How Benchmark Capital finds EXCEPTIONAL startups in the age of AI — A convo w/ Eric Vishria

AI companies that we're working with that have gone like zero to ten plus in a year, and contrast that with you never seen that before. Never saw that before. We have zero to thirty, zero to forty, zero to a hundred in like fifteen months. These are insane growth. Customers are like they see these products and are like holy shit this is magic.

Thanks for joining us, Eric.

Thank you for having me.

We've we've known each other on Twitter for many years. First time we've actually met in person. Thanks for having us at the. Benchmark office here in SF. I thought I'd kick it off by telling a quick story.

All right.

I hope you don't mind.

Yeah.

So I'm pulling up an email from February of 2019. So I guess 6 years ago, almost exactly. And again, we knew each other from Twitter, but I sent you this cold email. Hey Eric, I'm Tyler Hogue. Here's who I am. Quick background. We haven't met in person. I admire Benchmark 'cause of Andy Ratcliffe. I'm reaching out because a few weeks ago I had lunch with Matt Miller at Sequoia. and I asked Matt which investor he admires most, all in Silicon Valley. He immediately said your name. And then I saw your head shave tweet about Confluent, and it was pretty clear why Matt mentioned you. Anyways, I went on to ask you a couple questions, and I wondered if you'd be willing to tell the story here. Basically, how did you pick Confluent early on, and why did the founders end up going with Benchmark? And we have a couple follow-ups there, but I'd love to start it off with a story, if you're open to sharing.

Yeah, it's a good story. Totally. So a friend of mine, who's actually cousins or second cousins with Neha, one of the Confluent founders, is the one who introduced me, JP. And JP introduced me— I was 2 weeks into my investing career.

You were just barely starting. You were a founder before, right?

Yeah, I was a founder before. And so I was, like, literally 2 weeks into Benchmark. And there's always advice like, don't make an investment too fast, or whatever else. But you're meeting all these companies. And I had some experience with Kafka from my startup, Rockmelt.

Yeah.

And so we had used Kafka, and it was like this great piece of technology. And so we had some familiarity with the space, or at least with what it did. And I kind of thought going into the meeting, like, hey, this is kind of this nichey thing. It's a pipeline. It's a pipeline. that puts data into a data warehouse or into, at that time, Hadoop. And the thing that happened was, you know, Jay and Neha and June, but Jay led the conversation. He articulated a vision of why, like, yeah, it looks like this pipeline thing, but it's actually much, much more than that, and there's gonna be this big real-time data hub. This was— they still worked at LinkedIn when we first met, so it was the initial pitch.

And it was developed internally at LinkedIn, right, the technology?

Yeah, it was developed and open-sourced while they were at LinkedIn. And you're sitting there, and you're listening to this, and you're like, huh, I never thought of it that way. Nobody really has talked about it that way that I've talked to. or that I've talked to, at least in the investor community. And yet, it seems totally credible. Yeah. And, like, it's a cogent view. And then you kind of, like, dig into it, and you're like— you ask questions and everything else, and you're like— you know, the idea— Chris Dixon's Idea Maze concept is so good. And you're, like, asking them questions, and you're like, wow, these guys know everything. about this space. And, like, they know everything, including the things that they don't know, right? Like, it's like, we don't know how fast stream processing will happen. We don't know how this is. Like, we think this is our theory of how it monetizes, right?

This isn't—

it's not even a company yet. And, um, and so I walked out, and I sent a note to my partners, um—

By the way, I have that email in here.

OK, yeah.

I won't read the whole thing, but I would love— can I share 1 or 2 sentences from it?

Sure, yeah.

So this, this was really cool, 'cause, by the way, you're just engaging with me over a cold email, and you said, thanks for giving me the trip down memory lane. I also forwarded you the email I sent to Peter, Peter Fenton, after that meeting.

Yup.

And here's what's cool about this. You said, Peter, Zach and I met with the 3 Kafka founders today, and we both walked away thinking we have to do this deal, in bold. So you got to conviction pretty quick, it seems. Would really value your thoughts and help on this one. It will go fast, so can we meet as soon as you're back? And then you go on. It's a much longer email, but what stuck out to me, and this will get into the benchmark model, is like, you're immediately reaching out to your partner, Yeah. To help on this deal.

Yeah.

And that feels very— I don't know if that's super common in venture or not, but at Benchmark, it's like second nature.

Yeah, it's awesome.

You guys are equal partners. You're doing it as a team. Can you talk about that, and what you did here with Kafka, and why you reached out to Peter?

Yeah, but keep in mind, I'm 2 weeks in.

Yeah.

This is like, you know, like, a lot of this is, like, just better lucky than good, right?

Yeah.

Like, it's just better lucky than good. Peter's one of the great, um, great investors, great venture capitalists in the history of the business, but— But certainly amongst the top, I don't know, two greatest open source investors. And so he was like a perfect, perfect person to reach out to. And he replied, and they set up a meeting right away. And and and he actually said, if I can't remember the exact timeline, but it was like a Wednesday or something. He's like, "Hey, we should just like get everybody, like get everybody to meet together." So the other, the other their There were 4 other people besides me at that time. And, um, and so like, let's just everyone go meet. And, um, and so it's just like you get the benefit of everyone's experience on it and, and their assessments.

No competition between the partnerships? No, like, like, I own this deal, you don't?

Yeah, nobody cares. I mean, I think there's just like not a— not a— nobody cares about that. The economic model's set up that way, so we should we shouldn't care. And, you know, and you kind of like— and through that, I still serve on Confluence board, and it's been, like, an amazing thing to partner with Jay and the team from 3 people to, I don't know, 3,000 or whatever, from 0 to 10— to, you know, $1 billion in revenue, and more to go. And, like, I think it's just like when you get to go on those journeys, it's so special.

So it's a perfect segue, by the way, because I think at the time— you can fact-check me here— Confluent was one of, if not the fastest-growing tech startup Silicon Valley had ever seen. I think Sequoia said that's their fastest-growing company in their portfolio. And then you tweeted, just like 2, 3 weeks ago, this tweet, which I think should be the bulk of UnConvo here. You said, hey, do you remember when triple, triple, double, double, double was the gold standard for annual growth, SaaS companies, and Kafka maybe isn't a SaaS company, but I think the point stands. And then you said, that was cute.

Yeah.

All right, so let's have this discussion. This is what I would love to— what are you seeing? You say it's cute, I assume you mean that's a remnant of the past and things are different now. What's different? What are you seeing?

So I don't know how many AI companies that we're working with that have gone like 0 to 10 plus. In a year, but it's like 8 or 9, like it's some meaningful number. Like it's really like a lot.

And contrast that with, you'd never seen that before?

Never saw that before. Like 0 to, like you'd see 0 to 2 maybe, like you see those kinds of, but like this is like 0 to 10, we had 0 to 30, 0 to 40, like, you know, 0 to 100 in like 15 months. Like, I mean, these are, Insane, insane growth. And and we should just like be really clear that it's not the same. Okay, and it's not the same in good ways and bad ways. Okay, it's not the same in good ways in the sense that like there's a lot of experimental revenue. A lot of it is like you know MRR that's run rated, and so it's not like ARR. Like it's not what I think of as ARR and what anybody should think of as ARR, really. So it's like, it's not that, necessarily, in a bunch of cases. There's going to be more churn. There's like, people are figuring out value, and like, all these things. So like, that's the bad side, right? The good side is, like, it's like real customers. Like, it's not like hokey. A bunch of it isn't. You know, and each of these companies is different. Um, and, and I think, like, at the end of the day, customers are like— they see these products and are like, holy shit, this is magic. Like, I think that's really what's happening.

So you do think it's driven by actual product magic, not some experimental AI budgets just to test it out? There's real product value.

I think there's real— there's real perceived value, and some of it will get realized and some of it won't, but there's real perceived value for sure. And, and it feels magical. And the products feel magical to the customers, and so they're spending money. And I think this idea of magic and a magical experience is just like where people are like, whoa, here's my money. And that could be, whoa, here's my money, and then 3 months later, like, eh, it didn't quite work for me.

And there's the churn you're talking about.

And then there's the churn and everything else. But I was talking to an investor about this, and I was like, look, OK, take 30% off it. 40%, whatever. Like, it's still insane. And so, like, there's still something happening, and we shouldn't be dismissive of it. And the irony is, to me, that we also have some AI companies that are, like, zero, like, that are still, like, zero in revenue, and they're a year, 2 years, 3 years into development. And I'm not actually, like, bearish on those either. Like, part of me, like, I can make the case both ways. And when we talk about—

Well, you're not That old phrase of lemons ripen early, right? Like you get, you find success too, child star syndrome. You like don't grow into it.

Is this all— I think that's real. And I think there's a defensibility question that's really important here, which is, you know, some of these companies that are ripping are based, are like riding on top of the billions of dollars of investment going into LLMs. And I think that's one of the things you're saying, like you don't wanna be fighting that, that's for sure. Like you don't wanna be fighting the billions of investment. Great if you're riding it, but like you do have to make sure, like in most of these spaces, it's not like one company working. In some of these spaces, it's like 4, 5. Now, like generally speaking, as we all know through as students of history, like that doesn't end well.

Like even though 4 or 5 are initially working, and that's because the market usually has one big winner, generally speaking. Is that what you're saying?

Yeah. And you just get the gorilla.

So this is what's interesting to me, because Tyler and I were actually talking about this before. We're like, this year, will there— we were saying 0 to 10, how many? We put the over-under at 100. If Benchmark has 8 or 9, that—

Yeah.

OK, maybe 100. That's probably a 10x over what it was, you know, ever before, its all-time high. So here's the question, because that is— it's a really cool thing. It's real. It's going to be around. It should change how founders think about things, change how investors think about things. But does that mean you're, like, not a venture-scale business if you're not doing that, like, going forward?

No, I don't think so. I don't think so.

Because I don't either. But I think a lot of people are saying that's— T2D3 was the formula, and now it's, you know, speed of 0 to 10 or 0 to 100. And that's, like, where I kind of, like— that kind of rubs me wrong. I think that's stupid.

I think that's stupid, too. I think that's dumb. And I've never— like, the D2T— or whatever that was, like, the D2T3 thing— like, just to be clear, like, people talk about it and everything else, but it's never, never, ever how we looked at things. And, like, it just— we never look at things that way. We talked about Confluent. I think about Cerebras, or Fireworks, or some of these others that I work on, or Sierra, which Peter works on. And, like, those are all— those were all zero— like, they were founders and a deck. Like, there was nothing there. And, like, that's— you know, and those were, like, founders and a deck, and we just did them. And we did them with conviction. Like, we took a board seat.

Pre-revenue, generally speaking, these are?

Yeah, yeah. All of those are pre-revenue. They're all pre-revenue. They're pre-company.

Yeah.

And so I don't think that is it. I think there are different dynamics in each of these companies that have to be torn apart. But ultimately, for these in competitive spaces where they're riding the LLM thing, and they have momentum, some of those will work.

Yeah.

But you need to make sure, double and triple sure, that, like, the founder is an absolute assassin, and that they are shipping fast, and they have some theory of how the, like, quick sugar high turns into something durable. For some of the others, you know, that where they're like developing maybe their own AI and it's not in conflict with the LLM stuff, but it is also not necessarily benefiting in the same way from the LLM stuff and the billions of dollars investment there. It's gonna take time.

Mm-hmm.

And that's okay. But like, I think what you end up with on the other side of that, if the market works and the AI works and everything else is like, I think you might end up with something that's much, much more durable. And there's not gonna be like 5 of them. There's gonna be like 1 in most cases.

You brought up something twice there, and maybe I'm reading too much into it, but you used the phrase fighting against the LLM or working with the LLM. Feels like, how do you determine that? And one thought I've had is the more vertical you are, the more insulated you are from competing with the LLM. And so potentially vertical applications might be place to benefit from the LLM as opposed to fighting it. How do you think about that?

Yeah, I think this is, I think it's the right question. I kind of just think of it as like, okay, can I imagine if I fast forward to, I don't know, GPT-6, GPT-7? Like if you start thinking in that context, like, you know, or DeepSeek R4, like is that going to do this? Is it going to do a lot of what this product does? Is it just gonna eat it? You know, and we saw a lot of this with the marketing, uh, copywriting stuff early, right? Like, it's just like, it was like, yeah, it was, it was amazing. And then it was like just cannibalized by the thing it was built on. And, you know, and I think some of them are continuing to do all right, but like, like, but it's tough. Like, it's tough in that space. So I think there's things like, just to be concrete, like I work with a company, uh, Quilter AI that does printed circuit board layout. And like, they're building their own AI for it, and it's been a work in progress, like, and it's just, and its capability is kind of 10x'd over the last year. And I think there's like another maybe order of magnitude to go before you start getting to like real, like, oh my God. And I think that's coming soon 'cause we have a clear path for it, but it's not related to the LLM stuff. And I'm really confident that like, GPT-7 isn't gonna do like PCB layout. Like, you're just like, it doesn't, that's not part of the training set. It's not something that it can do, I think. And so like, you know, that feels like it's like once you cross over, you get to something really defensible. And so those are like maybe concrete examples.

Super helpful. So if the triple, triple, double, double, double thing's not like—

That was my question too.

What is the new—

What do you get excited about?

Yeah, what is the new one? What is the standard?

And let's take those, like, you know, when you're at zero and you're building for a long time, let's just take it off the table. When you are scaling—

Yeah, what should you be saying?

Yeah, what should— like, when you sit down and you're doing an AOP at a board meeting, it's like, when do you say, hey, guys, like, we've got to go— we've got to be more aggressive. We've got to be faster.

What's an AOP?

Annual operating plan.

Oh, yeah, OK. So I think this is, like, very company-specific. And I think one of the things that happened in the SaaS world— is it became a little formulaic. And so the investor-bankery-type investors, like the banker-type finance investors, they figured it out, and they have their formula.

They sucked the alpha out of the whole market.

Yeah, and it's like, oh, the magic number, and blah, blah, blah. And I don't even— it's just like— and they're like, eh, and the pencil, and then therefore, it should trade at 13.2 times forward. And it's just like, OK. And you know what? They were right, and they made a ton of money. investing in those companies, because there was, like, an arbitrage between people who understood this business model—

Yep.

—and how, like, the real power of the business wasn't reflected in the income statement, and, like, the durability, and all these things, and, like, different gross retention, net retention, all these things. So there was, like, there was an arbitrage from, like, I don't know, 2012 to 2018, that just, like, these— and these growth funds, and, like, everybody, they just exploited it. And they're all, like, bankery types, and they crushed it. And they crushed it. And that arb is gone. Like, you know, everyone figured it out and like, whatever, so it's over, as arbs—

Tend to do.

Tend to do. And I think this new AI one, maybe there is like, maybe there's just an arb on AI generally because like, it's just like so powerful. And if you just project it, like, it's just so incredible. But I don't think it's like, But it's not formulaic in that way. And so I just— I have trouble, more trouble with the plan, and just kind of saying, like, this is the standard on them.

Fair to say, like, benchmark class is classic, but, like, game on the field, so it's, like, a little dependent on, like, which wave you're riding. And so you're like, hey, this plan, and how we're thinking about growth, how do we think about it in terms of how does it position us to win, and be one of the— you know?

Totally.

It's still more like that. That makes sense.

Yeah, and you kind of think about it, and you can look around, and you can say, well, are there competitors dominating? Are there adjacencies that are crushing? How are we cannibalizing? What are the different things? I was talking to a really smart investor yesterday, and one of the things that he made the point on, which I thought was like— which I had never thought about, was one of the reasons that Cursor, which is obviously doing incredibly well, one of the reasons he thinks the adoption was so quick on it was that for the 3 years prior, Microsoft was like carpet bombing the market with Copilot. And Copilot came, and so then this thing came along that was like, actually like really worked well for this thing. And so like everybody was prepped. And I thought it was just like an amazing insight that you just said. It's like, I was like, oh, I hadn't thought about that. And so like, These are things that we don't— you just have to really look at them.

Tyler and I were talking about this on the way over, but there's a big difference between Cursor and Devin, for example, right? And one of the things that you're talking about is the entry point, and how prepped is the market. And they tend to move an inch or 2 at a time, sometimes a foot at a time. They don't move like 10 feet at one time.

You actually have a tweet— oh, finish your thought.

No, no. And so I'm just like, that to me is the point that you're making.

You have a tweet on this that I think similar concept. Eric, you said, while the copilot approach is all the rage, I think we're gonna see a bunch of areas where the AI just does, like completely does tedious work. So let's talk about that for a second. Copilot versus maybe pilot, or like it just does it.

Do you think they graduate or can you, in many cases, are we gonna see 'em skip to kind of the final version?

I think there's this, I have this idea in my head, which is there's a copilot. I think there's a set of areas where copilot is the right, approach, and there's a set of areas where we're going to be AI and done, and, and maybe some crossover as like the AI becomes more capable. The AI and done thing is like this PCB layout company, and, and where it's just like, you know what you don't really want? Like, you don't want to do that. Like, you just don't want to do it. Like, it's just better like for a computer go do that and, and come back and just be done with it. Um, but there are other things where, you know, the human Like guiding it or getting kind of superpowered on it is really powerful.

Or getting comfortable with it and then being able to put it in a mode—

Well, I think that's another important part of this enterprise adoption thing that's going to happen in a bunch of cases where it's going to take— the human will be in the loop. And then this is typical for automation. It's like, oh, I'm going to check it. I'm going to check it. I'm going to check it. And it's like, it takes about 30 days before people are like, oh, I didn't really check it anymore. So I think both models will be there. I think for code, we're gonna get to a place, you know, Dario made some comments, I think today or yesterday, yesterday about, you know, how much code generated is gonna be actually by AI versus people. I think that is clearly working and it's working really well. On the other hand, you know, you do have these cases, like there's one company I was talking to had hired, a relatively junior engineer, and like 90% of the code was written by AI, and it was garbage. It was just like garbage. And so there's this other kind of irony of this, which is like the great engineer is going to actually be more valuable, like the truly great engineer.

It's an amplifier.

It's an amplifier for them in the future. And there are other areas like accounting, which we have a company, Digits, you know, legal work, like I think Alayha Law and like other things where it's like in that stack, like constructing a brilliant agreement and a legal contract, I just, I think that's like a, at least for the time being, like that's a task for a human first. But there's a lot of underlying execution and rote legal work and paperwork and whatever that is like clearly AI can just do. Right? And so those are kind of the dynamics to me.

You mentioned something earlier. Basically, these companies that have scaled so well, a lot of them you funded just a deck and a team, pre-product, pre-revenue. What is Benchmark looking for in order to fund somebody pre-product, pre-revenue? Are these second-time founders only? Are there certain— how does it work at Benchmark for the partnership to get excited about a deck An idea, and a founder.

You know, our former partner, Mitch Lasky, he described it to me. He was teaching me in the, you know, whatever, 10 years of getting started. And he's like, look, it's not— there's no bar. It's not objective. He's like, the better analogy is it's falling in love.

Yeah.

It's falling in love.

I call it spiritual conversion.

Yeah, it's spiritual conversion.

You feel like you're baptized by these guys.

Yeah, you're just— you feel a connection. You know, each of us will articulate it a bit different, and I think each of us has, like, different tastes and whatever, but for me, you know, there's obviously this, like, big market thing. Like, what's happening in the market that is gonna allow this thing that really the world doesn't wanna have succeed to, like, break through? So, like, that's just, like, a big market thing, and, like, what's shifting, and, like, is there a seam, and all this stuff. You know, then you're, like, looking at the founder, and, you know, I just am like, is this the founder that is going to grow, and scale, and grow with this company? And I think that that's actually a really important thing to me, is, like, is the founder a hyper-learner? Because I find, like, you know, the job at 3 people is really different than 30, than 300, than 3,000. It's different at different scales. Like, these are really dynamic jobs, and they're changing.

Yeah.

And it's really tough. It's like, I said this before, like, I just, in some ways it would be better if there was like CEO 1 and then CEO 2 and then you graduated as CEO 3, 'cause it's the same title and same thing the whole time, but it's a totally different job, like every 6 months. And then the third thing is really like, can't be overstated, is an insight.

Mm-hmm.

And we kind of talked about the Jay example with Kafka and the real-time data hub and like, hey, this is gonna be a thing. Thing. I think of that almost for every investment, probably for every investment, there was an insight. There was somebody who— the founder came in, had seen stuff, and they were just like, no, but this. And you're sitting here, and, like, we meet companies, all of us meet companies every week, right? We're meeting companies every week, and you're talking to smart people all the time, and you're, like, reading all this content on Twitter and everything else. And then, like, someone comes in, in a domain that you know a little bit about, at least, and says something you've never heard. And you're like, wait a minute. And then you think about it, and then you ask questions about it, and you're like, that's totally cogent. I don't know if it's right. I don't know if it's right.

But it's at least logical.

But it's logical. It's well thought through. There's like— and you're like, if that's right, and this person has an insight, and they're a killer, And, like, this market is big. Like, that's the formula. That's the formula. That's the formula. And, you know, on those kinds of things, we've written really big checks, like, just like that. And I think, like, that, to me, is— it's rare. It doesn't happen that often, when you, like, fall in love, and you have those 3 things, and, like, all this stuff. But when it happens, it's magic.

It's so special. Do you know pretty quick when you meet it? Like, first meeting, you fall in love, or do you buy that, or does it take a while to get to the— Conviction.

Again, each of us is different. For me, I feel like— I just am like— I'm a romantic.

Yeah.

So I'm there, and I think about Lin and Fireworks. So—

They made the list 2 years in a row, right, the Exceptional Startups list?

Yeah, it's an exceptional company, and hypergrowth, and one of these crazy revenue-scale growth. She ran the PyTorch team inside of Facebook, the PyTorch engineering team, and the founding team kind of came from the PyTorch team. And she came in, and I generally don't like people from these oil state companies. Like, I just find—

Well, you did LinkedIn and I did Facebook, so I think you might like them.

Yeah, maybe. What do you think? Well, I would say LinkedIn to me was not an oil state company.

What it is now.

Well, it wasn't an oil state company in the sense that, like, They were commercial and they ground it out. Like, you know, like Google—

Yeah, private resource constraint monopolies. They were really duking it out.

No, I actually agree. Yeah, LinkedIn had to do some hard stuff. Google just collected money.

Oh, dude, so much, so much hard stuff. You know, these other companies, just like they poke a hole in the ground and oil starts coming out, and it's like, okay. Um, so I generally don't like them. And so, you know, she comes in and I tell her that in like the first 10 minutes.

That you don't like Facebook people, basically?

I just don't. It's not— they're incredibly smart and very talented.

But not great founders.

The zero-to-one thing's a different skill set.

Zero-to-one's a different skill set, yeah. And that's a much better PC way, a very euphemistic way to say it. And so, like, I— and just, like, the way she reacted and everything else. And then she pushed back at me, wait, wait, why? Why do you say that? No, this is what we're doing. This is how we did it in PyTorch, and this is how PyTorch came from behind and beat TensorFlow and everything else. And I was like, 10 minutes in, I walked out and told the receptionist, I was like, please call Chloe, my assistant, and push my next meeting. Push my next meeting, because I need all the time I can, because I'm doing this investment. And then it took 6 more weeks, and she still worked at Facebook, and the team was leaving.

She hadn't even left Facebook.

She hadn't even left Facebook yet.

That's the other part of your playbook. They're still employed. And they work at oil state companies.

Yeah. And so I, you know, I feel like, and that doesn't always happen, but when it happens, you feel like this. And, you know, she subsequently pivoted, the company's pivoted, and—

So was the insight wrong, or was it—

You know, the insight changed in this AI world.

Yeah.

It was like a PyTorch cloud idea. And then she was like, well, GenAI's happening, and really what people need help on is not building, not this like base layer, But one layer of abstraction up and they kind of moved to that.

So was it the fact that she pushed back on you and was inquisitive and like stood her ground? Like what was it that she did well there that caused you to be like, man, I'm kind of in love?

And part 2 to that is like, is it the insight or is it that they have the ability to have an insight and you're like, even if this isn't the one, the way you got to this one will lead you to the next one?

You know, they're all different. They're all different. This is the law of small numbers. Nothing statistically significant.

Exactly.

In that case, there was an insight. I never understood why PyTorch— TensorFlow came— I guess for your audience who didn't know, these are AI development frameworks. TensorFlow came out of Google and it was totally dominant. If you look at the NeurIPS papers from, I don't know, when TensorFlow was— it was like 99% of the NeurIPS papers were built on TensorFlow. And then like 3 or 4 years later, it's like 90% PyTorch. And like, these are big projects that are relevant to really good companies. And so like, how is it that a new framework came along and just, you know, beat it? Like, I understand Caffe and Theano and some of the previous ones before TensorFlow, 'cause they didn't have that kind of backing, but like, this is a totally different thing. And she explained why, and she just articulated why. So that was part of the insight.

Yeah. Yep.

And then why that insight would turn into something else. So I think that was, like, a really important part of it. And you're just, like, learning. It is the inquisitiveness part of it. It's the back and forth. Like, one of the most important things is, like, why do we do this? You know, and each of us has our own things. But for me, I want to work with entrepreneurs. Like, that's the part that gets me out of bed. Like, the investing part is, like, whatever. Like, it kind of— I'm like, that's a means to an end. For me, but I want to work with entrepreneurs. So like one of my other things is like if this person calls me at nine p.m. on a Saturday, like am I going to look at my phone and be like uh or am I going to pick it up? And and I just like that to me is like that's such a good test. You know another kind of question I always ask myself on these investments is. We've all worked with these amazing people in our lives that you trust and you feel— I feel a lot of loyalty to all the people at RockMelt, for example, who stuck with me through thick and thin. And I think they're excellent, and I want the best for them. And I was unable to deliver or lead us to an amazing company. And I want them to have that opportunity.

Yeah.

And so one of the things I think about is, like, would I talk RockMelt people into, like, working here?

Would you send a friend to this place?

Would I send a friend to this place?

Or would you work for them? Maybe age-adjusted, would you go there?

Yeah, yeah. Would I work for them? Or would I— and would I— do I believe that, like, I could very authentically convince people or sell people that this could be their life's work? And, like, there's a lot of businesses, like, good businesses, Which are— which I'm totally fine passing on and not working with, even though they're good businesses, and they'll be good returns, and they'll generate, like, venture returns, because they— like, I wouldn't want to work with that person, maybe, or, like, I don't have good chemistry with that person is probably a better way to say it, or I just find, like, it's just, like, just not a business that I get excited about.

Yeah, for whatever reason.

For whatever reason.

I'm actually very curious, too, This— what's the founder of Fireworks? What's her name?

Lynn.

Lynn. How did you meet Lynn? Because I— the sourcing part of this job is tricky because was she meeting with 5 other firms that day, or was it just you? She's still at Facebook. So, yeah, how did you source this meeting?

Um, so in that case, another founder introduced me.

Yeah, which is generally the best.

Which is the best. I think I get a lot of introductions from other founders.

Was the other founder someone you had invested in?

Jay Kreps. Oh, OK. Full circle. Yeah, so that is really special. I tend— I think a lot of stuff comes from founders, like other founders. And it's founders that I've invested in and founders I haven't invested in. And that's so nice when they come and you get that. It's the highest compliment that a founder can pay you. And and so it's it you know it's really special. I do cold outreach too. You know I I like doing that. I don't do it as much as I should anymore, but but you'll see things like there was a company called Octant Bio, and it's outside of my domain, but I really respected the founders, followed them on Twitter. And he said, like, hey, one of his early employees was going off to start a company. And he was so excited about it. And he thinks it was going to be amazing. And I just, like, DM'd him right away. And I was like, hey, I'd love an intro. Like, that sounds amazing. And I got to meet Nicholas Laura Stone through that. And he came in as EIR. And he started a company. And, like, it's a pleasure to serve on his board. Like, I just— and so, like, that's a really— That's a really special thing when that happens.

That's awesome. How often do you do the EIR play? Is that a common thing here?

Um, you know, it goes in fits and spurts. I think like right now, I don't know if we have any. I don't think we have any EIRs, but in different periods, yeah, we'll have them. I've done it 3 or 4 times.

Any other questions we should dig in on AI or this other stuff before we switch topics?

Oh, you know, just the classic, you know, you've No, no, no other ones.

We have a lot of debates on how this stuff's gonna play out, and—

It's hard to know.

It is very hard. Like, actually, I will ask one. Venture capital. Like, these founders are famously scaling much more efficiently, you know, fewer people at least. Maybe they're still raising big rounds, but do they need to? The direct question is, is venture capital even needed in a lot of this case? Will there be fewer firms in the future? What— how does the landscape change? And Sterling's point usually is like, won't it be the case that almost there's an adverse selection effect of the worst founders will raise venture capital now? Because the best ones can just get so far without it, because you can create something much easier than you ever could before. Now it's the stupid ones who need to raise venture capital.

It's also to your point earlier about— I love this— the investment bankification of—

Yeah.

Like, that doesn't exist right now.

No, it does not.

And so a lot of people aren't gonna know how to deal with that. Like, a lot of firms aren't gonna know how to deal with that. So at the same time as they're gonna be, like, struggling to adapt, you're having founders that there's a totally new way of building this stuff. And yeah, I love that question.

Um, I hope not, maybe is one way to say it. I think it varies. Like, I think there's gonna be some amazing bootstrapped companies, like, amazing, that will get very big and very large, and be able to, like, leverage AI to, like, to just, like, rip. I think, you know, there'll be lots of things. But to me, there's something that's lost in that, in the question, which is venture capital is 2 things. It's capital, and it's partnership. And I pay a lot— I put a lot of value in the partnership part. It's both a part that, like, I personally love, but I think it— I believe in it. I believe in that value proposition. And there are other ways to get that. There are other ways to get that, for sure. And, you know, most of the great companies, if you talk to those founders, they had partners, like, what we think of as partners around, who helped them and helped them make the business better and pushed them and asked questions and, you know, And we're sounding boards, which is my favorite analogy for it. But I think that's, like, a really important part of it. In some ways, the whole business— the business of venture capital kind of artificially tied these 2 things together.

Yeah, they're bundled.

They're bundled. And they don't necessarily have to be. And so maybe there's some future evolution that is where there's—

Well, you answer another question we debate, which is, like, what are the good reasons to raise venture capital?

And do you need— and is the money a pro or a con or a neutral part of that?

And what I'm hearing you say is one good reason that apparently some really, you know, obviously some fantastic founders have believed is that the value is in the partnership, because they can get money from anywhere, but they choose to have this non-executive partner called a venture capitalist on the journey with them. And so to believe that doesn't exist in the future would mean that you don't think that partnership's valuable.

Or that you can get it another way.

Or you can get it another way without the money, perhaps.

Yeah.

But I think the partnership component is pretty, like, yeah, if that didn't exist, then it's just money. But if you have that, you can see why founders prefer that.

Yeah, I think it's a very special part of the job, and it's a privilege to get to do that.

So we've probably got like 10, 15 minutes left is all. I would love to, we've got a bunch of early-stage founders that listen to this. Some may pitch Benchmark at some point, some may not, but for those who don't know much about how the firm works and some of the unique cultural aspects, Can you talk about a couple of things? So one, one, for example, I talked with Chatham at Sundance a couple years ago, and he, he mentions the famous Monday partner meeting, all day, free-flowing, unstructured. Is this— is this right? Talk about why. And it's a little different than other firms, right? How do you guys do Monday partner meetings?

Um, we do like 5 or 6 hours of, um, of just us in a room, um, and we're Just partners. Just partners, and that's all that's here. And we're just talking. Like, there'll be a company pitch, you know, maybe every other Monday or something. There'll be sometimes a company comes in to update. Sometimes there's a guest, lunch guest or something, but it's really just, and there's no agenda.

No one, there's never an agenda. You just show up.

You just show up.

You just show up and you're talking about, Companies, what happened in the week, what did you see, um, you know, things that are going on in your portfolio companies, like executive searches that you're working on, just—

So why not structure it? Why not have 5 topics and go through, uh, you know, like, what's the reason to keep it?

One of the advantages of it is, is there's 5 of us, right? There are 5 GPs and, and, and our, um, our operating partner, our lawyer in the room. And so there's no, like, the nice thing about that is like, you don't need to, like, you can, you can just like talk about what is most important.

I'll tell you what Jaden said at Sundance. He said, people have proposed or talked about changing things, but then the fear is that you'll break something that's working. And so no one will actually go through with it and you just kind of keep doing it.

Yeah. I don't like, you know, and if we're like yesterday, I think we were done like 30 or 40 minutes early and we just like, it was like, fine. Okay. We're done. Like, you just move on. So I don't know that there's any— and I think that part of what happens is, like, these firms aren't companies. Like, we're not a company, actually. And you want a loose and wandering exchange of ideas. Like, that's what our job is to do. It's to see the present very clearly. And to see the present very clearly, you have to just wander. And, like, you have to, like, understand, and you have to debate points, and you have to, like, do that. And so that's kind of what we do. And then we go and we go do some work. And then we typically have a dinner guest or have a dinner with a guest here around this table or a similar one in Woodside. And yeah, and it's just another opportunity to learn, exchange ideas.

You brought up that quote, the job is to not predict the future, but to see the present clearly. I think it's Peter Fenwick.

It's Kohler's. Oh, it's Kohler's.

OK. Talk about that, because a lot of people would say you're actually you're trying to predict future markets and invest for markets that don't exist yet. What does he mean by that?

Well, I think we should say like I think there are different different venture capital models work. Like there's not just one model that works. A few weeks ago, I was talking to Mike Spicer from Snowflake and Pure Storage.

We trying to get him on the podcast for two years now.

He's unbelievable. Like, in just unbelievable track record and unbelievable investor. You know, one of the things he's like, he likes to start companies and he likes to be the initial CEO. He was the initial CEO of Snowflake. He likes to start them and be the initial CEO of the companies and everything else. And, you know, I think he'd say like he doesn't like investing in other people's companies. That is, that is a, that is a capability. And here's which I mean, like what something to aspire to, but like I can't do that.

Yeah.

And so, like, it's amazing that he does that. I find it amazing to partner with founders in their earliest stages of development, but I can't help them, like, figure out the product. Like, I can't do that. I can't help them figure out the technology. I can't do that either. And so, like, that's a different element of it. I think I can say at times, like, that's not right. You know, like, that's not gonna work. Or like, we, you know, there's something busted here. Like, I can say that and have that pushback, but I can't actually help them figure it out, that part of it. But then there are other parts where I think we can help a lot. And so, like, that's the system. And so I think there's, like, different models that work.

My mind was going somewhere on this. Oh, oh. I would love it— Eric, would you mind sharing one thing you've learned from your other 4 general partners, if we go through them all? Like, what have you learned from Peter? What have you learned from Sarah? What have you learned from Chaython? And then the 4th partner— 5th partner is Victor. He's the newest, right?

Yeah, Victor's the newest.

Anything you can share or say for each of them?

Sure. So I'll go in reverse chronological order. So Victor's our newest partner. He's the founder of Wildlife Studios. which is a game company, and with his brother, who now runs it, and he's scaled it to hundreds of millions of revenue, bootstrapped, to your earlier point, and then partnered with us, actually, as well. And, you know, one of the things I think Victor's so— he's pushed me on so much is just, like, the job is, like, to work with excellent people, and, like, period, full stop. I mean, he's like, this is, like, that's the beginning and end of it. And he's also, and I just think that the clarity of that articulation is very useful. Like, I intuitively knew that, do that, but it's just like the clarity of it. And so I love that. And he has really pushed and continues to push me on, on like the risk-taking. Am I taking enough risk? And, and I appreciate that. You know, Chatham is He pushes in a lot of ways, too, but he is— I think he characterizes me by his resourcefulness on figuring things out. We cover a lot of the same stuff, and so his ability to see, like, what is moving early, and, like, what has actually started to break out, and, like, what's interesting, what's getting developer mindshare, everything else, he's maybe 2 or 3 levels more technical than I am. And, like, it's so powerful, I think. Um, Sarah, um, is kind of a— she's, she's been a venture capitalist. She started as an associate at Bessemer.

Um, found Pinterest there, right?

And yeah, she found Pinterest there. And, and, um, and then she joined Pinterest. And, you know, I think the thing that she has through that training is, like, coverage. like, meet companies, see a lot of stuff. And when you see a lot of stuff, like, you know, great things happen. Like, these connections happen. One of the conversations Sarah and I have all the time is, like, if I have free time, I tend to call a CEO that I work with. And— but the right thing to do, which is, like, nice, I think it's nice, but, like, part of what I should be doing is, like, just, like, casting a wide net. And spending time on that. And then Peter, who's been here the longest, there are so many things to have learned from him. But I think the most valuable from an investing perspective is we don't do memos, and we don't do this. And Peter always jokes, if you have a data room, we probably wouldn't open it. So the question is, why? And I think that the key thing is, for most of these companies, there's really only 1 or 2 questions that matter. Like, after you get through your founder thing, there's really only 1 or 2 questions that matter. And I think his superpower is just figuring out what those questions are. That's the whole game. And everything else is bullshit. And so you go through these long memos, and you go through these data rooms, and all this stuff, and it's just a bunch of— it's just a bunch of pollution in your mind of really what you need to figure out, which is, what are the 1 or 2 questions that matter?

You know, it's an interesting insight from— you know, Keith was saying the same thing. He's like, I like to invest early, because anybody can see and read the numbers, and that actually, like, makes me worse than if I am— because I'm not focused on what actually matters.

It's the same point you're making. It makes me worse, too.

And again, I think probably worth clarifying that this is for early-stage investing, right?

Yeah.

data rooms start to matter a whole hell of a lot, you know, when you're in your CED, you know, IPO kind of stuff. But, but sometimes the initial momentum, or lack thereof, is pollution in your brain. I think that's amazing.

And I think you're also saying the one or two things that matter after you've set aside the founder— like, the founder's like table stakes— what are the key things that matter of them being a big company, in addition, right? Do I understand that correct?

Absolutely. Yeah, that's a great, that's a great point is just like, what we care about is being part of the handful of companies that can turn into giants, and that can be, you know, multi-billion, multi-tens of billion, you know, maybe multi-hundreds of billion public companies. Like, that's what we're here to do, and, like, be part of those at the earliest stages that we can. And, you know, thinking through that, it's, like, a very difficult difficult task, and each one of them is different. I joke it's like you're looking at kindergarten playground and trying to like pick out LeBron, and it's like, you know, you're just like looking at the playground, you're like, that one, that one's gonna be like put up 50,000 points.

Like, it's just— that's funny because the analogy holds because like you'd probably go for the tallest one and that would, you know, probably be wrong, right?

Like, you want to look at other things, you want to look at other things too, and there's elements of it But it's— there's a lot of luck involved, and a lot of— but there's some really cool things that you can pick up on.

One question I have, because, you know, don't do memos, don't like data rooms, got it. How do you keep yourself honest about if your reason for investing, if that proved out, and regardless of the outcome, because it's all about the process, and how do you go and check yourself against what made you excited to invest? Versus whatever ended up happening?

Again, it's the right question. So Sarah actually writes notes on her, on the meetings she has so she can keep herself intellectually honest on it. I think she does it and she has a whole like system around it so she can go back and backcheck and everything else. I, you know, I understand that I am telling backstories when I like reflect on things and I think about them all the time. So, you know, I don't know. There has to be an element of intellectual honesty. Yeah, I sent you that. I think on our initial exchange with Confluent, part of the reason that was fun for me to dig up that old email is because it was like, what did I really think then? And I ask myself that all the time for the ones that work, you know, and like kind of go off to the moon. I ask myself that. And for the ones that don't, I also ask myself that. Like, what did I see, and what did I get wrong in the read? And would I do it again? And would I do it again? Like, one company— so Alex from Scale AI, I met him at the A, and I spent time on it, and I loved him. And it was just so clear that he was like— and I think he was like 20 or 21. Like, he was so young, and he was such a killer. And he had this autonomous driving labelling revenue at the time, and it's like people in the Philippines and all this stuff. And I was like, that revenue is not here for this. I just did not feel like that revenue was long for this world. And we kind of talked about it and everything else. So dumb.

And like—

And so I didn't pursue it. And I think I was— like, I think I should have said yes. Obviously, I should have said yes. But I think if I had said yes, he would have gone with me, I think. And like, fuck. Fucking idiot. Like, I'm such an idiot. And like, I know better.

So what'd you miss?

You know what I missed is that like—

Was the pollution you were talking about?

Is that guy—

Yeah.

Is like—

You had him.

He was such a winner, and he was gonna figure it out.

You had Lisanne Al-Ghaieb at the table, and you said, I don't like your business model.

Yeah, yeah, yeah. It was so dumb. It was so dumb, and just, like, so dumb. And, like, I know better. I know better. I know better than that. And I know I— and so, you're sitting there with someone at that age, and that sophistication and talent level, like, their curve is like this. There's no chance. And you know what? And even to go one step further, even if it didn't work, even if it didn't work, which obviously it has, but even if it didn't work, there's no chance I wouldn't have learned a ton.

Yeah.

And like—

'Cause of him.

Because of him. I would've learned a ton from him. And so, just so dumb.

Can I ask this question?

And then we gotta wrap up soon. You're out of time.

I know, this is awesome.

I'll miss a flight for this.

Do you regret the ones you didn't do, like the Alex ones, that in hindsight you feel like, process, independent of the outcome, you should have? Or do you regret the ones you did do, and you're like, oh, I ignored X, Y, and Z, and I did it, and I don't— like, which one hurts the worst?

Definitely the Alex ones hurt the worst, for sure.

Because you only miss out, you only lose one X.

You only lose one X. Like I said, I've been very lucky in that the people I've worked with, I'm happy to work with.

Yeah.

Whether it works or it doesn't work— Campbell said this to me. Bill Campbell was one of my mentors a long time ago, before he passed, and is a role model for me. That's what I want to be, I think, in time. 20 years from now, if people talk about me like 1/6 or 1/10 the way they talk about Bill, I'll be like, fuck, I did it. I did it. My career was successful. And he said this to me. And he was part of Apple, and he was part of Google, and everyone talks about those stories. But you know what is forgotten when people talk about it is he worked with schmucks like me also. And he gave us time. And he gave us time, and he gave us guidance. And I don't think it didn't matter to him that it like, you know, that like, I mean, did it matter? It mattered, of course, but like, it didn't, like, it wasn't like his, his like North Star. His North Star was like the people. Do I wanna work with that person or not?

Yeah.

And like, that was his North Star on it. And to me, like, I, I just, that's how I'm oriented. Like, I just wanna work with, and so I don't, I don't feel bad about that.

Love it. So we have 3 questions that we ask everybody. And you've actually talked about this throughout, so I think I could answer this one for you. But what— we call it the golden spur question, and it's like, what makes you the way you are? What motivates you? Why are you doing this?

I do it to work with the people. Like, I do it to work with entrepreneurs. Like, that's what I do. This is a vehicle to do that. And just like to be part of Benchmark, I think that's the ethos of Benchmark from the the very beginning is partnering with entrepreneurs and working with them. And I just get so much joy out of that. And I think my partners correctly push back. It was just like, hey, you want to be partnered with companies that don't need your help?

Yeah.

I don't know. I think that you spend a lot of time on the companies that aren't working and a lot of time on the companies that are really working, because small improvements or small enhancements on those make a big difference. And so, like, I really love that model.

I could have guessed you were going to say that. All right, the last 2 are pretty easy, I think. The IO Podcast, that's what we call it, investor-operator. So who is an investor you admire most outside of Benchmark? And then who is an operator you admire most outside your portfolio?

Well, the investor, I'd have to say Jim Goetz from Sequoia. and Accel previously. But he— for a few things. One, I'm hugely grateful to him because he told me to be a venture capitalist. And, like, he told me that in 2008, and it took me 6 years to figure out he was right, and, like, all this stuff. But, like, I just am, like, super grateful for him for that. And then two, you know, there are very, very few people who had successful enterprise companies and successful consumer companies, like WhatsApp and Palo Alto, I guess. And so just an absolute winner, and does it with class.

Love it. What about operator, a founder that you just love?

I mean, gosh.

That's Alex.

There's so many founders, but it would be— Mark and Ben gave me my career. And they gave me my start in my career. And Mark sat on my board at RockMelt. And I was Ben's assistant. And they're operators, they're entrepreneurs as investors. And so I—

Love that.

I have a ton of respect and gratefulness to those guys.

Eric, this has been a fun convo, man. Thanks for taking an hour with us. We owe you. We appreciate it.

Not at all. Thanks. Thank you.