The State of Consumer Investing With Benchmark's Sarah Tavel

Hey, it's Eric Newcomer. Welcome to the Newcomer podcast. This week I have Sarah Tovall from Benchmark. On the show, we're going to talk about the state of consumer investing, per investment in paparazzi, the photo sharing company. Just shut down. I hear the team went to Instagram. So it's a moment of hard reflection on what's going on in consumer world. We also talk about pop shop and be real. clubhouse, a bunch of the companies that are trying to make a play at consumer that have struggled. And then we turn our attention to Genre of AI, the one area everybody seems to want to invest. And finally, we talk about crypto. Sarah has not given up on the crypto space. She says she's made an investment in undisclosed NFT company. And we talk about her investment in chainalysis. And also we do at the very end talk about Benchmark and the state of the partnership there. Stay tuned. Here's the episode. Sarah, thanks so much for coming on the podcast. I really appreciate it. Yeah, I'm excited to be here. Right now I'm in my fiance's dad's office. So that's the sort of strange design aesthetic I've been like, I literally was eating cupcakes in vast quantities this morning for my wedding. So I've dragged you into the to the Benchmark offices. I think on a Sunday or a quarter. Yeah. Yeah. I'm in the right of home. So here I am. But I like the yellow. It's calming. Exactly. I've got boxers. I don't know exactly the story in the background. It's the right vibe. Yeah, exactly. Cool. Well, I mean a lot to talk about. You know, I think in the hype of AI world, I wanted to return for a little bit to the sort of more classic sort of consumer app and the fight to be, you know, the next Instagram and to look at some of your investments and to get a read on the consumer environment. And then certainly, you know, chain house is a great crypto company, maybe to talk about it at the end. But yeah, I mean, if you're, you know, a consumer startup right now chasing like mass engagement, is there investor money to raise or what would you say is sort of the mood given sort of the pessimistic period we've been experiencing? Yeah, it's, it is a really tough environment right now to build that type of company. And I think it's, you know, it's always been difficult, but the level of difficulty has been turned, you know, all the way on because right now anybody building something in consumer has to compete with the most addictive consumer format that we've ever had, which is short video. Right. And you see short video, you know, obviously with TikTok, but now it's everywhere. Like everybody has realized that if you want minutes on your platform and consumer engagement, you need short video. And so like if you think about it, like video is more engaging than photos. And the difference that you have with short video versus kind of the traditional social media that you would imagine, what you've seen traditionally on Instagram, Facebook, Snap, and then all these startups that are trying to kind of get into the space is that for, like let's exclude short video for now. Like, you know, I don't know about you, but when I go to Instagram now, you just realize like, okay, you watch your friends content, you consume your friends content because, you know, they're your friends and it's your way of staying connected and all those things. But really, can your friends content compete with like the global maxima of talent with short video? Like, it's really difficult. And so if, you know, if Instagram were to completely tune in the short term to engagement and time spent, then I would guess that it tilts very strongly to short video. It's just such an addictive. Which is a big reverse. I mean, for a while, it was like social is everything. I mean, that was sort of the Facebook view of the world. It's why Instagram was appealing to you. I mean, you sort of came into consumer with your early identification of Pinterest, right? Yes. How much did Pinterest try to lean into social or not? Or what was sort of the case study with Pinterest in terms of this sort of the social moment that they were competing with? Like when Pinterest started, I remember what was happening at the time was really that Twitter was ascendant. Like that's what social was at the time. It was really largely text. Even on Facebook, it was a very dominant text based platform. And so you had actually a lot of companies at the time, a lot of startups that saw Twitter's ascendants and then had different variations of Twitter or text based social networks. And then the second thing was that the idea of a list was always this small feature in a platform. Twitter, I think, did have lists back then, but who uses them? It's such a power user feature. And then I remember using Pinterest for the first time. And it was all images. It was a very different graph that they were building at the time than a social graph. Right. Do I care? It was very blogger based in the beginning. And so if I'm planning an event and I'm using Pinterest to kind of remember my best ideas of things that inspire me, am I going to follow my friends or am I going to follow, you know, a great event planner or a lifestyle blogger? Right. And so they had this kind of interesting thing where the content itself was very different than anything you saw elsewhere. The graph they were building was different, although certainly Facebook was a huge part of the follow graph that they did use to grow. And then the idea of the board was so different. Like the board is a list essentially. And so it was just a very different product. I still feel like it's a very different product. Like it is everywhere else you feel like you're always comparing yourself against other people. Like you kind of can't get away from that in most of social. Pinterest is this place where it really feels, even though there is a graph, it feels like it's just for you. Right. Like when I was experiment, I just like had a bunch of pictures of like pistachios in my brand of pictures. You know, I mean, mine is not totally like I look up stuff for my son all the time and it's I have bonds, I trees and snakes and all these crazy things, but it feels very different than any other social product. I mean, with the take off of short video, I mean, do you think the power is an obviously it's multi factor, but like there's the video format, just like the ability to power that. But a lot of the power to me feels like the sort of, you know, non network, the algorithm exactly, you know, you know, just like it gives you what you want. And because it's short, it's able to run the experiments at a scale that nothing else is. It's an incredible, you know, in the beginning, like generation one of social, what you saw was 100% based on who you follow or your friends, you know, the, you know, the explicit connections that you make as a user in the product. I actually built the algorithmic feed that was on Pinterest. I think we may have been one of the first and you just see when you use ML techniques to drive your recommendations, the relevance ranking and your feed that actually kind of constantly pruning your feed is a lot of work, like following someone unfollowing. Right. It's just a lot of work. And so if you can use ML to drive that, then you also get at this global maxima of all the content that's available, then ranking that content for each user individually. It's just an incredibly powerful thing. And so, yeah, it is your right. It's a combination of this, again, it's a format that is the most addictive that we've had. And then that algorithmic feed, which is why your friends content has such a hard time competing against the most talented people, you know, that you could otherwise. You invested in a makeup company that was trying to make a play. Super great. Yeah. Yeah. It's a beauty. Yeah. Is makeup not the word they want to use or it's beauty? Yeah, it's broader than it's broader than makeup. Sorry. But what's the state of that and what's the state sort of more broad? Like it seems with TikTok, there's a lot of like, you know, US investor envy about how great it is. The question is always, is there something to divine from it that you can extend to a new startup or is it that sort of TikTok has already captured, you know, whatever the value there is? Yeah. My question that I had for TikTok is, you know, commerce is something that like you look at all online retailers and they spend so much energy kind of optimizing the conversion funnels, making, you know, all the integrations and systems that you need in order to display the inventory that's actually available to make sure the pricing's correct and then to handle all the logistics, the pricing, all those things and to make that process as easy as possible so that you maximize the percentage of people who actually get all the way through payment to making the order, right? Which is really difficult. You know, it's when you have a platform like TikTok or any social product, you tend to have to be the, you know, lowest common denominator. It's really hard to build out that level of depth of functionality. Yeah. A consumer would want or need in order to make sure to get them to convert. The second thing is that I experienced this at least at Pinterest, which is that, you know, when you have, I think of commerce as like this very lean in thing to do. There's a number of examples I could give. Yelp is probably the best example where if you have a product that is not really oriented around an action you're taken, it's harder to get a consumer to go from this like lean back mode of going through videos as an example to this lean in mode of actually stopping themselves and completing an action, which is to buy something. It's really challenging. The experience of the Pinterest, I think Yelp, like you think about Yelp, I just, a product that couldn't have been better situated to take advantage of food delivery, of reservations. But it was always this mix of the things that were actionable and everything else. And so the hypothesis was that even though TikTok is so dominant in terms of time spent, if you could have a value proposition that was strong enough in commerce that you would be able to build a community outside of TikTok. Right. And that's still an open question. So that's Pop Shop, for example. Yes. For mobile, right? I don't know if you would use that phrase, but you have some shop owner sort of like on a video and people can like come in. What's the state of that model right now? So first of all, the product to me is still this incredible product. Like I invite everybody to take a look and you just feel an aliveness to it. The challenge that we've had to navigate is there was a little bit this question like you look at live shopping in China and it counts for 10% of all the e-commerce in China, which is one of those facts that is really astounding. But what's harder to kind of realize or you maybe underestimate the importance of this is that in China, you have a very mature supply chain infrastructure to handle the fulfillment of goods. And so you have Taobao as an example. Taobao is the largest platform. It's kind of, you know, could think of it simplistically as Amazon there where they partner then with these sellers, creators whom if you're a creator and you want to do live streaming, live shopping on Taobao, Taobao handles, you know, has the inventory and then they're pushing your deals all the time. Here are things to offer to sell. They handle the logistics. They provide unique value propositions like discounting or, you know, exclusive inventory. And so really what you need as a seller for a live shopping platform in China is just to be able to be amazing at selling and growing your community. And the challenge that, you know, we've had to navigate in the United States is that we don't have that same infrastructure. It's a lot of work. You get people hyped about your products and they're like, Oh man, I have to set up. Well, and so, you know, what you have is you have a situation where like, you know, you're finding the Venn diagram of like a seller who's motivated and like great, uncomfortable, great at getting from the camera, selling, building a community. You have to find somebody who has access to inventory that people would want and kind of go through the additional friction or, you know, feel that just the additional friction of going to a live shopping show versus just going to whatever website it could be. And then someone who is okay handling the logistics. So you know, you have a great sale on Pop Shop and then you've got to pick and pack a lot of boxes. Right. And then they were where Pop Shop leaned in the beginning where retailers and they were, you know, mom and pop, you know, Japan, LA. Okay. But these like very talented proprietors of shops in LA and elsewhere, when there was real product market fit there, but you know, those people, they are in the business already of doing that selling and then they can handle logistics. They know how to hire people to do that. But it is, it's just a harder equation here in the US than it is in China. Yeah. And in that category, I mean, what not was sort of the rival, right? I mean, that's tried to ride this sort of what? Well, car sort of Funko. Right. You know, kudos to them. Like what not, you know, what you see now is that their seller, the bar that they need for a seller is just much lower. Yeah. Right. Because most of the time you go on a what not show and it's just hands, you know, you see like somebody kind of put the like, okay, here's the card. Okay. They bid. And then okay, here's the next card. Here's the next card. And it's a very transactional, very like the relationship with the seller is almost fungible, right? Across the, you know, if you're buying cards, card breaking, whatever it is, it's a very fungible seller. But that wasn't the case for pop shop. That isn't the case for pop shop. Like it's very much a personality and that's a fit for the product they're selling. And so it's just a higher bar to be a seller. So we sort of got into this conversation immediately talking about TikTok because it looms so large and, you know, this sort of strategies compete with it. But then there was also this period of trying to find the new Instagram and sort of the belief that there would be another photo company. You know, obviously, I feel like Be Real was super hot, you know, and then you invested in paparazzi, which was the number one app at one point is now shut down. So I appreciate your willing, like, yeah, I mean, everybody would love to learn from the experience of that. Or what do you think was like this sort of hypothesis with like paparazzi and like, yeah, give us a little of this story of the journey there. Yeah. First of all, like Alex, the CEO is just one of those product founders that you love to work with. And I, you know, would make that same decision, you know, all over again. He's great. And he understands this kind of young generation of user incredibly well, just so deep, you know, all the ingredients that you look for. And I think the hypothesis, which also underlies Be Real and many other products is that, you know, you have this challenge in Instagram, which is the dominant platform for social, which is, you know, you feel constrained by Instagram because the bar just has gone so, so high to push something because you're like, Oh, it needs to be a top tier. It needs to be a top tier and you can't post too much because then you look like you're thirsty. So how do you have this plausible deniability of somebody, you know, posting content about you? And, you know, Be Real does it in a really clever mechanic of kind of saying, well, you don't have a choice. So you can't perfect it anyway. And so it kind of creates that plausible deniability to create content. And you know, I remember for paparazzi, it was a problem at some point that I had two pop companies I was working with. So at paparazzi, one of the experiences that I had, they were in a test flight for a while before we invested. And so they had a love use as I remember seeing the content of some of my friends. And so they, for the people who don't know the premise with paparazzi is that your friends populate your profile. So instead of you feeling like you're taking photos of yourself all the time, that actually your friends can take a photo and tag it to your profile. And so you get this profile of pictures. Have your little friend's stalkers all over the place. Yes. Yeah. And what I remember experiencing was going through paparazzi. And I would see way more content of some of my friends, like I remember Ryan Hoover was there where his girlfriend was taking photos of him and posting it to his paparazzi. And their photos he would never have taken of himself and posted. Never. But because his girlfriend was doing it, you know, there was that plausible deniability. And I got to have this view into his life that I wouldn't have otherwise had. And that felt really powerful to me. Right. Here this pitch, I'm like, Oh, I want this app. But then why does, yeah, why doesn't it make sense? You know, when we made the decision to kind of shut down the company and return the capital, I said to him, I was like, you know, one day this idea is like, I still believe, I still really believe that something like this could happen. And that's, you know, that core belief that the product is what drew, you know, my investment. Like, it does feel like the type of product that with the right iteration, the right calibration of the product could be a mass market product, you know, but the again, competition, it's like trying to start a new e-commerce website for books when you have Amazon, you know, it's just really hard. And it doesn't matter how incredible your product is. You're just competing. And so, you know, you're always competing against every substitute. And so even if like in a normal world, like you start here, like, and this is kind of user value at Instagram for like every minute, like the dopamine per minute that you get on Instagram or TikTok or whatever, like you have to start here in the beginning for a consumer social product and work your way up as you build more features and everything. And so it just is a really challenging task to not just get people interested in the beginning, which you see a lot of products grow. Everybody's willing to try things out, which is promising. But then to keep that interest alive into you fight for every minute of that user's life, you know, and that's really challenging. Right. I mean, people are excited about the idea of a new app or don't want to miss out on something cool, but then sustaining it beyond that sort of urge. And I think there is this, you know, people still desire authenticity. People still don't feel good about themselves when they spend too much time in these other products. Like, people still want to connect with their friends. Like, there's still all these things where you feel like I'm a big person, you know, some people have given up entirely. I'm not willing to give up yet. You know, I think just when you're about to give up is when you succeed, there is going to be someone, maybe it looks like a game in the beginning. Maybe it looks like AI companion that maybe it looks, you know, who knows what it looks like, but you feel like the fundamental architecture of these products is moving away from where they started. And does that create enough of an opportunity for a social product? It's TBD. It's funny that you say TBD. Yeah, like, there's an argument that like the super focused app where you just like ride that initial wave of enthusiasm and then try to like get out of it. Like, do you think that's like a replicable strategy or do you think we'll see a lot of these? We're seeing many variations. I think one of the things that people recognize and you see, I think this will be, you know, the challenge for anybody going after a consumer social is that you have to be at just for a tremendous scale in order to monetize via ads. You know, if you're at scale, it's an incredible business model. But to get there is really, you know, you have to have just tremendous scale. And so there is this other model, you see Discord being the dominant player in this monetization model, but then of course, you know, Tbh and, you know, NGL and all of it's kind of people who have been inspired by that mechanic is to charge users. And it's really, you know, there's a little bit of the leaky bucket. So it's one of those monetization strategies with, you know, NGL we saw where you eventually the cohorts run out, right? Like, you don't see a lot of companies trying to go after enduring businesses using that type of mechanic. You see a lot of people who recognize that there's a window of time where they can really monetize and then they take advantage of that window and then they move on to the next product and create a new window. And I think we'll see a lot of that. Yeah. It's almost like, I know I've never thought about it before, but I wonder to what extent it's almost like hyper casual games now, you know, where you recognize that some models acquire monetize and then move on to the next. Yeah. I mean, you know, just keeping the conversation on paparazzi, there's so many directions because all the, you know, there's so many like Venn diagrams of overlapping strategies and interesting points of connection. But just sticking to paparazzi, like wanting down, returning to investors. I mean, I'm sure you've got like a pretty good taste of, you know, the acquisition market right now for these apps that are going big. Like what's your read on, you know, the Facebook, Google's of the world and their interests in these companies and with the acquisition markets like all the public companies feel like they're in a defensive crouch right now to me. I mean, it's really hard. I think that companies are missing the boat. I would hope that this sentiment changes, but I understand that right now you're a public company, you know, you have way over hired over the last few years, way over hired and you've tried to fix that by doing layoffs, freezing, hiring. So in that climate, when people are coming to you and asking to, you know, add headcount and, you know, the leadership is having to say no, if not the opposite to then come and say we're going to make an acquisition of a team is politically really difficult. I think that's a mistake. I'm a huge believer. I wrote a blog post about this. I kind of think that everybody you hire in a company exists on this spectrum, which is between a mercenary and a founder and the mindset of those people are very different. Most of the traditional hiring channels that you have will skew towards, not all the way towards mercenary. Please don't take that as what I'm saying, but certainly not the 20% towards founder. And what's like a mercenary? It's more like I'll do my job, but it's a transactional relationship. You tell me what my job is and I will do that job. Like I care who my manager is, I care what my compensation is, and if someone else comes along and they offer me more money or a better manager or whatever it is, I'll self like optimize and make that decision. And it's a very rational kind of view. But it's a very kind of, you tend to see this more with people who work at big companies. It's kind of the training that you get in a big company is that you have to operate that way. And you want more people because more people on your team means you can get a promotion. It's that type of thing. Then you see the other side of the spectrum. And I experience this a lot of Pinterest, which is that the people who act more like founders, the people I call them the mitochondria in a company, the mitochondria is like the energy producing part of your cell. Like the people who act more like founders, they come to the company because it's like they really care about its purpose and its mission. In a way, the job title or job like here is your job is irrelevant because someone who's oriented, like, you know, their constitution is like a founder will just want to see what does the company need and they will do it. They break through process. That sense of urgency is just very different. And through good times and bad, they power a company. And I think that a lot of these companies, you know, I would take like for big jobs in a company, you want a founder, you want somebody who is oriented that way to take on that project or whatever it is to do the zero to one for a project. And so getting founders in these companies. I did a bunch of talent acquisitions when I was at Pinterest and you just saw how impactful founders were in the company giving feedback to the CEO, to the founders. And so I think it's a mistake to kind of see it as one head count, one head count. And I hope this changes. Lenny who has a great product podcast was interviewing one of the creators of Square Cash who I think hires a bunch of founders for his company. And on the one hand, founders are ferocious, you get a lot done, they take a lot of ownership. On the other hand, they leave two years later to start their own company and they'll tell you everything you're doing wrong. It's a personality type that can power a super successful company. And you want to mix. If you don't have people like that, then who's going to tell you the truth? When I was at Pinterest, I organized a quarterly dinner of all the founders that we had there that we had acquired with Ben. And it was just like you could tell that that dinner was really important because it was a breath of fresh air every time it happened. And by the way, yes, there's the kind of archetype of the person who's just waiting for their vesting to end and then they're going to move on. But there are many examples of founders going the distance in these companies. And that's up to the company to retain them. We Hunter Walk, and we were talking about this a little bit, wrote a blog post basically arguing that companies aren't acquiring. They'll just try and get the team. There's a piece of it where it's like, okay, maybe they'll go after a team once it's disbanded. But it used to be the case where you'd have this pre-bundled team and you'd sell them all to the company. I mean, I tweeted, I don't know if you can talk about it, but with paparazzi, it's like the company shut down return cash, but then a lot of the team goes to Instagram. Or what's the distinction there between selling and sort of them picking it up afterwards? Well, it's a premium. And so when you do a talent acquisition, you typically have a premium that the company pays to get the team together that then goes to the cap table. Typically, it's to repay some or all of the preferred. And when you hire, you don't need to do that. When there's not competition, you don't have to do it the expensive way. So having a talent acquisition is a function of there being competition for the most talented founding teams. I have heard, this is a tidbit. I've heard the Riplane, Parker Conrad's company is trying to get all the stray founders he can. Totally different category, but it'll be interesting to watch which companies are trying to just like pick up stray founders now as companies. And I think it's really smart. I mean, what they're trying to do at Riplane is there's a lot of SKUs. And so you want people to feel accountability to own the entire product that you're selling. And also not slow down as much by a company that grows a lot. I can only imagine what Riplane has to be a little decentralized in order to do it. To grow as quickly as it does. And what better archetype to handle that unstructuredness and pace than founders? Yeah, returning to an earlier point where we were talking about the growth challenge and getting to, you know, you get people on the app, but then competing with the dopamine of an Instagram or a TikTok. Like how do you do it? Part of the answer that TikTok initially found and now we're seen with Team U, you know, sort of this what's being seen, I guess is like the Chinese app growth strategy in the United States is just like spend hundreds of millions of dollars. Like, would you ever invest in a company that like came to you with the like we're going to blitz scale and consumer? It's funny. I think of like Silicon Valley as being famous for like, we'll burn a lot of money to try to grow something. But it seems like Chinese app developers are beating us on high spend consumer app strategy. Or what do you mean by that? I mean, never say never. I think, you know, be real did this, but if you do it after you have a retentive product. Right. I think that's actually what TikTok did is that they recognized finally that they had a product that was retentive. They had estimates on how well they'd be able to monetize that user base. And then they were able to kind of say, hey, we can spend up to one XL TV basically in order to grow quickly. And then bet that would have these kind of flywheel knock on effects. Be real figured out and spent a lot of money on this, which was doing campus recruiting. So they had a lot of people in these campuses that would then get other friends on the campus to use be real. But they did that after they realized that they had a retentive product. And so I think there is something to be said, but you really want to make sure you have a retentive product. And then I would really feel a lot more comfortable with the idea if you also kind of had a sense for what the effective monetization would be and use that to pencil out what you can spend. And obviously there's lots of intellectual games people can play with lifetime value where you're trying to guess what these customers are going to be worth. Yes. Yes. Always tricky judgment decisions made. The next frontier, I mean, you sort of alluded to this a little bit earlier, but it feels like the next frontier in consumer could come in this generative AI wave. Or like in some ways, it seems like mid journey is sort of a quiet, like dominant sort of consumer app built on top of discord, sort of mostly self-funded or entirely self-funded. I don't know, but I don't know what you read on the consumer landscape in generative AI AI. And yeah, is that a space you're actively hunting in at the moment? No question. It'd be hard. You know, look, usually consumer waves happen after a technology catalyst. And we have been using the same, I always think of it as like a palette. You know, you have these paint colors you can use. And so we're all for a long time, we've been using, okay, the same brand and the same green to like try to create something different. And now you have what's possible with these large language models. And so it does create a new opportunity. And you see, I mean, I would add chat GPT to your group like it's actually making money, which is making, well, making revenue. Yes. But I'm not to take anything away. Just generating revenue. You know, and what's fascinating to me about mid journey, I mean, it's just an unbelievable story is also one of the things that I feel is missing from the chat GPT experience personally is that we don't get to learn from each other on, oh, you figured out the best prompt to make some kind of requests to to chat GPT. And so I'm learning all the time from Twitter and other places what the tricks of the trade are, like what is possible with chat GPT. But then you have because of, you know, mid journey launch within discord, you have this experience all the time where you see people tweaking their prompt, changing it. You see what's possible, a new person can come in and start using mid journey and see very quickly what, you know, the tricks of the trade are. And so it's a very cool group learning type experience. And obviously they've progressed that model to like such, you know, such a crazy place. I mean, relative to opening eyes own effort, Dali, like you do the same query in both the same prompt, I should say in both of those experiences, the outcome is just way better in mid journey, both because they've been training the underlying model. But then also you actually then would really write a better prompt with maturity than you would with Dali because you've got to learn from everybody else doing it who's already at the expert level. Do you have a bet on this? I mean, one of the challenges of talking to you is you make early stage investments and then don't always announce them. So I don't always know. I try not to announce them. Not only. Yeah. Have you made an AI investment yet? Or not me personally. What are the explicit generative AI company? Of course, as you might imagine, every company that we have. All of them are pivoting. Not pivoting, but figuring out how to leverage this type of technology because you'd be, you know, it would be dangerous and foolish not to. I'm very excited or curious about other ways in which these technologies will, you know, change consumer social, like, you know, you certainly see these companion apps or role playing conversational bots. You could call them. Right. That there definitely feels like there's something there to me. I mean, it's hard not to see her and think about that. But you know, we do have a challenge in our country. I mean, probably globally of people being lonely. And if you have a bot that acts as your friend, that remembers you, that is, it makes you feel accepted, that makes you feel good about yourself that you can talk to and they know how to make you feel better, that's a powerful thing. And then you could only imagine how that, you know, what's possible, just keeps on expanding in that relationship. You know, character, probably with the most high profile and well-funded. I mean, how a benchmark do you invest in that type of company when right now they're building, you know, foundation models and they're just like huge rounds? Yeah, I don't know. How do you make a play? I mean, clearly, I guess you haven't personally made an investment yet. So I'm sure the question is the prices are high. How do you navigate the high prices in an area that you think is really exciting? Look, I think there's two things here, which is there's one with character. You'd have to make a bet that you believe that value will accrue to the foundation model. Yeah. Then there's a separate bet of the, you know, the experience itself, the first party product that they are building. The foundation model rounds have gotten very pricey and, you know, at benchmark, we look for exceptions and we are affirmed that believes in capital efficiency. And so it's just not a fit that that particular multi-hundred million dollar round to start, you know, I, were there questions that we could talk through on where does value accrue? I think it's an open question. And then there's a structure of the round itself. If you had a company that came to us, I mean, paparazzi was not a cheap deal, right? Like, we believe you got to play the game on the field. And if there's the right company, we'll play that game, but it would be less likely for us to do a multi-hundred million dollar round. That's for sure. Right. You sort of tease this and all I have to ask, like, where do you think the value will accrue? Or that's what everybody is debating right now. Do you have a point in view? I mean, I'm certainly biased towards whomever owns the user. Yeah. And so I'm spending a lot of time, you know, both in consumer, but then also in vertical applications of AI, to just see, you know, a company that creates an application where, you know, a large language model is driving some of the value proposition, that feels like a very valuable place, you know, then you're owning the user, you can have feedback loops, you can create other accruing assets, that feels like a really valuable place to be. And then of course on the consumer side as well, you know, you want to encourage as much competition though, at the foundation model level. And the wonderful thing that you're seeing more and more now is of course what's happening in the open source world. Right. You're like, if the foundation models can be commoditized and you can prevent the foundation models from stealing all the learnings from this vertical application and the vertical application has a consumer relationship, then you have a path. Or a user relationship. Yeah, user. Yeah. Yeah. It's interesting. Do you see any examples of that or like, are there companies? Oh, there's so there are a lot of examples. So certainly like you see what's happening with all the coding co-pilot type experiences, GitHub, Replib, you know, if there's any others, like that is an experience where you're using the underlying model. And I think that there's some even that use other foundation models as the source. And so you create an experience there. There's vertical applications. You know, you see so many things happening in law right now as an example, or gaming. Yes. Yeah. What's law? What lawyers do? A law. And so you have kind of two, it's interesting to me. There's like two modalities of end user value that you see, which is one modality is the co-pilot metaphor, right? Yeah. So you're already doing a job and you make them more productive at that job in ways that are only possible now, thanks to large language models. The second modality is you actually take a work product that would happen. And it's not about a co-pilot. It's actually taking that entire work product and doing it, what you call it, pilot. And that I think both modalities are super interesting to me. It's sort of user in the loop, user out of the loop sort of situation. Yeah. I mean, so I'll give a specific example. There's a company called Evenup in the personal injury space, really incredible team, remarkable what they've done. And if you are a personal injury lawyer and you have a case, what you end up doing is writing a demand letter is called. And the demand letter is like a summary of your case and then makes a demand to the insurer basically for a settlement amount. That work product, the demand letter itself, usually outsourced somewhere. The lawyer does it between all everything else they're having to do or they have junior people are a paralegal right, the demand letter. It's exactly the type of product that can be built, created with a lot, you know, a combination of other ML techniques, OCR as an example, and a large language model to summarize everything. And so here you are taking something that otherwise would have been a work product in the company and the law firm and making it much cheaper, faster, having, you know, a quality assurance bar that is, you know, stable and you don't have to manage people. You don't have to train people. It's just a product. And so that idea, I think we're going to see in a lot of different verticals and I'm super excited to kind of find companies like that. And then you also have the whole idea of taking people who are already doing their job and how do you take the workflows that they already have and either, you know, replace them with something better that gives them superpowers or assists them in the workflows that they already have. And so you're going to see a generation of really interesting companies that look like that too. Yeah. I'm going to ask you about crypto in a second, but I want to talk, you know, chain analysis, you know, is one of your, certainly your most highly valued, right? One of the most successful investments today. I think they had a great New York Times profile the other day. I mean, to me, it's an interesting company in that it's like, it's almost like a crypto skeptical crypto company or it's like, oh, there's a lot of sketchy stuff going on in crypto. It's clearly taking off, but it's going to help in, you know, the criminal justice system. It's going to investigate. It's going to track all these things. But is there like a metaphor company or a comparable company for AI where it's like, okay, there's a lot of hype. These like foundation models are like wild and maybe they're not going to, you know, be that differentiated. Is there a way to like see all the mania and AI and still make an investment like chain analysis or other companies like that where it's like, no matter how this nets out, I still think, you know, this is going to be a good company or yeah, I can find one in AI. I think there's absolutely going to be opportunities. You know, I think a little bit, it's just, it's more, I would think of it more as like a picks and shovels type investment, which is like, okay, you know, when we made the investment in chain analysis, I was very bullish on crypto. What I struggled with at the time was that there was a lot of companies that were raising money via ICOs on Ethereum and to build out the infrastructure without really knowing what the use cases were going to be. And that felt like a really challenging way to invest to me. And our deep belief at benchmark is that our job is not to predict the future, but to try as best we can to see the present clearly. And so when you think about what were the use cases that exist in crypto, you know, you had store of value with Bitcoin, you had fundraising with Ethereum, and then you had criminal activity. And of course, we weren't going to invest in like the next Silk Road, but I got lucky about hearing, you know, I met with Katie Han when she was still at the DOJ. She mentioned that she had used chain analysis as the software tool in order to do the investigations that she pursued there, Silk Road, Mt. Gox. And then I met with Michael and realized that chain analysis was both this thing that you would need to do investigations in these different blockchains, but it also was an enabling force. If you had something like chain analysis, then regulators would be more inclined to use something like chain analysis and like have a scalpel to get the bad activity out, then to use like the atomic bomb or regulation to kill everything. And so it felt like chain analysis was actually a bullish, necessary foundational piece of this ecosystem in order to enable its growth. In AI, I think there's a lot of parallels. And one of the things that I think about a lot certainly is, you know, how there's a lot of questions right now about, you know, you alluded to it, how do people use your data? And so that's both kind of a B2B setting, you know, if you have stuff that gets crawled on the web, like you kind of have this situation right now where anything is, if it's in the public domain, it's open game for any of these models. A lot of people are realizing that and not too happy about it. So how does that work out? And then you have this question that really is a pretty important question to me about truth and deep fakes. There's, you know, we know that these large language models hallucinate and there's going to be a lot of things that emerge as a result of this, especially as a lot of the large language model content creates content on the web and then fed back into the model. Like, we're just going to have a lot of things to figure out. And so I think there's going to be all types of opportunities along this. On the truth, Brian, I mean, I was very excited about like perplexity, which is trying to footnote all its results. I will say, you know, bad information in bad information out, you know, there are all these websites that try to guess like, you know, biographical facts about people that have your like average height, you know, face on just like, we don't know, we'll put something up there. And like, I do find perplexity inhales some of that stuff. So like the footnotes, I mean, you can, you know, go to the footnote and then learn that it's not a very good source. But the truth problem for these systems is going to be so complicated. I thought perplexity's product experience was like the way that they saw the hallucination problem. I thought it was really elegant. It reminds me a little bit of like how waymo in the beginning, they had no steering wheel in the car. It was like the idea. And then they realized that the technology wasn't actually ready and consumers weren't ready for the experience of having no steering wheel there. And you have kind of the new generation crews being just incredible what they've done where there's a steering wheel. And the citations and perplexity felt to me like a steering wheel. Like we're not yet ready for an experience that has no citations, no ability to really fact check. And we're doing people a disservice, I think by not having that because the risk is you read answers and chat, I have this experience all the time. And it sounds right because that's what it's made to do. But that's an outcome of its design is architecture. But you've got to train people to check and people aren't going to do that. Right. No, it makes you so lazy. It's like, oh, of course. If I see my job is just outputting text, it's like, oh, you know, chat GPT is so good at that. But, you know, when it's outputting, you know, coherent thoughts and argue in it, you know, true. Yeah. And so the kudos to the perplexity team, I thought it was well done. And of course, Bing did something similar in their own experience. Okay. So I said, I'd ask you about crypto. You know, like it's certainly still the winter there. Are you looking at new crypto investments or I'm curious what you think is interesting right now? Yeah. And, you know, I still believe NFTs are, you know, I was talking about that color palette is a new color. And I still, you know, we have a couple of investments. So rare, right? Is it so rare as one of them? And then I have a stealth company that I haven't announced in the gaming world. It's an incredibly talented team that I'm very excited about. Hopefully we'll be launching soon. And when you think about like an NFT, like there's something there about man, crypto really did itself no favors over the last few years. Like crypto is a bad word now. Yeah. And if these are a bad word and it's really hard to train consumers to trust something again. You know, once a consumer has a first impression, it's like much easier to teach a user a first impression than to rewrite that first impression. And so I think there's like two things that have to change in crypto. One is, you know, this kind of a new culture of building. I think that the culture of building in crypto was completely dominated by speculation. Right. And well, I shouldn't say completely 90 something percent dominated by speculation. Yeah. Okay. And there are certainly incredible founders that have been building in crypto. And so I feel bad kind of pulling them all in, but it has been, I mean, you know, just what people what absorbed all the energy and emotion and venture dollars was speculation right under the day. People described this differently, but it really was speculation. And the create a culture of building that I think actually started with ICOs. And so we've got to change that culture. I think embrace actually web two principles. I think there was always this like antagonism that you'd see play out on Twitter of like web two versus web three. I think web three builders have to embrace web two builders. It's the only way to build something enduring. And so you've got this culture that has to change. And then we're going to have to have a branding change, you know, if crypto is a bad word, if NFTs are a little bit of a dicey acronym, what's the new package that I would think maybe doesn't you don't even know it's crypto when you're engaging with it in the product that will get some of the benefits, which I think are real with this, you know, what's possible on blockchains. So you're saying crypto is not at the front of the consumer's mind. It's like a useful technology that uses the power, but it's not the branding is not really around. It's that I think that's the most likely instantiation that we'll see that it's moved to the background. We will have some crypto elements. It's hard to get away from it entirely. And I wouldn't say you need to, but it's not all about. And there was just kind of this belief like, Oh, well, crypto, it's, you know, it's obvious the benefits. That's not true. There was this whole set of arguments that the sort of financial speculation was powering the adoption. So that sort of justified. Yeah, I just don't believe. Yeah. And by the way, regulation made that argument also a lot harder. Unpalatable. Yeah, it's a term sheet, you know, just had an article about Bill Gurley, I think moving to Texas, obviously even like back and forth. I mean, you know, people are always fascinated by the benchmark partnership, you know, this sort of partnership of equals sort of keeping the fun, small event sort of in the crazy times. I'm just, I don't know. I'm curious to hear how the firm is like evolving or like, what can you tell us from inside benchmark right now? I mean, we've always had a pretty simple idea, which is that, you know, there's this creative destruction. This Peter says that has to happen with the partnership where there is, you know, once you start, there's no training wheels. You're thrown into the deep end, you're an equal partner. And you're expected to be a hundred percent until the minute that you retire, you raise your hand and say you're not. And when you have an equal partnership, it kind of pushes you in the direction of just recognizing, you know, as Bill said in that interview, the hustle may not be in you anymore. And if you feel that way, then the model is, you know, as was set up by the founders, such that it's time that to raise your hand and move on. Of course, it should be said that like all of our retired partners are not very good at stopping working. They're not. It's like an affliction. I think, you know, the reason they're here in the first place was because, you know, the curiosity and competitiveness and, you know, drive for learning and relevance in the mix. And that never leaves you. And so they're all our biggest, you know, there are significant portion of our LP base. Right. And you're still there on Mondays. And I'm texting all of them all the time as we look at companies and kind of think through. What is the partnership right now? Peter is sort of the old guard one. They're still there. He's the one who's been, yes, says Peter, Eric Bishria, me, Chaithen, Pudokuta, and the Miles Grimscha. Basically for like four or five are the new generation, basically. Yes. Yeah. Well, it's hard to say new generation because it's sort of famous evolving. Right. You know, it's always it's like this, you know, it's a living organism. And you know, if you ask us internally, we'd say there's been several generations before that. I think there was an information story or like, you know, was there debate about sizing up the fund when things were getting really manic in 2021 or how do you feel about, yeah, the sort of keeping the fun size the same? I mean, we have debates about every part of benchmark structure all the time. There's no resting on laurels. There's no assuming that the model that we have is the model that we should keep. We say whenever a new GP joins the partnership that they're a founder of benchmark. We always think of ourselves as co-founders at benchmark, trying to think through, you know, not to rest our laurels to be skeptical, to be paranoid. I'm probably I may be the most paranoid one. I give it to my New Yorker tendencies. And so certainly when you saw everything happening in 2021, our model had both never felt more different and more differentiated, but it felt on times uncomfortable because everybody was investing a lot of capital. Evaluations were much higher. Rounds were happening much faster. Burns were going up and it felt like a euphoria. It didn't feel as much like what we were built for. But then at the same time, as I said before, you've got to play the game on the field. And so you have to ask yourself, should we do something different? We have debates all the time and still find ourselves coming back to the core that is benchmark, which is being that first board member. Usually that's the series A, being, you know, no more than six GPs at any one time, it being an equal partnership where we respect each other and feel affection for each other and love being together, and then to find the best founders of our generation. Just focus on that at the early stage and say no to everything else. Great. Sarah, thank you so much for coming on the podcast. It's good to see you. Hey, that's our episode. I'm Eric Newcomer. Thanks so much to Sarah Tovall for coming on the show. Shout out to Tommy Haren, our audio editor. Riley Cansell, my chief of staff, Jung Chomsky, for the wonderful music. This has been the Newcomer podcast. Follow us on Substack at Newcomer.co. See you next week. Goodbye. Goodbye. Goodbye. Goodbye. Goodbye. .