason we’re able to retain incredible people like my partners Liy and Yas, and others who have come up behind me, as well as me, Joe, Santi, Kevin, and all the people who have been grown within the firm, is that our founders made the very generous choice to forfeit their carry when they stepped away from the business and retired.

This isn’t something that’s talked about in venture. I had no idea about it when I was considering which firm to join 11 years ago, but the vast majority of founders of firms retain a meaningful portion of the ownership of that firm when they retire.

That creates really bad incentives for the high performers. If you’re a really high performer, why would you stay at that place? You go start your own thing. That’s part of the reason we’ve seen such a proliferation of new funds pop up.

The amazing thing about our place is that I have no reason to leave, because of the generosity of the founders who stepped down and said, “You know what? We want to empower the next generation. We’ve made enough money. Here’s our carry.”

It means it’s ours to run. It means I can look you in the eye if I’m recruiting you to be a principal and grooming you into my next partner, and say, “You have a real chance to be an equal partner alongside me.”

Final one. When you look at the next 10 years and think about excitement, I’d like to end on a theme of positivity. When you think about the exciting things that are coming, what are you most excited about when you look forward to the next 10 years?

I’m very excited by drug discovery, especially around MS. My mother has MS, and I’m excited about what will be enabled there.

I had a conversation with Sam [likely Altman] 3 months ago. He was talking about the cognitive dissonance he lives with every day, knowing that the gains and improvements between GPT-2, GPT-3, and likely GPT-4, which took just a few years and were incredibly exponential—and obviously GPT-3.5 was the moment the world changed—are very likely to be matched or probably dwarfed by the improvements we’ll see over the next 2 or 3 years.

How do you make decisions right now about what to invest in, how to live your life, and how to raise your kids, knowing that this change is coming? That cognitive dissonance is very hard to live with for all of us, and certainly for Sam, who has a courtside seat.

I asked Sam, “How should we think about raising our kids, knowing that this is changing so quickly?” His first reaction was, “Don’t teach them to code.” Then he said, “What I mean by that is, teach them the logic of how to think, but you don’t necessarily need to teach them the mechanics of coding, because that’s obviously likely to go away.”

He said you need to teach them how to understand how people are thinking and feeling, and how to influence that.

I excitedly came home and told my wife, who is a career leader of big revenue teams and is now teaching the course on sales and persuasion at Stanford Business School, “You’re the future.” I’m grateful that I had kids with you, because it means our girls are going to get this inherently by having you as their mom.

When I think about the future and what I’m excited about, I’m excited for all of the things that are more rote and less creative to go away. I’m excited for the fact that we can hopefully be more human.

We really can connect more. I can spend more of my time understanding how you’re thinking, how you’re feeling, how I can make you feel better, and how you can make me feel better.

Hopefully, we’ll have more moments like the one I played for you before, where my 3-year-old sings “Shake It Off,” the Taylor Swift song. We can have the machines put the music behind it, and we can focus on her and elevate her.

Jake, I so appreciate you coming to London. I so appreciate the walk around London. The loveliest thing is making a new friend. I know it sounds strange, but it’s a really cool thing.

I feel the same way. I really enjoyed it.

I so appreciate you, and thank you for being so great.

Thanks for having me.