8. Sasha Kaletsky, Creator Ventures

What creators really like spending their time and money on in terms of software is creative tools. Tools that can help them create content better, faster, easier, cheaper, whatever. And Beehive was a really good example of that. Then gradually it just became clear that your friends were less entertaining than people whose entire job it was, was to be entertaining. And by the way, there's not just one person who's entertaining you. There's an entire universe of everyone in the world trying to make the most entertaining possible content for you and then it's choosing the best. So how's your friends ever gonna compete with that? More and more AI-focused on the one hand. And so there'll be more content that's generated by AI. However, I think what's an underdiscussed trend, which I think will happen, is the exact opposite.

Hey, my name is Michelangelo. I'm an investor at Stride and today I'm talking to Sasha Kalinsky. He's the founder of Creator Ventures, a $20 million seed fund focused on consumer internet companies that he co-founded with his cousin Casper Lee, influencer with over 50 million followers across social channels. Today we're going to do a deep dive into consumer investing and the impact that AI is going to have in the space. As always, thank you so much to the sponsor of the show, Citizen of Tajikistan. They're an underground community of thinkers and builders, and you can go access their Discord through the link in description. Enjoy. Let's start from death and rebirth. Yeah. What's the status of creator economy? What's happened in the past few years?

Well, so this is all linked to a series of blog posts I wrote like halfway through last year. Yeah, that basically started with this idea that social media, traditional social media as we know it, was moving from follower platforms to like content platforms. Follower platforms was Twitter back in the day. You and your friends would all follow each other, talk about the fact you had eggs and bacon for breakfast, and then everybody would not really give a shit, but they would be somewhat entertained by it because it's the first time you're seeing your friends post on social media, and they would, you know, like it and engage and everything like that. People would follow their friends. And then gradually over time, those follower platforms turned into content platforms where people went on not just to be entertained by their friends, But they got entertained by professional entertainers who were obviously more entertaining than their friends, or more knowledgeable, or more whatever people wanted to follow.

Yeah.

And so that basically led to this what's called like an unbundling of the social and the entertainment, which were two things that were together in old-school social networks. They started to get unbundled where a platform like TikTok became— and Instagram Reels and YouTube Shorts and everything like that became became super, super focused on purely just the entertainment. Yeah. And it worried a lot less about the social.

I must say though that TikTok really does try to drive those social dynamics. So there is a constant like, add to friends, you know, share with friends. Or so I feel like maybe they didn't necessarily engineer it to go only entertainment way, they just failed at the social way. Is that fair?

Yeah, they failed, but it's also, it's probably a better business. the entertainment side.

Okay.

You know, the business model just makes a lot more sense. You can engage people, you know, that people want to socialize for only so long, you know, maybe 30 minutes, 40 minutes. They want to see what their friends are up to, all the weddings and engagements, things like that, all the funny stuff their friend says. But people can be entertained for 6 hours straight. You know, you can scroll random stuff for so much longer than you can learn about what your friends are up to. And so the business model just makes more sense. the entertainment platform.

Ideally, if you're able to get entertained by your friends, that's the sweet spot.

But that was the sweet spot, yeah. But then people— then gradually it just became clear that your friends were less entertaining than people whose entire job it was, was to be entertaining. And by the way, there's not just one person who's entertaining you. There's an entire universe of everyone in the world trying to make the most entertaining possible content for you, and then it's choosing the best. So how's your friends ever gonna compete with that in terms of entertainment?

I guess, I guess, yeah, it's true. I wonder whether the solution is finding the in-between, which is what Instagram today probably is kind of leaning towards, that you have the massive entertainment aspect and then the social. Yeah, you know, if you can merge the two of them, I think that's possible.

So that's, that's exactly, and that's where we've been on Instagram for like the last, basically since it was founded. It started being super social. Your friends would take photos themselves on holiday or whatever they had for dinner. And then gradually it became more and more entertainment. And now it really is significantly more entertainment than it used to be. It's creeping, creeping, creeping slowly towards being more of an entertainment, a content platform. And so what does that mean for creators? So the creator economy was built in this world in kind of late 2010s or mid-2010s when the follower platform and the content platform were the same thing, kind of like what you just said about Instagram. where you'd have your friend's post and right next to it would be an influencer's post. And so as a user, you're seeing those 2 things together and you're sort of blending them together in your head.

You—

these social networks were a great place to become part of somebody's community, much like a friend, in a kind of parasocial type way. But then gradually, as these platforms are becoming more and more about entertainment and less and less about keeping up with your friends, they're instead of showing you the same creator each time that you follow, they're showing you random creators you've never seen before who just post something super viral. And so your relationship as a user with those creators becomes a lot more transactional and it's more like, entertain me. And your addiction as a user is no longer to following what's going on with these creators and how are they doing and are they— what are their friends and what do they have, you know, what do they see in the cinema? And it becomes your addiction becomes to the platform. you wanna know what the next video is.

Interesting, so the content marketplace, the platform becomes the main value capturing thing.

Exactly.

Which is why, hence, it's a better business model, because it has more of the stack. It used to be the creators have the power, now it's the platform, so the algorithmic suggestion of what value do you wanna tap into right now, that if you have 100 people to follow, it's likely not going to happen. If you have the whole world, you're much more effectively going to provide that.

100%.

Interesting. And so, but then does TikTok successfully tap into the value capture? Like, are they able to actually compete from a monetization standpoint compared to the social platforms?

So, so yeah, I mean, it's a great business model, TikTok, and Instagram is increasingly becoming an even better business model as people spend more time on it to be entertained. And so yeah, they're all great business models now, these entertainment platforms. But what's a really interesting trend now, because I don't want to be too negative about the present and the future of the creator economy, because there is this rebirth happening, which is something we've talked about before, which is, you know, people still have that itch to see, to get deeper relationships with creators like they used to. They used to— people used to watch 11-minute, 12-minute YouTube videos about what their creators' lives were doing and what they're up to. That hole is being filled now equally importantly in terms of a trend by podcasts, newsletters, things like that where you can truly be part of somebody's community. But it's just not happening on traditional social media anymore. It's just a new vertical that's coming to replace it.

I wonder, there is From a monetization standpoint, the relationship, if you have a deeper relationship, that's probably much more valuable actually, I think, in terms of like longer views actually equate to higher ad spend, I think.

Yeah, yeah, absolutely. So that's exactly right. So that's why it's a crisis for your average TikTok, Instagram creator who doesn't have a newsletter, doesn't have a podcast, and just makes content for social media. they have to basically optimize for completely different things. They have to now optimize for going super viral, saying super controversial shit, being super entertaining in a short amount of time to get viral. And that can help, but sometimes maybe they go do a viral video with a brand in it and they get paid. But yeah, it's much harder for them.

Sorry, and as a rebirth, you said it's a whole new category. Can you explain that better? I don't think I got it fully.

Yeah, so the rebirth is the social— the new platforms that have taken the place of the old social platforms. And so, so the way, the way I wrote about it was the social platforms became more and more like entertainment platforms, and so they kind of moved up the curve of becoming more entertainment-based. That left a gap, and now we can fill a whole category. What are some examples of these Well, so the first simple examples are podcast platforms, even Spotify. You know, people can really seriously engage with creators deeply.

What about BeReal?

Well, exactly. BeReal is a great example of that. There was a yearning there for that to happen. And, you know, BeReal, the retention wasn't there. And, you know, it's maybe too early for a full postmortem, but, but effectively, like, that to me proved that there was a real itch to be scratched by a platform that was like that. And so yeah, I think there's definitely a space for a properly social platform, whether it's BeReal or another, in the next 10 years.

Unless TikTok is successfully going to drive that socialization, although they may not be interested into it and they may just double down on the entertainment.

I think that's right. I think it's going to be very hard for them to make a transition from being more entertainment-based to be more social play because the social posts just won't do as well. You know, TikTok's algorithm is about optimizing for retention, for time on platform, for usage, all that stuff. And most of your friends' posts are going to be boring, especially compared to this entertainment. So it won't fit the metrics as well.

And so what kind of metrics are we going to see from this new gap, like the new category of apps that are going to fill this new social layer where the connections are again, the most important thing. Are we going to see shorter sessions, do you think? I know it's hard because we're talking about impractical, but no, probably longer sessions because it's about the depth of connections. But then I don't know how much you can be updated by your friends.

Yeah, so I think there's probably 2 separate things happening. One of them is that there's one gap, which is for people to follow on what's happening with their friends. That's the BeReal gap. And that's the one that BeReal tried to go after and others will also. That one is going to be all about super short time in-app. Well, it might be about super short time in-app, just seeing quickly what they're up to. They're not trying to entertain me. I just want to know what's up. Okay, click, click, click, done. And that's the friend gap. And then there's the creator deepened relationship gap, which may go the exact opposite way, which is like you're doing hour-and-a-half recordings and deep relationship-building newsletters, which you craft every day and spend effort into that goes into everybody's inboxes. That kind of stuff. So it's almost like Barbell.

Was that, was that the part of your thesis in investing in Beehive?

Yeah, definitely, definitely. That's definitely part of it. There's a huge gap for like owning, you know, owning your audience in the creator economy.

When did you invest first?

We first invested at the end of '21 in their first, their first round. That was before they launched. Wow. Yeah, we did that because, you know, we've been looking at creator tools for a while. And we've always been, you know, my co-founder Casper, you know, we're cousins. I've known him for a long time. We've always kind of talked about business stuff. And he's really smart on like what works for creators and what doesn't. And he and I have always agreed, and it was mostly from his initial initiative, that a lot of tools that serve creators don't really have that much of a pain point. And the kind of view we started to form was that what creators really like spending their time and money on in terms of software is creative tools, tools that can help them create content better, faster, easier, cheaper, whatever. And Beehive was a really good example of that.

Of a really good tool that was enabling them. What were the main ways that you've seen Beehive Help creators create better content?

Yeah, so I mean, it's, it's, it's a creation platform. So everything from the newsletter editor itself to the formatting that you can optimize— they've got a referral program which helps other people spread the word. It's, it's a fun thing to use. And so it's like when we made the investment, one of the things we talked about was the joy that it elicited. And it's like what I used to say was that when people would receive their monthly invoice from paying Beehive, $40 a month, $100 a month, whatever it was, they'd feel good. They'd be like, okay, I'm glad I'm paying that. Which is not what you feel when you get your accounting invoice or your, you know, developer, you know, whatever boring stuff invoice or your garage that you're renting to film the content. But it's the bit that they really enjoy. And that's why we were super excited about investing in Beehive before they launched.

Interesting.

And I'm sorry, and also very importantly, the team. I mean, just obviously they're super, super strong. We tend to love teams that have done the exact thing that they're trying to do their startup around before. And I've got a whole series of reasons for why domain-specific founders, in my opinion, are much stronger founders for this 2023 era than they were in 2010.

Tell me about that.

So, so in the kind of first post-2000 age of startups that kind of we we kind of grew up following. Well, well, that maybe I'm not.

But tell me.

But yeah, so so no, were you not into startups when you were five?

I mean, until six years ago, this is a true story. I didn't know what a startup was. Wow.

Yeah, we should talk about that sometime. Maybe have. But um. So, so, so yeah, that first era of startups from the early 2000s onwards, like, setting up a company was really fucking hard. There was like, you had to find a way to get the servers in place, you had to find a way to hire engineers, you had to like write your own— a lot of custom code to make things happen. Basically, you had to be just a very domain-general genius to start a company, right? You need to be able to solve stuff that was coming at you from every direction, and you had to plan it all yourself. Then in about— in the 2010s, somewhere in the 2010s, basically AWS was built. People didn't have to host their own servers. Engineer— there were engineer hiring marketplaces, Y Combinator. There was networks formed that that could basically teach people all the general stuff about building a startup. And so that made building startups a lot easier, maybe 100 times easier in terms of a general way. And so what used to be so hard building a startup so that you needed to be a super general person to do all this stuff, it became less like, hey, who's the most general genius who can find out, figure out a way to build stuff? It's more like, who's the more specific genius who can build exactly that startup in exactly that space, because there's going to be 20 other people, all of whom are building the same thing. So, who's going to be the best of them? So, it's basically become a lot more competitive, which— and then if you look at the decacorns that have come from the last 5, 10 years, it's like Rippling and even Ramp, even, I think even Brex, Fair, like every single one. It's actually quite incredible. Every single one comes, and obviously we hope Beehive one day, every one comes from a founding team that deeply knew their space already going in. And so I think the era of the domain-general founder is over and the era of the domain-specific founder has begun.

This is interesting because I wonder whether it's a generalizing statement for every single industry, or I wonder whether there is some industry that are still on their way there. I guess for consumer offerings, yes, it does feel that way. I there is obviously the question of AI, which is a new added piece to the stack, and so we can see right now to build a really defensible offering with AI. Yeah, it seems as though you need to have that AI expertise so that you can either train your own models or take some open source models and just from the ground up improve it with. Your own fine-tuning and do your own thing, it feels like a generalist founder is not necessarily the most adequate person. So, but if you're applying that to another specific area, then you may want to have a generalist actually that knows what they're doing there. So I wonder whether you're thinking about it as a 10-year horizon type of process that we still have a few parts of the stack, and then are not new parts of the stack going to arise? are you not always going to need someone that can adapt to the technical improvements that we're seeing?

Yeah, very possibly. And that, you know, I definitely— and I think, you know, you have these exploratory moments, startup exploratory moments, where you have to work out what the areas of interest are and what are the new hills to be climbed. And then you have the execution moments where it's like, it's pretty clear what needs to be done, you need to do it.

Interesting.

I think we've been in an execution period in the last 10 years or so, or maybe last 5 years. I think that execution period will continue. It's possible that with a new, you know, a whole new domain, then maybe with AI and stuff, maybe it'll change.

Interesting. So basically the pendulum of whether you want a generalist or domain-specific founders shifts depending on whether we're going through a platform shift or whether we have already established a platform and now it's about the specific offering. You don't need to figure out the platform, the platform's figured out. We can just work on the application layer.

I think that's right, because I think, or at least it's definitely a big determinant, because it depends on what enabling technologies have been built for startups to make their lives easier. You know, the Shopifys and AWS and things like that. And we're in a VC world where, as you know very well, you know, those tools get funded very fast. Those enabling, those startup-enabling technologies are like the ideal venture investments. So those are almost happening sometimes before the businesses that they'll actually sell to.

Yeah, which is why hype is a thing.

Yeah, exactly.

In the sense that you— it's not there yet, but we already know that there is— it's not the consumer buy-in is not there yet, but we already know that there is a need because the developers or the people that are building are feeling that general struggle that they need to figure out so many things. And so you got the tool, but it's a derivative of a derivative, but you still— you don't have that consumer buy-in. So we don't really know. Although it seems likely that it's going to happen with AI. I mean, you do feel that there is some things that are out of the blue, just much better compared to what we used to have.

100%.

So yeah, interesting. So since we actually, before going into AI, I wanted to talk about Threads because we, you know, we talked about new types of consumer applications and And Threads was the big thing that happened in the past 6 months that just exploded at the beginning, and now we don't really hear about it that much. So was Threads more entertainment-focused platform? Was it more social platform? What happened there?

So yeah, I stuck my neck out a bit on Threads because the day after it launched, when everyone was kind of going crazy for Threads, I was like, hold on a minute, this is like, this is not sustainable. Because, you know, the metrics were incredible. Everybody was using it, got 100 million signups, and then everyone was even retaining even the next day and stuff. But the problem was all the content was about Threads on Threads, which was really entertaining for a day and hooked everybody. But then you're like, okay, wait, wait, now this is boring to read about Threads. And so what are we actually gonna talk about here? And I think what Threads really, really misunderstood was they misunderstood what— and this relates to your question. They misunderstood what Twitter was. They think— they thought Twitter was a feed of posts, basically a feed of text that people would follow and about the topics they're interested in, but just a feed. That was not— that's not the core of Twitter. The core of Twitter is a collection of communities that happen to be organized in a feed. So you've got Black Twitter, you've got Anime Twitter, you've got all kinds of weird stuff, unspeakable stuff. Twitter, you know, all kind of fashion and everything. But, and the algorithm has done a great job organizing you into one feed, but that's what it is. And what the criminal mistake that Threads made is that they treated it like Twitter was a follower platform, that basically you could just take your Instagram existing follower network, follower graph, and just import that into a text-based platform and And that would be entertaining. And the crucial mistake they made is they didn't include any topic search, hashtags, or any ability for people who are into weird shit to find each other. So I wasn't able to find the people who are into Arsenal or, well, whatever it is, brown shirts, or all the things that I'm into. I wasn't able to find them. So I just had to look at everyone talking about Threads.

Wait, is that because actually I felt like Threads was a bit of a shit— it became very quickly like full of shit. Like if you see on your— I call it For You, but I don't know what it is, like the thread discovery page. So that one is not just based on following, right? That one is based on following plus suggestions, right?

Yeah, yeah. But they didn't give people— they didn't give people a way to find the type of shit Right. They gave everyone the same generic garbage, but I want my garbage, and you probably want your garbage.

Oh, definitely. I don't care about the water was told on Twitter. But it's interesting because it's a dilemma that you— we don't— we haven't really solved before because no other platform has the chance of just pulling 100 million users from one side to another.

Yeah, well, actually, the time when this has happened before was Instagram from Facebook. And they took it, but I like the same shit. Back when Instagram is what it used to be, which is a friend-follower platform, I wanted the same shit on Facebook as I wanted on Instagram, which is my friends telling me what they're doing with their lives. So that worked.

But then it became entertainment first, and the second it became entertainment first, then you're creating a new experience reliant on social graphs, then it becomes messy.

Exactly.

Ah, interesting.

And then the people that I wanna follow from a photo perspective are not the same as the people that I want to follow from a text perspective.

That is also interesting.

Yeah, they're very different groups. And so I was, you know, not just me, but everybody was stuck with their Instagram followers with their Twitter type of content. It was Instagram followees with their Twitter type content. It was totally wrong. So I think the fact that they didn't find a way to get people to find their communities was a criminal mistake in terms of strategy, in my opinion.

Is it fixable?

I think it's gonna be hard now. I think it is probably fixable, but I think it's gonna be really, really hard now because they almost had their shot. And getting— convincing creators to invest time in a platform, especially big creators, it's not easy once there's no audience there.

That's the problem. But also, maybe they thought that the creators would stay there because their followership translated immediately. They didn't have to build a new thing. Like, I think Casper got millions of followers immediately just because they translated from Instagram to Threads, right?

Yeah. But what they forget is that the creators, they don't just want the number. They actually want people to look at what they're doing. They want people to— Yeah.

What do you do with like millions of followers if nobody actually—

Yeah, exactly.

Actually, that's even more sad if there is no— Exactly.

It's more embarrassing. It's like 3 million followers, 2 likes, you know what I mean?

And it's the platform's fault though.

Yeah, yeah, it's the platform's fault. And then everyone's like, oh. And then also then people get too embarrassed to start posting on it because it's like, this is lame. And like, yeah, we all hate Elon or whatever. If you're a creator, we don't like Elon, but like, come on, let's go back to that place where people are actually listening.

Well, have you felt Twitter, like Twitter's main feed, also getting more and more polluted? What changed there?

Did you notice Yeah, I mean, they've done exactly that transition from a follower platform that's based on the people that you follow, basically, and either you're friends with or you specifically want to follow, or moving that to content that's more viral and set to entertain you. You know, I personally, I think I more prefer the follower version where I can be more specific about what I want to see.

That's right.

But, you know, I can't argue with, I can't argue with their algorithms. I'm sure what they're doing is probably better optimized to keep people in the app. Yeah.

I wonder whether, because with TikToks, you understand immediately whether someone has liking for something within it or not. I mean, you have many, the visual input is much faster compared to text where I need to consume something. and you don't know my reaction or what I think because you don't know if I'm scrolling away because there is 5 tweets within. So I think the quality of the data that you can capture if you're TikTok is probably higher quality compared to Twitter. But maybe I'm wrong. It's just, I don't know if text-based is the optimal ground for data ingestion.

I think that's definitely right. I think that's definitely right. But I do think, you know, in their defense, and they've fucked up a whole load of different stuff. We could do a whole laundry list of things that they've done wrong.

Twitter.

Yeah, but in their defense, they've made it easier for creators to make viral content. That is true. And growing on Twitter, if you are prepared to pull the right viral levers, is easier than it used to be.

Yes, that is definitely— I mean, are you talking about the Thread boys that are like, this is how I made $10 million? Hey, here he goes. Yeah, there's some of that. Yeah, exactly.

But just generally, they've— yeah, that's the one. And some— that's some of the perverse incentives. But I think they have done quite a good job aligning the incentives of the creators to get followers and the audience to be engaged. Because the unfortunate thing about the thread boys— and you can mute them, by the way, I have some, not all— but yeah, so, you know, you could— but, but people, people like that. And you may hate it, but—

No, I completely— totally. Totally. But then, what do they say? I think Linda. I don't know if her name is Linda, the new CEO. Yeah, Carino. She says it's the world's. It's the world's open place to discuss the topics and everything.

Oh, it's not that.

Yeah.

Because well, the thing about that is the town square. Town square.

There you go.

But the thing about the funny thing about town squares. So what I used to say is Twitter is a town square, Instagram is a Disneyland. I would much rather invest. I'd rather be— if I was to own a town square or Disneyland, I'd definitely rather own the Disneyland than the town square. A town square is a terrible business model. In a town square, everybody's supposed to be equal, and then everyone gets pissed off with me when one person starts making more money than somebody else, and it's all supposed to be about democracy. That's a terrible business, and unfortunately, that's the business they built. That's the business they bought, and then they try to turn that into Disneyland and people get pissed.

Yes, you turn that into Disneyland and then the few followers of the town squares get completely disappointed. Yeah, but I guess that's— Disneyland's a much better business model.

Yes, and that's why, like, with the verified stuff on Twitter, you know, like, they started giving everyone blue checks, and then the people who didn't have blue checks got pissed off, and then people who actually did already have blue checks but then it got taken away, yeah, and it's like everyone's just like pissed all the time.

Yeah.

Whereas like with Instagram, because there's a social contract with Twitter that it's gonna be fair, that it's gonna be egalitarian, that it's gonna be right— whereas Instagram, they fucking did the same thing.

They introduced blue checks, but nobody cared.

Mark—

yeah, it's just Mark.

It's like money. It's okay. Yeah. So it's, you know, you just—

they bought a business that is not really a business and trying to portray themselves as the— the town squares locked them up in to this thing, and now there is an asymmetry between what people expect and what is becoming.

Yeah.

Although it may be much more attractive. And so, sorry, so right now today, would you rather invest in a Disneyland or in a town square? Because on the other side, you said that there is an opportunity for new social bondage and graphs and a platform sustaining that to be born. And so that's more like a town square, right?

That's a good point, yeah. I think I would want it to be More of a maybe more of a playground than a town square. I think the problem with the town square is that it's like debate and like intellectual discussion and politics and all that stuff.

Chaotic, hopefully.

Yeah, yeah, exactly. And and by the way, that can be great for creators doing that. There's a lot of people who make a lot of money talking about politics and making content about it, about controversial topics and all kinds of stuff. But I don't think you want to be the platform hosting that. Because of the whole level of expectations, whereas I feel like these pure social platforms, these follower platforms that are purely about socializing their friends, maybe they're more like a playground, and that can be a good business model. I think certainly better than like an egalitarian town square.

Playground that maybe you're not expecting a fifty billion company, but you are expecting a potential unicorn.

You can make some money. I mean, maybe you could make a. I mean, even old Instagram or Snapchat, for example, and I know that Snapchat, Snap. Obviously they are now prioritizing some of this creator stuff, but they've built, whether it's really worth the $10 billion or whatever it is in market cap, but they've built a decent business from that playground.

Well, the Snapchat there is also, yeah, Snapchat is interesting because people chat on Snapchat in the US especially. Oh yeah.

It's kind of WhatsApp for kids kind of thing.

Yeah, but you still think WhatsApp is gonna take over the US, right?

I do, I do think WhatsApp will take over the US.

Why are the network effects on WhatsApp so strong compared to Snapchat? Why is it— why is it gonna work better?

Yeah, it's really interesting. I think— well, I think WhatsApp will eventually work because it will start with group threads, which even if there's only 30% of Americans that are on Android, it just— more and more groups are gonna need to be there because otherwise you've got the green bubble problem and all the stuff. Yeah, features are better than SMS. That's issue one topic. And then the group threads become one-to-one and then you start using it. I think topic— the reason, the second reason is that SMS is getting really spammy. Like, I don't know, I don't know about you, but my SMS is full with like brands and shit that I don't want to see.

Deliveries.

And deliveries and like incoming stuff that I don't initiate, right? And WhatsApp have been super smart. Meta have been super smart and super deliberate about not allowing Companies to just send you WhatsApp without you having to first engage them. Yeah, it's beautiful. It's a great experience. Everything I'm seeing there is actually relevant to me.

I think it's really hard to build on top of WhatsApp anything at all, right? Like any sort of distribution for— I mean, there is WhatsApp for businesses, and there is a few companies that try to allow you to automate conversations. Yeah, it's like, it's not easy at all. I don't know. I don't know how the API allowance there.

No, it's a really good point. Actually, it's It's a space that we went super hard into. WhatsApp, building on WhatsApp, building sites on WhatsApp. That was like a key theme of ours when we started this fund, and it was something that we were really exploring. And increasingly we realized that WhatsApp are being super intentional about not letting brands give you spammy inbound, which is the one thing that keeps it relevant versus SMS.

Telegram is not doing that though. Telegram is quite spammy. It's quite spammy.

Yeah, Telegram's super spammy. Spammy, yeah. And also, like, I think there's also trust issues with Telegram. People don't really know about the ownership structure, and they're a bit worried about it.

I don't know too much about that. I know it used to be protocol, and it still is to some extent. I don't know exactly.

So my family group chat has been on Telegram for, like, I guess maybe 10 years or something like that.

And you've been trying to— too much friction to change to WhatsApp?

No, I think it's good to have all the people, like, My fiancée and my family.

I have them all on Telegram because WhatsApp is for VCs and WhatsApp is for VCs.

Maybe we can do Telegram. It'll be my fiancée, my family, and you as well.

No, but I wonder the reason I do struggle with inbox management on WhatsApp for sure. Because on one side they want to keep the interface extremely clean, but then on the other side—

You got too many friends.

Come on, don't put me on the spot like that. It's not about friends, you know. No, no, I know.

It's about VCs. No, you're absolutely right. And you are right. I think that's genuinely— that's a kind of you and me and us and generally VC problem.

Also a white problem niche.

But, but even then, if it was— if it were not WhatsApp, it would be SMS. Like, there's no— you can't— it's very hard to solve that with technology. Like, there's just too many people trying to hit you up, basically.

Yeah. We'll see in a year whether WhatsApp takes over the US.

Yeah, I think it'll take time. I think it'll take time. It's already happening. You can see the numbers of the active users growing year on year.

I mean, I remember when I went to the US in 2019 for university. Yeah, nobody had WhatsApp at all. And now today, if I have to chat to an editor or chat to someone, do you have WhatsApp? They know of it. They may not have it, but they're like, all right, what's up?

If you push them hard enough, they'll teach you.

Oh, they will. They'll give in. They'll give in. Okay, so we've spoken about that there is a potential opportunity for new playgrounds now, and you've started to invest in them. You've invested into Beehive, which is a great example of what could be a playground, maybe even a Disneyland someday if it shapes up that way. I wonder whether we can also talk— I don't know if Beehive already has, but I know at some point they'll be thinking about integrating AI features for copywriting and just making writing much easier. And there's going to be a whole new wave of AI tools of some sort in the entire creative tooling industry. So creative tooling, we've seen Runway, we've seen a bunch of examples that have been completely disrupted work— existing workflows, with new workflows, and running adjacent to existing platforms.

Yeah.

What's your view on how AI is impacting creative tooling today?

Yeah, so, as you said, AI is getting integrated into all these creative tools. It's making you make the same content better, faster, cheaper, higher quality. And so And yeah, Beehive, for example, does have AI features where you can automate, automate AI creation of photos and text and translate things like that.

Okay.

So we're obviously going to move to a world, there's no doubt about this, where AI is going to become more and more integrated into the workflow of a traditional creator. And so creator content creation is going to become more and more AI-focused on the one hand. But, and so there'll be more content that's generated by AI online. However, I think what's an underdiscussed trend, which I think will happen, is the exact opposite, which is once you get more and more AI content and it becomes less and less personal and relatable, um, and trustworthy, there's going to be more and more platforms that are going to be like no AI and going to be making sure things are pure, no filter. Like real true representations of the creator or your friend, because there's a level at which people just don't— people just don't want to— people don't just want to see garbage AI all day and they want to know what the creator's talents truly are.

Interesting. I wonder whether that's the case because in the back of my mind I'm thinking, is AI just a paintbrush?

Yes.

You know, before the paintbrush we were painting with our hands and then all of a sudden we have a paintbrush. brush, and people go, oh, that's— I want to see the hand painting because that's how we always used to do.

Yes.

And so I wonder whether that's just gonna be a one caveman type of things that at the beginning we're like, no, no painting with paintbrushes, but then over time it develops into actually being part of the entire workflow for also that original content.

Yeah, so a good analogy for the impact of AI on content creation is Picasso. So if you look at Picasso's first ever self-portrait that he did of himself, late 19th century, it was a beautiful drawing. It was like super specific and he looked exactly like, you know, whatever he was, early 20s Spanish man. And he represented it perfectly. If you look at his last ever self-portrait, it's like his jaw is over there. You can't even tell. You can't even see what's happening. And, and that's obviously a big part of that, is the fact that photography allowed people to represent content in a lifelike way. So there became no more point making lifelike drawings.

Ha!

And so I think something similar is gonna happen with, with, with AI in, in video content, or just general content creation, where As soon as AI allows you to make content in this incredible, efficient, and high-quality and relatable way, the format of non-AI content is going to have to— it's not going to die. It's not. It's not. That's certainly not going to die. There's going to have to be another way for people to express themselves, to engage their audiences. And it's not going to look anything like this AI stuff.

It's going to look like a jar.

It's going to— yeah, it's going to— Yeah, exactly.

That's interesting. I wonder, I wonder what is going to fuck up the Disneyland model.

Yeah.

Like if, if, because it is reliant to a certain extent on individual originality. I mean, wait, are we, are we assuming that there is going to be a convergence in the type of content? Because actually AI arguably could allow you to do the Picasso drawing with the jaw that is moved.

Yeah, yeah. And so, and so, yeah, so in terms of the actual content itself on traditional social— on some social platforms, it will, you know, it's not— the analogy isn't that it will be more crazy or less crazy. The AI stuff probably can be even more crazy, or crazy in a certain— in the way that we see it now. The analogy is like there's gonna be some new formats that will be impossible because The thing about the jaw over here and the paintbrush and you can't see their face, you can't compete with that with a photo. There's maybe a million photographers in the world in the— whenever Picasso died and when he wrote— when he made that last self-portrait, not one of those photographers can create that.

Right.

Only he can do that. And so there'll be— the equivalent will happen where people will want to see stuff that you can't make with AI, in my opinion. And, you know, you're already seeing that with like watermarking rules. And, you know, obviously government, for example, the government in China have put this regulation in place that means that AI content, AI-generated content, needs to be watermarked.

You know, I— it's interesting because sometimes I wonder whether all of art at the end of the day is just defined by its limitations, in the sense that the reason we appreciate something is because it's partially done with the limited capacity, like tools and abilities and you know, you can showcase how good you are because we're in a limited environment. If you could just conceptualize something, so it's not impressive.

I think that's an excellent point, and I completely agree. It's like in the world of animal behavior, like evolutionary biology, a signal is only a signal when it's costly.

Ha!

You can only— you can only— it only— you're only transmitting information Really, when it costs you something to create it, to transmit the information. As soon as everybody can transmit the information without any cost, like, you know, the bear, the gorilla that's beating his chest, if everybody can beat it just as loudly and make their chest just as big, that doesn't mean anything. But you have to actually be big and scary and the biggest gorilla to make that intimidating. And if AI just allows everybody to do whatever they want and make whatever content, you know, for example, the beach in Saint-Tropez. If I could— if we could just AI this setup and make it look like we're on the beach in Saint-Tropez living the good life with our private jet in the background, and we're— that's no longer impressive. And we're both VRs. That's no longer impressive because nobody cares. But they act— but, but my contention is that people still want to know if you can do that. They still want to know if you can go to Saint-Tropez and have the beach and the private jet and all that kind of stuff. But they're just gonna need that proven to them in a different way now. So that's, that's, that's the basis for the argument. That, that non-provably non-AI content is gonna get bigger in this AI age.

Why do people want to prove that you can do it? Why do you think we have that human need since we're talking about animal behavior?

Yeah, I think, I think people are fundamentally— a lot of the reason they enjoy social media content is they want to compare themselves to each other. They're fundamentally, you know, there's a lot, there's a lot about humans that are collaborative, and those are kind of the best parts. And those are the parts that make us all as successful as we are. But there's also a lot of human nature that's kind of competitive, which is more the unfortunate parts. But I think those competitive instincts are more the ones that drive usage of social media platforms rather than the collaborative ones.

So proving that you've had these limits— look how cool this photography looks. Yeah, there is no AI here. Look how impressive it is. I do wonder It's an interesting contention that you have, that we're still going to be interested in that watermarking. I wonder whether it's going to be too distracting to have this incredible unlimited creativity, or whatever you want to call it, this new age of content. Whether I can see myself getting distracted fairly easily and not caring that much, but maybe it's just me.

No, you're 100% right. Of course you don't want your whole feed made up of AI stuff.

Of course not.

Because you don't know who's written it. It's like the classic analogy is nobody wants to watch 2 AIs play chess against each other, whereas the World Chess Championship gets all these millions of streams and everyone's really excited. Or let's say cyborgs learn to play soccer, football, better than Ronaldo and Messi. Nobody's gonna watch them play against each other one-to-one, but you wanna know who's on top.

However, isn't watching 2 AIs play chess play against each other what we're doing when we're looking at the outcome. When we're looking at the AI, the trained model that over time has played against itself hundreds of thousands of times, we just, we look at the outcome of that model, which is at the end of the day we are still looking at that. Yeah, in another form, we are looking at the AI playing itself.

That's true. And, and, but that's facilitating the thing that we really want to to see, which is one person having glory and another person dying, and all this emotion that it's really not possible for AI-generated content to convey.

You think it's not possible?

I think it's not possible. I mean, so I saw a really funny, um, Reddit post that was like, I don't know what to do, my boss is sending all his emails via ChatGPT and I can tell and I don't know what to do. And it was this like super long detailed explanation of how somebody's boss was just using ChatGPT to send all emails. And like those emails are probably almost exactly the same as the emails that he would've sent himself anyway. But there's something super infuriating about the fact that he didn't actually write them himself and he didn't put the effort and he didn't craft it. And I think we really, things are gonna move further and further in that direction.

To prove your point that we're gonna need to see direction that is not enhanced by AI. I actually think over time we'll see people sending emails that are completely the opposite because you don't want to write a glorious email because the more glorious, the more likely it's done by AI, etc. So it's just going to be boiled down to an extremely functional detail.

It's like, dude, 100%.

Yes. Yeah, yeah, yeah. Which right now is not— does not happen on email.

You don't need the yes.

Yeah.

Why? Yeah, exactly. Why?

Lowercase. Well, because on the other side you have the I route, which is all the glamorous, etc.

That's why I always— that's why I joke that Sam Altman always tweets in lowercase. It's so he can prove that it's not ChatGPT writing his tweets.

Is that a joke? Yeah, yeah. Because it has some merit.

He does always tweet in lowercase. I mean, it may be true as well. Yeah.

I mean, you can prompt it fairly easily.

Yeah, yeah. Just do That's a good point. Yeah, yeah, yeah, that's true. Now you can— and how you can make it all— you can do those precooked prompts where it applies it to all of your—

Oh yeah, yeah, yeah, of course.

So yeah, I mean, but there's gonna have to be— but this is the problem, is that it's an arms race between the user who doesn't want to see this AI stuff, in my opinion, as much as they're gonna get it, and then the creator and the platforms that are down for it. So I think the solution is gonna be is going to be watermarking of some case, in some way, or it's going to be a new direction in the form of content.

Interesting. I personally don't know where I land. I've been battling with this for a long time, and I feel I do see the compelling argument of unlimited creativity enabled by AI.

Yeah, yeah, exactly. Me too. 100%.

But I still don't know whether we want that.

If they—

yeah, I don't know whether humans will want will care about human, like, oh, I wanna see that this is 95% human-built. I generally don't care if people are, people are quite lazy.

No, so I completely agree. So I think what we're gonna see is like, you know, in the art market, you have like 99% of art sold in the world is the fucking picture on the wall that says, I love my family, and like, heart to my home. And then it's like a big juicy red heart. That's like most art. And that's the AI equivalent of like shit that people want because it's easy and they can just digest it. They don't have to worry about it. That's gonna be huge. I'm not saying that's not gonna be huge.

It will.

But there's gonna be a kind of reaction against that in the form of more exploratory type of content where it will matter who made it. And that will proliferate.

Although a model sophisticated enough should be able to understand that that content is shit, it won't have enough entropy and divergence to not like to suggest content for that audience that needs that as well.

That's exactly right. But people will want to know that it's not generated by a model.

So the— So you'll have models trying to break into that operating system?

I think so. I think, I think people will start to, start to actually care about— this is, this is way in the future, by the way. We're nowhere near that now, of course. But people will start to care about what is created by the model and what isn't, because they want to know if somebody's skilled. By the way—

Well, teachers, teachers, right? They're like, they don't want essays written by ChatGPT because there you're testing whether someone is skilled.

And then, and then even in, in, in the form of art, I think a teacher analogy is a great one. In the form of art, the reason that the Picasso example was an interesting one is because he was really good at drawing. And the fact that, you know, when you have these modern artists who paint really abstract pictures, you know, like just blue canvas or whatever.

Yeah.

And it takes them 10 seconds or whatever. But often, often they actually are really good at drawing. And I think at some level, if you're a collector or you're really into the art world, it's somehow important to you, the fact that they can also draw and they're choosing to do the blue canvas. Like Picasso, he chose to do this abstract stuff that was much easier functionally, Than what he used to do. But somehow it's important to the viewer that they could do really fancy art that looks really accurate, but they've chosen not to.

It's not out of lack of ability that you're doing this canvas. It's out of choice, of range within your ability. You're choosing that because it's your creative representation of what you want.

Exactly. Yes.

So, but that doesn't necessarily need to be the artwork proven that is human-made. Well, what you're trying to prove is proof of capacity, proof of ability. And if you have that proof of ability, then you're fine using AI. Maybe that's also more allowing for AI. That's an interesting point. Who knows? Honestly, I don't know. But back to AI as a— also as a layer within the creative stack of creators that are creating all sorts of content. You've also invested in ElevenLabs, right?

Yeah.

ElevenLabs is part of the stack that does that automating of— or I mean, it doesn't, you know, you can sort of use someone else's voice to utilize that. What are the use cases for that that you think are particularly helpful for?

I think right now there's a lot of consumer usage of people wanting to get Joe Biden to play against Donald Trump and then Barack on a video game, and then Barack Obama comes in. Hilarious, right? Hilarious, hilarious. The first, second, third, all the way up to a thousandth time, and then maybe less hilarious thousandth one. But yes, they've done super well on nailing that consumer use case. But I think the long-term business model is going to be more in kind of enterprise B2B type use cases. Where whether that's a gaming platform that can have an NPC talk creatively about everything and you don't like have the NPC being like, the victory is over there, and then you come back, it's like, the victory is over there, and it's the same thing again and again. They can be more wordy. So I think gaming is a really interesting one. I think that like sales enablement videos and things like that, and even podcasts, like form of entertainment, audiobooks, that kind of stuff.

I mean, you are powering the Disneyland, so you are, you are just a tool in the creation of that. You still believe there is a massive market of AI-generated content, which a piece of it can be generated by— it can be— you can use voice also. You see all sorts of TikToks even today with AI-generated voices.

Yeah. And it's great content and it's amazing because it empowers— like, imagine if you were a kid in India or Nigeria and you had a really good idea for a video, but your accent isn't necessarily the one that's going to work for this viral TikTok, you can now just edit it together, use clips from the internet which you've taken, and then put an ElevenLabs video on that and make it super viral, make great content. And so yeah, it's like everyone talks about democratizing, um, democratizing creative tools, but ElevenLabs is a perfect example of companies doing that really well.

Yeah, I mean, I can see myself very Recording the intro with— I don't need to be there.

Yeah, exactly.

Your intro is great.

You can—

I'll just narrate it. I do wonder when it comes to differentiation and defensibility again, though, not necessarily applied to ElevenLabs. I mean more in general, AI and creative tools. So the thing that worries me here when investing in new creative tools that are powered by AI is that they tend not to own part of the— they don't own the technology usually, unless you're investing at the model layer.

Yeah.

So it tends to become a thin UI layer on top of existing models that are third-party. That is cool because you can create new workflows and new, completely new products that, you know, like Lensa.ai at the beginning, right? Everybody went fucking nuts. I think you covered it as well. Yeah, yeah, everybody goes nuts. But then you can— it's super fungible. You can get a new Lensa AI because the model is not theirs.

Um, yeah, well, actually, I think there were 2 things happening there with Lensa. One of them was the point you're making exactly, which is anyone could just put a wrapper on Stable Diffusion or whatever, right? And that caused people to use AI profile pictures from other tools, right? But actually, I think the bigger reason that Lensa AI collapsed is that people got sick of AI profile pictures, is that it stopped being cool. Because— and I think that comes back to the threads, to the barbell. Oh, it comes back to the AI barbell, where it's like, once it's really easy to look like a beautiful spaceman or like a cowboy on a thing, or like you as a rich guy throwing out money at the strip club or whatever crazy stuff people want to have themselves doing. As soon as that becomes easy, it's like, wait, that's not cool. That person's just done that in Lensa. That's like not sick. And so that's what for me collapsed Lensa, not the fact that they're just a thin wrapper on the AI software.

That is also interesting. Yeah, it's the— because it's not unique, it's not cool anymore. Because The one thing you do is democratize access, but if you rely on lack of access because that's what cool is, then you're kind of shooting yourself in the foot. Although my point around lack of, like, being wrappers on top of existing third-party models is still as good.

Crucial point.

So how do you navigate unique differentiated offerings in creative tools that are powered by AI?

Yeah, it's a— it's, it's the, the, the trillion-dollar question. Yeah, I mean, yeah, so I think in, in terms of, uh, you know, ElevenLabs specifically, just briefly to cover them, I mean, they, they have built a lot of bespoke technology. And, and also with them and Midjourney and others, the way they can improve long term, in theory at least, is they offer you a generation And then you accept or reject it, and then you might use another one. And so over time, they can start to build that into their technology to improve it over time based on the ones that people picked. And that's the same reason why Midjourney is by far the best image generator, is because they have the benefit of all that data of everyone being like, that's— option 1 was the best, option 4 was the best, option 3 is best. So they can feed that back into the models.

Because they own the user base, and so they have that data feedback loop.

Yeah. So there is, so there, you know, that is at least theoretically a defensibility.

It becomes.

But having said that, I 100% agree with your implied, the implicit point you're making, which is that a lot of AI companies are not gonna make good venture investments at all. And I think the reason for that is number one, what your point about defensibility, which is like the core thing that VC-backed investments have had is like software, to mean that other people can't do the same thing as quickly. Number 2 is— this is where it gets a bit financial— but number 2 is like gross margins. Like, software is a beautiful business model because there's no incremental cost to serve a new customer. Gross margin, maybe a little bit of hosting, but it's like 80%+ gross margins. In AI, these GPUs are not free.

I mean, I wonder what the gross margins of initial software companies looked like. Yeah, because, because as we go through that pendulum that we start—

exactly right. But those— but that back then, in those days when, when software costs a lot of money to serve, that was one-time licenses.

There was like—

there wasn't SaaS. That was pre-SaaS. That was when you pay for the installation, and then when you want a new version, you pay for it again. So that business model, that SaaS thing, was only became possible when you had the rise of cloud and And third-party hosting.

Yeah.

But yeah, so like gross margins of AI companies are just going to be structurally lower. There's no getting around it. It's just going to be the case. They will have lower margins, especially as the models become more and more and more intensive. Maybe in a super long-term world, the gross margins will improve over time. But sorry, but that's the second one. And the third one is the retention, where the problem with AI tools is that people want to try them. They just want— they're interested in the new technology and That's meant that the retention rate of all of these AI apps, even the absolute best, has been pretty low, certainly lower than any kind of software company that we've been used to. And so you have the general low defensibility, you have the lower gross margins, and you have the lower retention. Those are all the 3 things that made SaaS and software a beautiful investment thesis. And so you don't have any of them. And so I think there's going to be a lot of money burned in this AI rush, and there'll be some great companies built. But there'll be some flameouts.

But I like it. It's a good mental model to go at it. If we dissect them, let me take first gross margins. You don't need to run a model necessarily. You don't need to build from scratch your own model, right?

Yeah.

Arguably, you're going to go more and more in the direction of taking third-party open-source models, and then you're just running that on the cloud and you do your own thing, right? And to that, I don't know to which extent it actually affects your margin if you don't need to build from scratch.

Somebody's got to pay for the GPUs, somebody's got to pay for the training, somebody's got to pay for the usage.

Yeah, but as there is more demand, the supply usually adjusts and then price goes down. So it depends how long it's going to take. Yeah, I guess that's the question. It's like, for the foreseeable future, the gross margins are possibly going to be affected by that until that becomes integral parts of smooth of the cool, like fair enough. In terms of retention, if you're looking at retention in terms of like the hype, it's like, oh, let me try Lensa, and then you don't have the retention. The retention looks bad because you've had a massive boom because it still looks so cool because it's such a new fucking cool thing. But then maybe it doesn't substitute fully and it doesn't integrate in your workflow. So it's this cool A thing that you're willing to try out, but then you're not really integrating that. I do think over time we will see products that stick and integrate fully within workflows. So that— it's not like it's a native characteristic of AI products that they have poor retention. It's because for now, most of them are just jewels that haven't really found their necklace.

I think that's a great point. I think that's a great point. And the ones that will work will be the ones with high retention. It's just right now there's just a sea of ones with low retention that are still getting funded as if they have high retention. That's what more scares me, but everything you're saying is true.

And that is the trick of the trade. It's very easy to fall into the trap of thinking, like, as a VC, I mean, I'm sure you see as much as I do of tools that are like, that you can see the journey. And then you can see the FOMO because it starts to pick up and then it will just crumble like the others.

Yeah, I mean, revenue growth is not the whole game. You know, it's like you've got to look at the quality of the revenue, which is easy to say in retrospect, but it's sometimes harder to do at the time.

Do you have any comments in terms of incumbents' predisposition to win in the AI space? Because I think in previous waves we've seen, we've seen AIs Sorry, we've seen startups and incumbents kind of split revenues accordingly. But in this one, in this new wave with AI, there is an argument that incumbents will capture the majority of value, and that is because of distribution power. You know, they've got users and also I think data, you know, they have so much more data. What do you think?

So I 100% agree with that. With that. So if you look at the history of new technologies that have been disruptive, you know, starting with even just the idea of hardware and software in the, you know, late 20th century, that enabled new product experiences sold in new ways. You know, CD-ROMs, computers— these were all things that you couldn't sell if you were a, you know, an analog company. It was— these are digital experiences. Then the internet wave came along, and then you started to be able to sell things online, and that was like inaccessible in a lot of ways to the old-school companies because these are new— like Google, you couldn't build a Google if you were a hardware company. It just didn't, it just didn't make sense. And then the next one was like cloud and mobile, which obviously mobile, the clue's in the name, that's a new distribution channel. Cloud allowed, as we were talking about earlier, cloud allows for software-as-a-service subscription. You don't have to pay upfront, which basically allows a whole new type of company to come and replace the legacy companies that were doing one-time installations. So all of these things allowed for a new type of company to add value in a new way that was inaccessible to the old incumbents. I think the difference now with AI and the kind of one of the red flags that we're starting to see is that this technology works just as well for the incumbent as it does for the disruptor. In many ways, it actually works better for the incumbent because they already have the users, they already have the data, they can already have more financial resources to build this stuff. And the great thing is, because of all these VC-funded startups, it's becoming really plug-and-play for them. They're also the best customers for these VC-funded developer tool startups, the big companies. So all these companies are building for enterprise. They're not building for startups anymore. And so my personal take is that the vast majority of value created by AI software is going to go to incumbents, and the best AI tools in many cases are going to be pre-existing companies.

Yes, it's very interesting. I wonder whether there is still a case to work on an AI company just by how large the market size of these, you know, the Googles— even if you can capture 10% of that and create something and use it. It's interesting.

Definitely.

It's enough, right?

Yeah, 100%. 100%. And there will be examples which will be inaccessible to big companies. For example, I think, you know, we think AI voice is an example. I mean, we would say that given that we've invested in the space before, but it's gonna be quite difficult for big companies to do voice cloning at scale. just because it's quite controversial. You can end up with some things that you don't necessarily want from a PR perspective coming out of your platform with AI voice. And there might be some other examples like that where the big companies, for whatever reason, are strategically not able to work on it for some reason. And those would be really attractive.

Well, yeah, I mean, I think there is an element of risk. Like, if I'm Google and I'm pushing out a new AI search function, I guess you can scale it out and start with 1,000 users, 1 million, etc. But there is a risk that you're fucking up your most valuable asset. So there is obviously more friction in adopting new elements within your tool stack, and that's possibly something that can hold incumbents back as compared to startups.

Absolutely. So like, for example, Google versus ChatGPT. ChatGPT is OpenAI is a startup and they've managed to do this. Obviously they've built better software and everything, but— If Google launched this— It's gonna be hard for Google to launch it because—

Well, Bard.

It does, but Bard sucks if we're being real. And they, you know, they haven't prioritized making Bard super good because a lot of their revenue is dependent on not being able to— not getting the answer immediately. It's dependent on being able to search and click ads that will give you the answer for you. And so—

But I'm sure they know it's not sustainable. Like, no?

No, yeah, you probably—

There is probably value capturing right now. They're like, we still have a few years of possibly this, you know, incredible engine that we've built suggesting trash and monetizing on the time spent. But then over time, I don't know if that's necessarily the choice they want to take. So yeah.

And they can both coexist. I mean, there can still be— because there's a lot of people, for example, like me, who want to see the Google results and also want to do the work themselves to understand why that's the answer and to see the sources. And to— like, if I get— if I'm in an accident and I fall off my ladder or whatever and I want a good lawyer who's going to represent me to the ladder company, I don't just want ChatGPT to tell me that one lawyer. I want to go Google it, I want to look at their website, see how legit they are, look at their reviews, all that stuff.

Although arguably you do want to have an assistant that is trained on lawyer data for what you want specifically, so that you can get to— you can maybe save 50% of the cost and only get there. You know, you don't have to go to the GP if you already know what's wrong with you.

Yeah, you can just skip that step, right? And so I think— and that's why I think there'll be space for There'll be space for both.

Interesting. Cool. Okay. I wanted to go into the last segment that is more around creators specifically and their— and the work they've been doing around venture. And I think you've got a really good, you know, overview of the land at the moment because I assume most of your LPs are creators. No?

No, our LPs are like traditional VC LPs.

Oh, massive miss, right? Okay, well, Casper, your co-founder, is a creator. To what extent do— so is it by choice that these creators are not your LPs, or did you do it on purpose?

Was that— We did it on purpose, yeah. So, you know, I'll tell the story of it basically. We started 2019 investing together, Casper and I. Casper is a content creator.

Yep.

Grew a big audience, created quite a large business called influencer.com, which does influencer marketing for a load of big enterprises. They employ like 150 people. It's quite a big company now. And he's co-founded various other businesses as well. So he's got the creator thing and the founder thing. And we were cousins, so we started investing together. Um, and we created an investment club. It was called Creative Investment Club. I was working full-time in private equity at the time. And we invited other creators to come and invest with us. Like people with over a million followers could come join us to the deals we were investing in. We didn't charge any fees or carry, or we weren't trying to profit from it in the short term. We were just trying to invest in the best businesses, and we figured that having a load of value-add people investing with us could help us get into those kinds of deals.

So it did start with creators?

Exactly. So you— yeah, so you're right. So we— so we did that back in 2019 as a side gig. We did some deals, we got lucky with some of them, some of them went well, and we developed a good track record. And so then we started thinking, okay, maybe we can turn this into a venture capital fund. So I left my job in private equity to go start this fund, which we renamed from the terrible name Creative Investment Club to call it Creative Ventures. Much better name. It was so cringy having it, by the way, having a terrible name for your startup is is the worst thing ever because every single time you introduce yourself, it's like makes you embarrassed. You know, you cringe every time you say it. Can you imagine how bad that is?

Uh, it's— I can, I can. But how, how are you guys like, yeah, good to go?

This is it. On the— yeah, well, actually, to be honest, we— I've skipped out a whole nother even worse name that we had called, uh, Creator Collective Capital, which was halfway through. But that, that we just like erased that in the history of the Yeah, so that was, that was bad. But so in any case, so then we started, then we launched our venture fund, which Casper and I are the co-founders of, and that we invest in consumer internet startups. Our LPs are not creators. Our LPs are funds of funds, family offices, other VCs, my old colleagues, and, you know, people like that. And And then in terms of creator involvement, we still sometimes invest with creators from time to time when there's some value they can add, but it's not like a formal thing. It's more of an unstructured thing.

So did you, did you figure that creators weren't making the best business partners after that attempt of syndicating your way through to then build a fund?

No, creators were really good investment partners for us. Creators, I think creators add a lot of value when Doing investments, but I don't think it's good for a VC fund to be reliant on them. I think nobody should be reliant on creators.

Okay, tell me about that.

The right ones, I think, you know, with with with some creators, for example Casper, he's super professional, super businessy. He's a chill guy, but he he understands business and he's a founder, right? And he kind of like gets it from business perspective. But which is great for us because he's He understands that stuff, and he's also professional. I think with some other creators, they're really smart and they really get it, but it's not always easy to get them to sign docs on time. And then they may just have a reason that they're not interested in a deal for some reason that doesn't necessarily represent whether it's a good deal or not.

They're really good on the creative side, but then when it comes to actually getting stuff done on time and behind the scenes— okay, that's right.

So, but in terms of your broader question of where, where do I, what do I think think about creators in startups and VC. There's so many different angles to attack that question. It's such a complex question. I mean, starting— I think one of the most interesting places to go with that is like creators as founders of companies. So in mid-2021, at the height of the creator economy hype, when it seemed like every creator-backed business was going to be a huge thing and all these creator software companies were going to unicorns. We did a study which actually we never published, but we did a study of every single creator-founded business in the world. We created this Google Sheet and we mapped every single one. And it was like 700 businesses, 600 to 700 businesses, each founded by a creator. And we did— we ran some numbers on them. So the first one was 40% of them were fashion businesses. 40%. 20% were beauty. So that's 60% just fashion and beauty. More than half were just fashion and beauty.

Aesthetics.

Aesthetics, exactly. And then there was like 35% mixed, like toys and all these kind of other categories and stuff. Less than 5% were B2B businesses. Very small number.

I mean, of course, because their audience is—

Ah, the B2C. You're right, that's the natural thing. But then what's funny is we relooked at it again Like a year and a half later, I think we actually looked at a year and then we keep looking at it now. It's like they all failed. They've all failed.

No, they've all failed.

But pretty much, I wouldn't say—

which category, like B2B or—

well, so all of the consumer ones are doing— I'd say maybe not all. I'd say 95% of consumer ones are doing dramatically less well by every possible metric you could find online than they were. The B2B ones like influencer.com and the others are doing much better than the others.

By Casper, influencer.com?

Yeah, Casper, that's his business. And so, you know, you, you start to realize like creators as founders are incredible at like the first 3 months. But then the problem is when you— business is about building momentum, as you know as a VC. Like you need to compound the wins and it's actually very hard to build a sustainable business that starts like that. You know, and so, and that's what a lot of these— that's what all these consumer products creative businesses does. And so what's happened is that there's been a whole load of cash flow businesses that thought they were equity businesses, that like businesses that just generated a ton of short-term cash but didn't actually build anything lasting. Right.

So as a VC, you think it's interesting because you potentially see the opportunity in distribution, but then that's immediate distribution that is They're cash flow generating, but not necessarily building towards something more long-term.

Exactly. In the vast majority of cases.

In the vast majority of cases. I do wonder, I mean, so the very strong virality effects, of course, of course, very strong at having an audience from day zero that they can distribute to. The missing part of the puzzle is possibly retention and network effects around there that can keep users engaged. So it does make sense, as you're talking about it, to have a combination of creators that helps with the uplift of a product that then has product-market fit.

Yes.

So that is an interesting combo, where you have creators as a co-founder or as an initial partner that can help with the zero-to-one of the immediate audience appeal, but then the thing has a life on its own.

That's exactly it. And that second part is so important, of having something that's actually good or different or valuable or positioned different. And the creator itself is not going to create magic from nothing.

Yeah, yeah. Any— do you have any example at all of someone that has done something good, or like from the product-market fit on its own, that has partnered with creators?

Yeah, absolutely. Yeah, there's loads. And, you know, it's still early in a way, so you never know what happens with these things. But everything from, you know, Fenty Beauty has done— has at least built something somewhat large.

Which one?

That's Rihanna's beauty line. And, you know, I happen to think that maybe, you know, that it did benefit from a lot. And then who knows how— where it will settle out in the end. But they did build something objectively pretty big. Skims from Kim Kardashian has done really well. And there are some others that have, you know, especially ones that don't seem to be based too much around the creator themselves.

Right. Because otherwise they tap and saturate their audience. Like, once you tell your audience too much about something, it's like you've saturated that market. And so easy to get there, but then you can't keep doing that over time.

Yeah, and then there's others like, for example, like Summer Fridays. There's like, it's a, you know, beauty slash skincare brand. There are quite a few that have actually done well and built lasting things, but it's the tiny minority.

But it's mostly still on direct-to-consumer. Yeah, it's It's very rarely some other non-direct-to-consumer offering, as you've seen with the 5%. Yeah, exactly.

That's a big opportunity. I really do believe B2B creator-founded companies are the biggest, most untapped opportunity within this creator-founded space.

But why? What's the benefit from having a creator on a B2B company?

I'll give you an example. Influencer.com, when they were getting going, they, they went to a web summit, a big conference in there. I don't know if it was their second year or their third year.

He had Casper there.

So all of that, all of their competitors were paying €5,000 for this booth where they would like beg people to come talk to them and say, look, like, I'm an influencer marketing company, come talk to me. Casper got them on the main stage to announce their new product and And everybody was coming to them all for free. Because the amazing thing that people don't understand is that the B2B world cares more about creators than the B2C world because they're so not used to it. It's like so much fun for them to have somebody who's like got followers and knows how to engage people. Everyone's so fucking boring. So if you're not boring, you have this huge, huge advantage over all these super boring companies that don't understand anything about social media.

For now.

Interesting.

Yeah, but then it gets saturated as you have the interns that take over the TikTok profiles of Ryanair.

That's a great point. Yeah, and that's— they're benefiting from the same thing, which is like people expect a TikTok of an airline to be boring, but it's not. And so people are like surprised, and then they give it kind of additional credit.

But then how— yes, fair enough. And then if that becomes the norm, it becomes obsolete fairly quickly.

Yeah, yeah, maybe that's right. Maybe in 10 years' time we won't be talking about that anymore because the opportunity will be gone. Who knows?

The opportunity will be gone, possibly. I still wonder how you manage brand. I don't know how that intern initially got the good-to-go from the social media team.

It's brave, it's good.

It's good stuff, good for them. And okay, we've discussed founders, we've discussed LPs. Oh, founders. Do we mean what kind of equity splits can we think about in these cases when there is a co-founder? Obviously, it's super case-dependent.

That's a really, really interesting question. It is super case-dependent. And it depends on— usually businesses that have the creator as a more— not the majority founder are going to be better ones. Because they're going to be ones where the core is the idea and the execution team and the people building it, as opposed to the creator. The kind of more merch-like companies where the creator owns 80% and then you just get some underling to do it, or their management company for 20%, those are less likely to be sort of high-quality businesses. So I think in general, ones where the creator, funnily enough, owns less are usually going to be better companies.

I wonder whether there is a a studio-like model to be built here where you have, you know, ideas and then creators and, you know, you can kind of, if you can tap into that and find aligned audiences that you can tap into that, it becomes much better than a clothing line for a creator over time. But it's also very beneficial from the other standpoint, from non-creators. Yeah.

So yeah, there are studio businesses that do that.

Really?

Yeah, yeah, that kind of incubate businesses for creators.

Oh, is that a thing?

Yeah, and some of them have done really well, but, you know, the way they've done well is they've been smart about knowing when the cash flows are gonna end and not pretending that they're building these big businesses.

Right.

And just more like, okay, we're gonna make 70%, we're gonna make 60% gross margins on this product. We're not gonna make too much because we know it's We don't want to be holding a load of inventory that doesn't get sold. So we're just going to make as much cash as we can from the first 3 drops, and then we're going to stop.

Because it's still going to— well, because there is no inherent retention. And so, but then the second it gets sold, that it's like the worst acquisition in history because it's fully reliant on the creator.

Yes.

And it's not part of the— it's not incentivized anymore. So they're non-assets. They're not a— because you still— it's just like the creator selling. And so the second it's out of the— it's a zero-worth acquisition then.

Yeah. So it's a really— it's, it's a model that you need to be really careful if you have. There are companies that do this really well, but they need to be really careful about how they manage the whole process.

Got it. Got it. Interesting. Okay. I've got one last question. Why does MrBeast follow you on Twitter?

Because I once tweeted something about MrBeast. MrBeast, but actually it was about— it was when, um, it was when Twitter started, um, showing the— you remember when you started being able to subscribe to specific creators? Yeah. And I basically went on to Sensor Tower and I found the top 10 most subscribed to creators and I tweeted about it and like MrBeast was number 3 or whatever and then I tweeted about it and then He saw the post and then DM'd me like, "Yo, where did you get that data?" And I was like, "I got it from here," and then he followed me.

Nice.

That was it.

Good. And now you chat on him with him on a on a regular.

Yeah, exactly. Everything you know he's actually dialed in here. We can bring him in.

And you're working on a on a on co-founding something with him?

No, I'm loyal.

I'm loyal. All right.

Loyal to Casper.

All right. Good. Well, Sasha, thank you so much for coming on.

It was great. Yeah, it was good fun.

Appreciate it. Thanks.