How AI Will Rewire Consumer Internet | Deep Dives with a16z

Right now, AI has been so much about productivity, job replacement, work replacement. And I think we have a moment to shift that, which is how does these new tools help you get more out of your own day and your own life and the things you want to do?

So it sounds like you're saying retention is still the most important metric. Is that true?

I mean, I think it's always been.

But you've been a part of really every product cycle, the big ones and the small ones, and now most recently leading product marketing for a lot of the AI efforts at Apple.

One of the things I learned a lot at Apple was how do we tell a compelling story to regular people? One of those things that the moment you use it, you're like, oh my God, I can never go back to the old way. And we forget that like technology's no longer the underdog. The parts of the products we've made are running the world.

You know, we used to say real life, but like online life is real life now.

ChatGPT was amazing for a whole bunch of searches that we thought were totally solved through Google.

Yeah.

And all of a sudden you start putting those searches into ChatGPT, you're like, this answer is more than 10 times better. This experience is more than 10 times better than Google links back and forth.

Yeah.

And I think that's why ChatGPT blew up.

One topic that we discuss a lot is, are the labs just gonna win it all? Should we all just go home? And I know it feels like we have a version of this conversation every time there's a new product cycle. What's your view?

So I think a funny way to think about this is—

Welcome back to the Andreessen Horowitz Podcast. I'm here with a person that I've been friends with for a long time. I admire for a long time, was my investor for a brief time. Um, and now I'm proud to call him partner, Josh Elman. Welcome.

Thanks so much for having me. I'm really excited about this.

It's gonna be awesome. Yeah. So Josh, uh, everybody sort of probably already knows him, but you know, you, you've been a part of every important consumer story. I don't know if that's because you've been the catalyzing force behind them, um, or perhaps you're a good picker.

Um, I like to think it's sometimes it's like just Forrest Gump showing up at the right place at the right time and being on the journeys. But it's been really incredible to get to be part of creating technology that, that we all use every day in our lives.

Yeah, I was, uh, I was reading some of the background and it sounds like you sort of set an intention at Stanford to be a part of the intersection of culture and technology, and you've kind of done that.

Yeah, it's kind of crazy. I, I think my resume, um, had this objective. It was like to create tech, great technology that changes people's lives. Yeah. And as I, as I think back on that, I'm like, I wish it was even more like benefits people's lives, that makes people's lives even better. 'Cause I think changes lives, technology's kind of neutral, but there's so much that we've done that's just tried to like make the world better, more connected, more human. Yes. And give us all these new tools and powers that we never could have had before.

But it's interesting because you came from a time, you know, when you graduated, technology was this, you know, sort of these weirdos in the back of the classroom. You know, it always felt like a toy. And now it's something that, you know, sort of some of the most important conversations in our society are about technology.

Yeah, they're about technology. They're happening through technology. It's this trend. I mean, we used to joke that if you got to like 10 million users, you were impacting so many people in the world. And then it became 100 million and then it became a billion and then it became billions every hour. And we forget that like technology is no longer the underdog, like the part Parts of the products we've made are running the world.

It's like, you know, we used to say real life, but like online life is real life now.

Yeah.

I think it is sort of default.

I think the pandemic proved that for us most of all. We kind of maxed out on how much you could actually do in the online life. And there were so many things where we would meet over Zoom and we would all have these conversations, even with like family reunions and friend groups and like birthdays and dinners. Totally. We kind of maxed out what that was. And I think what's been really cool as we've come out of that is so much of that has lasted and has made a lot of things more natural. And we still like to sometimes even connect over video, but now we also are getting much more back in the real world and technology can do a lot more, I think, to bring that together.

So before we get too deep into it, let me make sure I kind of frame up the background. So you've been a part of really every product cycle, the big ones and the small ones. So Facebook, LinkedIn, and Twitter, now known as X, in varying roles there, which we should talk through. Um, TikTok or Musical.ly, Discord, Robinhood, and now most recently leading product marketing for a lot of the AI efforts at Apple.

Yeah. Yeah. I've, wow. I mean, when you put all that together, it's been an incredible run, but what I just came off with Apple was a real change for me. About 6 years ago, I decided I'd never worked at a large company and I thought we were in a place where startups were actually gonna struggle against some of the big companies. And I was like, let me go try the other side and learn how that works. I got this incredible opportunity to work at Apple. What a fabulous company. I've had the best time working there.

It's interesting because even within, you know, I'd say information's pretty fluid in our industry, but Apple's famously opaque. I mean, what, maybe what did you change your mind on through the Apple experience? And actually before that, tee up exactly what you worked on, what shipped, you know, tell us more.

Yeah. So, you know, when I joined Apple about 6 years ago, I started in the App Store business, which is a giant business for Apple, you know, between both discovery through the actual App Store client as well as a whole platform for in-app purchases. Through games and apps. And the question was, how do we actually help people discover more? And how do we help people get more out of their apps? And working on features that surfaced more things that were happening in apps to users, both through the store and through other products. We actually had this vision to kind of launch the Games app, which was this idea of something that you could go every day when you want to figure out what to play and what was happening in the games that were already downloaded on your device or what your friends were actually doing. And that launched a couple of years ago. I'd gotten that started. But then AI started happening, and I got an opportunity to move over and sort of start up the product marketing function or restart the product marketing function around AI for Apple. And we had these visions of what was personal intelligence going to be? How could you really bring that to people embedded in the device where all your content is, all your data? Siri had been around for a long time. It hadn't been as successful or well-reputed as I think people had hoped it could be. And I think there was a huge opportunity with what was happening in GenAI.

Mm-hmm.

to rebuild Siri. And I got to be part of that over the past 2 and a half years. It's been a very interesting roller coaster. But I'm so proud with what was just announced 2 weeks ago at WWDC. I mean, the story of what Siri AI is, and I'm using it every day and it's fantastic.

Yeah. Yeah. So we've had a lot of intelligence over the last 3 years, but arguably less personal intelligence. Talk about the personal part. What does that uniquely unlock for you?

Yeah, and I think what's really special, you know, with these assistants is not just what they can help you do or navigate in the world and kind of bring world knowledge to place, but what they can understand about you. And it's not just what you ask it over time, but having access to all your information. And when you trust it, your device has your mail, your messages, your calendar, notes, anything that you save, being able to just have an assistant that can tap into that and answer questions. I had this experience about a month ago when I was beta testing this, and I was driving to a dinner. And all of a sudden freaked out and I was like, oh my God, am I going to the right place? And I hadn't added it to my calendar. I just knew I was meeting a friend for dinner. We texted about it. I just said, navigate me to dinner. And it found that message and got us and got me right there. And I was a few blocks off. I would've gone to the wrong restaurant 'cause I just had remembered it wrong. Um, this morning I was trying to figure out where to park and someone had texted me a whole bunch of different parking locations.

Hmm.

I'm around and I just was able to go, hey, where should I park? And it just goes and finds the information that you know is there. And pulls it together. And that's what you want from an assistant.

Yeah. Very, very cool. Well, I'm on the product. I think it's fabulous. And it's, uh, it's also, you know, it feels like it lives in a unique part of the consumer ecosystem, something that doesn't feel directly overlapping with the startup today.

No, I think that's right. And I think, you know, we've always kind of expected Siri to be great and understand you and understand everything. And I'm so glad. I can't wait for everyone to get to play with it. Um, because I'm so glad with what it can do, but It also reminded me just how many more opportunities there are in the world.

Yeah. So, so maybe diving right in, you know, there's been obviously a lot happening in the product cycle around AI machine intelligence broadly. I'd say a lot of it, at least for startups, has been more B2B oriented than consumer oriented. What do you think is the sort of state of consumer today?

Yeah, I mean, I think we're in a really, really interesting spot. I mean, ChatGPT in many ways is a consumer app and it's one of the biggest consumer apps that's come on like a storm just in, you know, 3, 3 and a half short years.

Yes.

Because I think it's how do we navigate the world, the information that we have access to, things we want to do, places that we want to go spend our time and our energy. How do we connect with other people? I think right now AI has been so much about productivity, job replacement, work replacement. And I think we have a moment to shift that, which is how does these new tools help you get more out of your own day and your own life and the things you want to do? And it's a massive paradigm shift in interface. We haven't even all process what it's going to mean when everything can start with natural language, when you get options and you're able to explore across a variety of options right there in a really dynamic interface that gets generated for you. When you have an idea and you just want to like create a little app or a website or an experience that's very personal to you and you can just express what you want. And every day I think they're talking about how you can create something from scratch. It's amazing.

Yeah.

You can, you can code in ways you never could before, but Even if you've, something's already coded and is working, you can tinker with it and ask it to be customized in a way that you want. And I don't think we've unpacked much of this for consumer yet. I think most of the hard work and effort has gone into all these productivity tools. And I think there's a massive set of opportunities.

So, but okay, so maybe to wind it back to the social era, we had a lot of entertainment apps, we had a lot of social apps, we had a lot of connection apps. We're not seeing that today with AI. Is that like an ecosystem problem? Are founders just not interested in those problems? Why?

You know, I think one part of it is some of us think a lot of those problems have been solved.

Okay.

You know, we, we had to find ways in the early days of social networking with LinkedIn and with Facebook, you know, and then with Twitter, people didn't have a way to even connect. You know, LinkedIn was like, you're going to put your resume online.

Yeah.

We're like, what? I would never put where I've worked online. That's only if I'm looking for a job. Facebook was like, connect to everybody you've met in high school and college and throughout your life.

Terrifying.

And we did. And then we figured out how much more fluid that makes the world, how you can get information that isn't just brokered through a few channels, but through the people you care about and the people you learn about and the people you're interested in. We've seen a lot of the negative implications of that too. But I think on the positive side, we're so much more kind of connected to who we are and what we're interested in and what we can discover about the world. And so when you think of AI right now, you think of AI as just like short-circuiting things you already do.

Yeah.

And that's a lot of people are building, but they're not exposing us to new ways to connect, new ways to get out in the world, new ways to solve problems that we're trying to solve in our lives, whether it's how-to or everything else. And we're going to have to figure out those 10x things that replace some of those habits that we've already formed.

Yeah, it's interesting. I think a bunch about this sort of great man, great woman theory of history, and it feels like at least the intersection of technology and us as people has mostly been intermediated through groups. As you said, in social, and now we have a chance to kind of be that great man, great woman via this technology. It's a very different shape of it. And that's not necessarily even just a comment on productivity. It's a different shape of entertainment and exploration as well. Like, what are the sides of yourself that are just totally undiscovered still?

Yeah. Well, I think with, with AI, you can go down these self journeys and go down and, and I think one of the challenges with AI is it can also be very self-reinforcing in a way that if you are going down this like very solo journey with a companion app or something, you can get like, in this mindset that you don't need the rest of the world. And I think the, the products that can kind of take that energy and take that exploration and take those groups and take that sort of great man theory and put us in a place where it's like encouraging us to actually do different things and get out in the world and connect with other people and do things in the background that help us be in a better place to do that.

Right.

I think there's a ton of things that still are unexplored.

Do you think we're going to have agents in group chats with us? Like, what is that gonna look like? What are, do we have to establish social norms for how we interact with them?

I mean, I ultimately think that we're gonna have this set of agents or things working on our behalf or, um, things in the background that'll be solving things for us that we'll talk to just like we talk to regular people, but we'll expect different things from it. We'll expect a set of understanding that, you know, I think you understand a lot of me, but you don't understand all of me. And I'll expect my agent to have an even deeper understanding of that. So when I give it hints or things, it'll be much better. And then I think it'll actually help us interact with other people. It'll be like, hey, I'm gonna go talk to Anisha's agent or negotiate this thing. And then we're gonna make sure that you guys are set up to have this great conversation or whatever it is.

Yeah.

Yeah.

It's, it's interesting as a level, it, it like, it's a level of indirection.

Yeah.

Which I think actually has a lot of interesting implications for work. You know, you think like work, there's so much conflation of what is the truth with what is sort of personal to me. And, you know, the likes and the dislikes and the disagreements, even often intellectual, can lead to these sort of emotionally charged situations. I'm optimistic, or at least hopeful, that agents will defuse a lot of that and work can be more about the honest pursuit of truth.

And I hope so too, because I think a lot of it is the best work happens when you've got the right information in front of you and you're actually really able to have the points of views and information. You know, one of the things I learned a lot at Apple was how important, um, the culture of debate was. To get to the best answers, you need to really discuss them. But the only way you can have real debate is to have as much information and thought prepared ahead of time.

Yeah.

And that was really the hard work. There. And I think we're going to have agents that are able to help us be more prepared and ready to have better debates. And when we go down different paths and we're like, actually, we're missing this piece of information, the ability for an agent to go retrieve that for us very, very quickly and help kind of lay out some of the pros and cons or whatever the sides of the issue are, is going to be tremendous.

Do you think there's a tension between agents and apps? Like, are they overlapping? Is, is, are agents like the final app?

I mean, a lot of people have used this moment to go, is this the death of apps? Because you'll just get to everything through a command line.

Yes.

And I'm very much a believer that apps and experiences and being able to visualize and explore information or spend time in an app consuming content or interacting with things, none of that's going to go away. Now, what I think we'll be able to do through these agents, through these types of experiences, is more quickly get to what we're looking for, more quickly get into an app where everything's set up right and I can further customize it and explore explore from there rather than have to start everything with my own set of tabs. But I definitely think that we're going to want to spend more time in rich experiences instead of just back and forth in chat.

Yeah, that's interesting. So is chat an intermediary then, or is it, is it sort of an interface that's having this renaissance and it's here to stay?

I mean, I think it's an incredible starting point, an incredible substrate. I mean, if you think of human language, how do we have most of our conversations? They are dialogues back and forth, but we don't But after we're dialoguing, we then want to explore information and see things in bigger ways and be able to actually experience it. And, you know, I look at things like, you know, games, like an app— chat's never going to replace a game. I'm like, a chat game like Zork, like I used to play when I was a kid.

Zork, great reference.

Like, that is very different than a rich, immersive, experiential game. And so you're never going to replace that, but chat is a great way to kind of get started and to help me think through something before I'm ready to have that experience. Yeah.

And it's interesting. I mean, games is a great way to think of how these 2 interfaces can reinforce each other. Because if you're playing not Zork, but maybe Witcher or a modern RPG, you know, there's obviously, there's these very like graphic heavy experiences, but then there's a lot of fun in just talking to the NPCs and kind of exploring the dialogue and all of that should get a lot richer.

Well, and I think this is a huge area for, for a lot of consumer invention too, is this ability to have these immersive stories and experiences and new kinds of entertainment. Right. Become these beginnings of worlds, but then you're exploring in all new ways. You know, we have a generation of kids growing up that grew up on Minecraft and Roblox.

Yeah.

And they were used to playing a game where they controlled the world, where they chose to join the world. You know, Roblox is this incredible co-experience platform where you get a couple people together and they're playing, you know, games and having conversation at the same time with their friends. And they're going to expect is that group grows up, more and more experiences like that, that are rich, that allow you that sense of control, that sense of ownership, that sense of tinkering, that sense of privacy in this bigger world. But yet the world should be so much richer because of AI and so much more open-ended.

So does that mean that the sort of Gen Alpha that's coming up, the Roblox, Minecraft generation, will they expect to be able to kind of fork and remix all their utility apps? I mean, what does that mean in practice?

I mean, honestly, I think so.

Yeah.

I think that's actually a huge opportunity is Is every piece of software, you know, it's still going to have to do a lot of its core functionality and not every Gen Alpha is going to go, I need to start my accounting software from zero. I need to start my travel planner from zero. But once they've got the thing that does most of those functions, they're going to want to tinker with it. They're going to want that flexibility and the apps that will win in the next generation will give people that sense of ownership at the end.

Yeah. Yeah. Really, really interesting. I mean, it leads to a natural question, which is when it's trivial to create anything, How do you discover things? Is, you know, how scarce is attention? So maybe just talk about distribution broadly.

Yeah. I mean, I think this has been this really interesting wave of the internet of how much, so much of it's powered brand new distribution methods. We both grew up through the social era where most of it was virality and there was a lot of techniques and tactics and testing and learning of how to drive more virality, how to get one person so excited about something they would share it. And sometimes some people overhacked this with how to get one person to inadvertently invite their whole address book too, but that never worked. It was always the intentional viral sharing. This is so great, I want you to join me.

Yeah.

And then we got so many of those that everybody got a little bit jaded and you're like, why are you inviting me to another thing, Anish? And like all of a sudden you gotta stop doing that. And so we had to move into this world, you know, search was the other big wave, search and, you know, being able to be search optimized for Google was a massive thing. What we've seen in the past 5 or 10 years that are really the working methods now are the trust we have now in these new creator relationships.

Yeah.

Sometimes they're true social relationships, sometimes they're somewhat parasocial, but the creators are able to talk about the things that they're doing and exploring and figuring out. And you learn to trust that this person isn't just doing it to make money or an endorsement. They're doing it because they're passionate about what they're tinkering with. And if I like what they're doing, whether it's in beauty or you know, health or restaurants or technology. I follow some really cool Apple creators. Um, they'll go and tinker with these things and talk about them to the world. That's the new path of discovery is you have to be so good that you're getting those people to talk about it. And sometimes you can even pay for those endorsements and try to get that started. But honestly, everything only ultimately works if it's true organic word of mouth. I love this so much. I'm telling you about it at some level.

Yeah.

And everything then you do is you feed a lot into the engine to get to that point where your product's so good it's spreading itself.

So, okay, so let me challenge you on this a little bit because there's something intellectually dissatisfying about creators being the next-gen channel versus the traditional networks. You know, in the old days, the old days, I mean, the value of the Insta network or the Facebook network or the X network is so powerful. And you could argue that the ones who own those networks have sort of maybe abused their powers or used them in ways that benefited them more than the ecosystem. Creators just feels like a weaker form of that?

I find that they're the, the, like, the biggest form of that because actually most creators have earned their reputation based on trust and based on a trusted relationship. Now, I remember in the olden days, you know, it'd be a very famous athlete that would then get up and go, I endorse this cereal. And you're like, there's no way that that athlete is eating that cereal every day. Like, they're just getting paid to endorse it in the commercial. But now when a creator says, here's my health routine, Or here's my travel routine and I went here, here, here, and these 2 I liked and this one wasn't as good.

Yeah.

Like you're actually, if you've learned to trust them and build that relationship and follow them along, you actually believe it much more. And yet it scales all the way down. There are influencers in college, the ones who actually influence their college community, the ones in high school, and there were always that. And so it's not, it is still about community. It is still about influencing your friends.

Yeah.

But it is getting there. I mean, Discord, for example, another good example.

Yeah.

Through all through influence, all through this kind of influence. But it wasn't like these mass influencers. Sometimes it was streamers going, hey, we're playing on Discord now. But most of it was the person who gathers everybody for games would go, hey, I'm trying this new tool. It's way better than what we've been using before. Everybody jump on it. And that one person would bring everybody in their group who they played games with. And then a few people would go, that was really cool. And I also play games with this group. I'm gonna get this group on and then they do that. So that's even, you know, call it, uh, it's not a micro influencer, it's a Word of mouth.

Yeah.

That's pure word of mouth.

Yeah.

But I think it scales all the way down to that and all the way up. Yeah.

That's interesting. So do you think then that that's the channel sort of for the AI-native consumer products, or are they gonna, is it gonna be something like ChatGPT Apps Directory or Codex, or is there some other thing that's going to happen?

Well, I mean, I kind of brushed over this, but we also talked about virality and search, and search was sort of the single-purpose tool. I think we're gonna be back in the same world where people are gonna need this sort of spread, this word of mouth, whether that's accelerated through creators or just through individuals being so excited. Yeah. At the same time, You're going to need to figure out how to do, you know, what they call it, gen, you know, GEO, generative, you know, optimization where the agents talk about you. And when someone's trying to solve a problem, the agents will actually do that. And I think there's both huge opportunity for whether it's the directories or whether the, um, whether it's just getting into the content for the model training or the, the RAG, or ultimately whether these products become really smart and they realize that referring things out makes them better.

Hmm.

And I, right now, I don't think any of the core assistants kind of have that mentality that referring people out, yeah, makes them actually stronger. But I think that's the biggest opportunity.

What do you mean referring people out?

You know, when I start trying to solve a problem with, uh, ChatGPT or Claude, it does everything it can to solve my problem.

Yeah.

Hey, I wanna plan a trip. Hey, I wanna figure out a good bottle of wine. I wanna do this. Yeah. But, but what would be really cool is if it said, hey, I know you're trying to plan a trip. Here's a bunch of things. By the way, here's the resources to go deeper. And, and maybe they'll want to own a few of those verticals, but they can't own them all. And actually being even smarter about knowing when it's good to hand you off, I think will be great. And then it'll create a lot of opportunity for those tools.

And perhaps those tools are apps.

Yeah. I think a lot of them will become apps in their own services. They'll have some chat experience, but a lot, a lot more, you know, when you're trying to plan a trip, you know, chat is great to get some lists of stuff, but then like, it's not awesome. You gotta like turn it into like, bookings and lots of other stuff. And, you know, when you're trying to deal with some financial plan, look, it's great to like work through some problems, but you still need to turn it into like a lot of actual details. And I think those artifacts and managing those are going to be actually better in other services.

Yeah, that's really interesting. So then would you say that creators are sufficient for the AI consumer products to get to scale and we don't need an AI-native distribution channel, or do you think there's still something on the come?

I think this is what we gotta go figure out over the next couple years.

Yeah.

I think, I think the short answer is there's a lot of ways to get to scale today. People can get to 10 million, 50 million users in ways you never could before. And all that really matters is if you get to them, do they actually use you? Do you stick? Do you become part of their lives? So, so I'm less worried about finding just the distribution methods for awareness as much as like, what are the products that are actually gonna move our habits and change us? And Part of being an AI product, you know, today is it has to be so much better than the products that were before. ChatGPT was amazing for a whole bunch of searches that we thought were totally solved through Google.

Yeah.

And all of a sudden you start putting those searches into ChatGPT, you're like, this answer is more than 10 times better. This experience is more than 10 times better than Google links back and forth.

Yeah.

And I think that's why ChatGPT blew up. And I think for every experience consumers have today, with these new possibilities of building things differently, of presenting things differently, of experiencing them differently. One of those things that the moment you use it, you're like, oh my God, I can never go back to the old way.

Yeah. Yeah. Really interesting. So it sounds like you're saying retention is still the most important metric. Is that true?

I mean, I think it's always been this idea of, do you really use the product? You know, and if you can get me to try something and willingness to try, what's really cool about this era, willingness to try, is bigger than it's always been.

Yes, that's true.

And, uh, that's part of what got me so excited to come back to the, the startup side and the venture side was we can actually get people to try new things.

Yes.

But trying new things only matters if people stick and bring it into their lives and it becomes part of their lives in a way that they're so excited to tell about the next people. And everybody is like, how do you get to this many users so fast? And the answer is don't.

Yeah.

Get to a smaller number who have moved their lives over to your product. Yeah.

Yeah.

Who have truly found something new that they are doing that they will not want to go back from.

That's right.

you find that, then it becomes how do we repeat that? And everything I've gotten to work on in my career has been really fortunate to have found its way to that and then build and compound and set the company up to really scale from that.

Yeah, there's some non-intuitive wisdom in that, like Paul Graham's whole thing about monopolizing a really small market being much better than being diffuse across a big market. Similarly, Andrew talks a bunch about the way to build a big network is to build a bunch of small networks, which makes total sense. But You know, we're so overtrained on this, like, go big, billion users. It's hard to do things at small scale.

I mean, the problem is you have to be able to think in both at the same time, right? Like, like the best founders I've gotten to work with figured out how to both have that vision of what is billion user scale and why is the world different when we're actually, everybody's using it, but then go, but I'm not gonna get there today. Today I've just gotta get to this and I have to win, you know, this area and convince this group to become passionate. Yeah. Robinhood's an incredible example where, where they had a vision to democratize finance for everybody.

Mm-hmm.

You can imagine anybody in the world who didn't formally have access to the stock market, to the type of investing products that other people were able to do, to be able to give that to everybody. But that certainly didn't mean overnight you need to be getting every single person to be able to do it, you know, in your retirement account and everything else. It was like, no, it started very small with Hey, if you are actively thinking about wanting to invest or are actively already trading, we should be a better product for you. Yeah. And then, hey, for those people who aren't there, who might not even ever wanna be investing in single stocks, how can we over time get the right to expand the platform to everybody else?

Right.

But not, let's go out with 30 products today, 'cause that would've confused everybody and landed no one.

I'm glad you actually mentioned Robinhood, 'cause so there's 2 companies we sort of talk about as the exceptions in the consumer lore from a distribution perspective. Yeah. One is Robinhood and one is TikTok.

Yeah.

You've obviously been involved in both. So the way we tell the story, you should tell me if this is correct or not, for Robinhood is that they did do paid acquisition, significant paid acquisition, but because it was this very brilliant mechanic of give a stock, get a stock, you sort of had this non-inflationary customer acquisition cost versus being in an auction with Facebook and Google, et cetera. So you were, you know, they were able to scale it and it worked even though it was paid. And historically we sort of say, well, paid is a problem at scale.

Yeah. You know, I think what's interesting with Robinhood is it didn't start paid. They, they started with a product that they thought was just a better way to invest in the stock market. Let's remove the cost of commission. I mean, back then it cost $5 or more a trade with any of the other big platforms.

Right.

And so they said, we're gonna figure out how to give it away for free. We know there's gonna be other ways that we can monetize, including, you know, they found several, you know, mechanisms through the, through the ecosystem that still were better trading costs for individual platforms. Sure. Yeah.

PFOF.

Yep. But, but the key was, It was all word of mouth at the beginning, and they figured out how they were on Reddit. There were Reddit communities that were really excited about it. Anybody just word of mouth on mobile was saying, ah, I kind of want to invest in the stock market. This app is great. And by the way, it also had beautiful design. Like one of the things that you needed when you were dealing with people's money was trust. And so beautiful design, an app that showed that people cared and put a lot into it.

That's interesting.

You know, really helped Robinhood have that initial credibility. Over time, as you're figuring out, okay, how do we grow from here? Tried a lot of things, definitely tried some paid ads and, and you know what? They worked. Free trading was a message that worked for people who were already kind of in a trading mindset.

Right.

But if you weren't, it wasn't. But the referral program was really magic. And the secret to the referral program wasn't— it was, there's a lot of referral programs that were big at the time. Airbnb, Uber had ones like give 10, get 10.

Yeah.

You'll get $10 for getting somebody, you'll give $10. The $10 are the starter for the person who does it.

Yeah.

And like, Those are nice 'cause you have a little bit of a discount, but it's a dollar amount and you're like, is it worth $10? Am I trading off my friend for $10?

Yeah.

Robin, here's $10 to invest. What are you gonna invest in for $10?

Yeah.

You know, nobody knows. Can't, you know, at that point you couldn't even buy a share. So, so the team came up with this mechanic to figure out how to make it give and get a share.

Mm-hmm.

An actual share. And then was able to use the fact that that share was a lottery. Sometimes it would be a $2 stock, sometimes it would be a $100 stock.

Yeah.

And enough people got the $100 stock that they raved about it and that made them very, very excited. Yeah. People get the $2 stock and they're, oh, I have a whole share of stock. I can sell that. I have $2 in my account that I didn't have before.

Yeah.

And Robinhood's ability to balance that, um, made it a very, very cost-effective mechanism. So look, when you're doing paid ads to bring people in who are already susceptible to the message, when you're bringing in this referral program to then have people share it and advocate it. And again, referral is not quite as pure as word of mouth, but it's pretty close. And I'm only gonna, I'm only gonna bother somebody if I at least believe in it.

Yeah.

Um, or the incentive for me is so good, which they know the incentives, they're getting the same one.

Yeah.

And so it really became this powerful mechanism. And then the brand just kept building and building and building, and people started to trust it more, and the design got even better. And, and all of a sudden, one day you're like, whoa, this is no longer this little upstart financial thing, but every other company is dropping commissions because they now see the threat. You know, Schwab did it. Um, Ameritrade did it. Schwab then went and bought Ameritrade. Like it transformed, you know, the brokerages.

Yeah.

All from this little plucky, let's try to democratize finance.

Yeah.

But get the people who we can early.

Well, it really demonstrates the power of a founder-led company even being post-public.

Yeah.

'Cause now you see what Vlad is doing, sort of agentic trading stuff. I mean, he, it seems like he's leaned the most into it. Yeah. Out of everyone.

I mean, I, I'm just blown away by what they've done with the company. You know, since I've been gone, I left at the end of 2019 and it's turned into this incredible, realizing the vision, multi-platform.

Yeah.

It's no longer just trading, it's, you know, retirement, it's credit card.

Yeah.

It's got some banking components and they have the, like all these products that bring in the same care, design, trust for the user.

Mm-hmm.

Truly trying to bring value. And Vlad has just done an incredible job navigating the seas. It went through some real bumpy times too. I mean, so many great companies have had these existential crisis moments, and Robinhood's had a few, but both between some crazy downtimes that happened at critical moments in the stock market, what happened around GameStop that felt like that almost might be existential for the brand, but they came back to their core values. Great design, great product, trust for the user, something people are proud to talk about and share, and they've really been able to win people back and grow a ton from there.

It's amazing.

It's awesome.

They've done so well. Okay, so we've established that Robinhood was a referral program. So now talk about, maybe just tee it up, talk about Musical.ly, TikTok. Not everybody may know connection and then how they grew, because they are the other great exception, which is anytime we say, well, paid doesn't really work at scale, people say, but TikTok.

Yeah. Well, and this is a crazy kind of multi-part story. You know, Musical.ly was a group of a small team in China that was trying to build for the Western markets. That was actually their initial vision. They started trying to build an education app, hadn't really worked. They had seen some fun things happening around lip syncing. And so they decided that they were going to build a little lip sync app. And they built this and they put it on the App Store and it said Lip Sync. And they were seeing spikes every week because Alex, who was the founder of Musical.ly and actually went and ran a lot of TikTok, was telling me this, that they were so confused. And they found out finally that like the spikes on Thursday night, I think it was Thursday, was because of when Lip Sync Battle appeared on like MTV.

Okay.

Lip Sync Battle would run and then people would go type Lip Sync into the App Store. And they download Musical.ly and they start creating lip syncs.

Interesting.

And Musical.ly though designed their product in 2 really smart ways. The first was it was really easy to take your video and still go post it elsewhere. So you could create this great lip sync video and post it elsewhere. But the second thing was they also curated a feed of the great lip sync videos. So when you'd open the app, you'd swipe through, you'd add some friends, and you'd create your video and you might go post it on Instagram. And so that loop was, Create something you couldn't create anywhere else at the level of quality, then be able to post it out to Instagram and your friends, you'd start following your friends. So you'd see what your friends had posted alongside some other really cool stuff they were curating. So they really created this loop. And then they found out that there were a bunch of, I'd call them like B+ stars on Instagram who weren't quite getting enough credibility and were only posting pictures. And Instagram video was just coming out at the time. And they were like, let me use Musical.ly to create some cool content for my Instagram feed.

Got it.

And then on Musical.ly, they'd become the number one stars. They'd become the A-listers, the ones with the most followers, 'cause they were getting featured in the feed. So now you had this loop where the influencers were posting a ton of stuff on Instagram. So you were starting to see Musical.ly on Instagram. You're starting to have a better experience in the feed, and people were talking about it with their friends, and it all started to compound. And the other secret was there was always a little Musical.ly bug at the bottom of the Instagram video.

Right, right, of course.

And you could actually download the video. It was like unheard of at the time. You'd create the video in Musical.ly, download it, upload it to Instagram.

Yeah.

And it would keep the Musical.ly bug.

Right.

People weren't cutting that out.

Right.

And it created this, this great loop. And the company got really big. We got to, you know, tens and tens of millions of users daily, you know, hundreds had tried it and it was in incredible shape. And then they started thinking about whether or not to keep going independently, raise a lot more money or sell and got this incredible offer from ByteDance to go and join ByteDance. And ByteDance had had a Uh, product Douyin that was very, very large in China. And so they had kind of both modeled some after Musical.ly, had had their own invention and had this great AI team. And they were like, we can help Musical.ly scale even more worldwide.

Yeah.

And so they, they bought Musical.ly and the team that was already in China became part of that, that ByteDancing. And then they, they spent about 6 months or so kind of tinkering, rebuilding the product, moving Musical.ly and Douyin into a shared codebase.

Yeah.

And then they relaunched it with ads, with dollars, because what they figured out was to get from here to where we want to go, we need to pour a lot more into it.

Okay.

They got, they spent, they were probably one of the biggest spenders on Facebook, Twitter, kind of every social media to get people to download TikTok at that time. But the product loop was so good that everybody retained because you would sign up for it. And all of a sudden with a couple of swipes, you were deeply personalized. Your friends who were already there on Musical.ly were coming back and trying new things. So you were seeing that and it was really energizing and the whole loop worked. Now it still took them pouring a lot of money in to get that distribution, but the retention really worked. And I joke at the same time, Quibi was pouring similar amounts of money to get people to download Quibi, which didn't totally work.

Didn't work as well. Yeah. Well, okay. So is the lesson from that then that if you have a highly retained working loop, you can do paid and it's okay?

I have no problem with people doing paid if they understand it. Paid to both get people in who will become very good retained users and who will spread the word for you can be a great catalyst. What you can't do is spend a lot of money to burn 10 people to find the one who actually stuck with your product and think that that's going to scale because that only gets more and more expensive.

That's right. Yeah, it's interesting. It's sort of Adjacent to this is a debate we've had, which is on one hand, it's amazing that consumers are willing to pay for software and in a big way. If you look at the top ChatGPT SKU, it's $250 a month. The top Grok SKU is $300 a month. Gemini Ultra, I think, is $300 a month. So wow, consumers are paying for software. On the other hand, because there's no zero marginal cost of distribution, inference costs money, you can't really build a mass market free product without a massive balance sheet. Better or worse? Is it just changing the shape of startups? I mean, what's your view?

I mean, I think it's, I think there's a willingness to pay and necessity to pay is definitely shifting. You know, we had this period, you know, 15, 20 years ago where everything was shifting to the cloud and all of a sudden the startup cost for a company became much less than ever before. You didn't have to buy a bunch of servers, rack 'em, and, you know, sure, serving costs a little bit of money and bandwidth and compute use, but it was actually Relatively low for what you thought you were getting as a startup. Obviously now, if you're running a ton of inference, you need your customers to, to be willing to pay for it. But I think they are. And the ability to pay in apps is so much easier than it's ever been. You know, if you, you know, getting payments is easy if you're providing value.

Yeah.

And I think you have to, I think it hurts companies that need massive network effects because part of the benefit of these massive network effects was Free distribution to everybody helps you get there. And so haven't totally figured out what that balance is, but I think that's what I'm excited to go work through over the next few years with various startups.

Yeah, it seems like one of the things that could really change that is if iPhones get really powerful GPUs.

That's right.

It would seem crazy for that to not happen at some point in the future.

But I think we're going to be figuring out with the models, even that you can run on iPhones or on laptops, how to push more and more of the complex or the the less complex inference that needs to be done onto device to lower the cost. I mean, there's no reason that everything you're ever trying to do, whether it's refining a title or just rewriting a paragraph or answering some basic questions, those can actually use models that are incredibly powerful that come out on the devices now. And then you only send the more complex stuff up. And I think both from the enterprise side, you're gonna see a ton of that. And I think from the consumer side, I think really smart founders are gonna optimize that too.

Yeah, it'll be interesting. I know that Karpathy has done a bunch of work, a few folks have, showing that you can actually train and run models on 386s and 486s.

Incredible.

So I'm sure there's also just a lot of good old-fashioned engineering that will make the models push more and more to the edge.

That's right.

Okay, so one topic that we discuss a lot is, are the labs just going to win it all? Should we all just go home? And I know it feels like we have a version of this conversation every time there's a new product cycle. What's your view?

So I think a funny way to think about this is, 20 years, everybody was like, oh, well, won't Google just do it?

Right.

And then all of a sudden, Google could do everything. And certainly there's a bunch of categories that Google has absolutely expanded on and done great. There's a— 20 years ago, there's a ton of startups that have grown and become massive companies since then. You know, uh, 15 years ago, um, 2010 era, everybody was like, well, isn't Apple just going to do it all? It's going to be everything's going to be their own mobile apps, right? Have this great world of, you know, mobile-enabled apps that we could have never imagined before and these social apps. Um, but about 10 years ago, I kind of started thinking that too.

Yeah.

I kind of started thinking about, I'm not sure if there's a world in these networks that are now 1 billion plus users.

Yeah.

If you can truly build another consumer startup from scratch. It's part of why I actually left venture to go and I said, I just wanna build again. And I went to Robinhood.

Yeah.

That led me over to Apple and said, gosh, I'm not sure if you can actually build something big enough. In kind of pure consumer areas. And I'm not sure if I was wrong. I mean, ChatGPT is obviously a massive— What about Discord? Well, I had invested in Discord and we had just sold Musical.ly. And I was like, those are going to make it through. I joined Robinhood, which I said, that one's going to make it through. But starting from scratch, you have to get enough momentum to get to that billion user scale. I was like, I'm not so sure. And this is part of my own, like, if I didn't really believe that I could find that path in venture, I was going to go and work on them instead.

Yeah.

And, you know, part of why I went to Apple was to go learn and work at the biggest scale you can and try to deliver that value to everybody's pocket. But I think right now, I think the world is so open again.

Yeah.

I think that the main chatbots, the main assistants are going to do a lot for us.

Yeah.

We're going to rely on them for a lot of things, but I think we're going to expect everything to work like them. And we're going to rewire so much else of what we do in life and our services and the way we get things done. That the opportunities to fill all of those areas of our lives and help us connect and help us get information that we want and help us explore things are not just going to go to like, there's only 2 products we ever use.

Yes. Yeah. Yeah. No, no, look, I totally agree. And I also think there are ways that startups can play that incumbents can't and labs can't, you know, to me, Companion is such a great example of this because I feel like Companion touches on everything that every committee at a big tech company doesn't want you to do. You know, there's disagreement, there's probabilistic behavior, there's sexuality, there's, you know, There's just all of these ways that you can interact with a product that are very human. And as you know, when you're at really big scale, it's hard to ship things like that.

You know, I think it is, although having been part of, you know, one of the biggest companies, learning to ship things that matter and make the world better is kind of the fundamental value of what we do and the fundamental value of what you can do when you get to work at that scale. But when I think about things like the companion products and entertainment and how AI can sort of take us on these journeys, I think it's incredible how much different it can be than everything we've had in the past. We've never had an experience like that. On the other hand, I think it's also a high level of responsibility.

Yeah.

We have to be very, very careful with kids, with safety, with sexuality, with letting people who might be going through any mental challenges to feel like they're getting something served through a product that isn't bringing them back to the community or helping them get the help they need.

Right.

So I think we have to very much find that balance, but then Once you accept that that's just the baseline, there's so much to explore that not every product wants to be. You know, lots of products don't want to be your best friend. I mean, if you've used the new Siri, it's incredible, but it's very much trying to help you get things done. It's designed to help you get things done. It's incredible helping you get things done with the experiences and the content that you already have and what it knows about you and what it can go access in the world. But it doesn't need to be sycophantic and it doesn't need to be like, you look great today, and create that personality.

Yeah.

But there are other experiences where you actually might want opinions on a wardrobe and how you dress and give you tips and treat you a little bit more in that human relationship. I think it's a very different product.

Yeah, it's interesting because if you almost unbundle the two, you— there's a product that's just about getting things done, you know, which sort of almost has no soul by design.

Yeah.

Um, and that's great and we need that. But then there's also these products that might mirror things that we do, you know, in the rest of the world, which is I might join a volleyball team because I'm looking for friends and volleyball is like a level of indirection to building new adult male friendships. So I think there will also be companion products that look like that.

I agree 100%. I think there's so many that can help us do more of that and explore the world and use, you know, in the background, do things with other people and try to figure out what makes sense for you. And by the way, I would argue that even the that get things done has a soul, but it has a very different soul. It's not a, it's not trying to replace any human relationships. And none of this really is trying to, it's all trying to augment relationships. It's not even trying to like make you perceive it in some sort of human way. It's trying to like really be a tool. And that's what we are as technologists. We're toolmakers. We make things and they can be net neutral and it's our job to infuse humanity into them no matter what we do.

I love that, man. Yeah, it does feel like this is a technology to explore our humanity. And we've never had that before, actually. You know, we've had a lot of technology to extend our intellect, and that's let us do incredible things. I mean, SpaceX this week is the best example of it, but we really haven't had any technology that's sort of, I don't know, helped us be a little more self-reflective. Social products are fascinating, but sort of exploring us as groups instead of us as individuals.

No, I think that that's a really profound way to think about what we can do with this new technology. And so much of it is helping you be better. I mean, I've seen these great little apps that are like helping people learn to date by practicing first date conversations.

Right, totally.

And like, it's one of those things that it sounds funny and we're probably both, you know, older and not in that lifestyle right now, but like, if you're nervous about something, being able to like practice is incredible. I mean, I haven't done one of these in a long time. Yeah, I have to I have to admit, I did use a little bit of going back and forth to ChatGPT to just practice talking and getting back into this kind of mindset, because that's how these things help us be better.

You know, it would just be so funny and ironic if the result of all this technology is that Gen Alpha, they're just like the most incredible social apex predators. Like the art of conversation is back, you know, it's like the 1950s.

I mean, I think that would be amazing. And I think we, We always forget, but that's the generation they've grown up with. They've grown up fully online, fully mobile. And we've seen some of the downsides of this.

Sure.

I mean, I think there's a lot of, there's a lot of things happening right now to even like prevent social media from people under certain ages, mostly because we haven't all done the work to make these as safe and as additive as possible. But on the other hand, this generation is going to know how to adapt and get the most out of this technology and use it hopefully in better service of others and hopefully Look at some of the examples of what's gone right in the world that they can live up to, and hopefully what's gone wrong in the world that they can say, it's our turn to go fix.

Yeah, it's fascinating. So this class who's graduating right now is the first class that grew up, uh, had ChatGPT for all of college.

Yeah. Yeah.

Which is really, really interesting, you know? And actually what I think that ChatGPT lets you do, or, or the sort of broad AI is actually, you know, what they called the greats in college, which is studying ancient Roman, ancient Greek, sort of, uh, everything related to it. It's a sort of individualistic, very abstract intellectual exploration of things just for the sake of it. And it's a lot of what the technology lets you do. So I hope they're as enthusiastic about it as we are.

I mean, I really hope so. And one of the things that's really heartwarming for me whenever I talk to kids is when they tell me, hey, I don't use it to do my work because honestly, I just know that like, that's not what I'm going to get graded on. I might get caught, something else. What I use it is to help me make my work better. And the fact that you can turn to something like a ChatGPT when you've done some writing and go, how do I improve this, is such a powerful tool to have in your arsenal. I mean, that's one of the things that, that I've been using with new Siri on my, my Mac is, you know, you write something and they just go, how's this sound? How could I spice this up? How could I make this feel, you know, get my point across a little bit better?

Yeah.

And the tips and the things it comes back with Really help you go, okay, that's actually a really good tip. I'm going to put that into what I'm writing. And when you're using it that way, these tools are very additive as opposed to replacement.

I totally agree. Yeah, it's fun. Somebody that we both know, Eugenia, who you might've actually introduced me to her or at least whispered a good word in her ear on my behalf. Thank you for doing that. We're so proud to work with her. She just said this thing that was so profound and I keep repeating it, which is, look, people are not trying to save time. They're trying to spend time.

Oh, I love that.

And the entire sort of original sin mistake we're all making in assessing these companies is that the average consumer doesn't know what to do with the blinking cursor. They actually need to have things pushed to them so that they actually can explore and not have to be so high agency.

Yeah. I mean, I think this idea of time well spent is exactly the right method of what consumers want to get. And this is why building for consumers is so different than building for enterprises is you can't just start me with something and I'm like, I got a problem, I gotta solve it. Give me a blank cursor. I'll just dig into it. It's what are the ways that you wanna best spend time? And what we've learned is there can always be a lowest common denominator here. A way to spend more time is more scrolling, more anything. Just keep going infinitely, or just keep pushing this button in this game and see what happens. Like, Yeah. you can pull people and lull people into that, but the best products can be smarter and can actually take us from from, you know, idling away time to actually spending it better. And, and, you know, you, you want to say, hey, to an AI, like, hey, I'm a little bored, or I need some ideas of what to do. You want to, like, spark you and give you motivation and encourage you to go do things. Yeah. You know, either it's get out in the world or just, you know, do something new or think about something or be entertained. You know, it doesn't always have to be work or learning.

Yeah.

It can just be fun, but still things that, that add value where you reflect on it after that and you're like, I'm really glad I did that.

Totally amazing. Okay, well, maybe to bring us home, uh, give us your kind of request for startups. You know, what are the areas that you're personally most interested in? What do you think's underexplored?

I mean, coming out of, you know, having worked inside of Apple for a long time and at Robinhood and coming back to venture, I'm thinking a lot about what is this next wave of all the ways that we connect as people.

Mm-hmm.

Spend time well, whether that helps us, you know, things that help us do things like travel better, um, you know, manage our finances better, manage our health better, get things done, learn how to do our how-tos, you know, and get through that list. All these ways that we can use all this new technology to help us get through our lives.

Mm-hmm.

Spend it really well and connect with other people. Pretty open-ended because so much right now is built for the productivity and the enterprise, and I can make this team more effective and here's 20 workflows and here's 30 prompts that help everybody. And I'm like, what are the things though that make me so happy to go, I'm using this every day and you should too. And you should too. And I can't wait till we all are because it makes it better for everyone.

So Josh, take us through what you're going to work on here at Andreessen Horowitz.

Look, I am so excited to be back in the startup world and really get this opportunity to help Andreessen Horowitz portfolio companies figure out their way through this consumer landscape, figure out how to grow, figure out how to tell their stories. Figure out how to build their products. And then I'm really excited to help all the future Andreessen Horowitz portfolio companies. Let's figure out who they are, what they're doing, how I can help them, how we can help them figure out their paths. And then, you know, when it's the right time, make sure that they come in and we, you know, as an Andreessen team, really fund them.

I love that, dude. Talk about maybe for a moment, what are the kind of specific challenges you've worked through with companies? You know, is it about distribution? Is it about M&A, what are some of the ways?

I mean, I think the answer is that the stuff I love helping with the most is thinking about the product and the product loop. How do you onboard people to get to that retained experience? What do you need to do to get people all the way up the ladder? How do you let them bring other people into that network? What does a core product do for them every day? And how do you evolve a single-use product into multiple uses and scale that as a product platform? So helping people right in the heart of what that product and product strategy is. But also happy to help talk about how you organize your product team and growth team. And it's completely changed in this new era. I mean, everything is different. Things are more different for a product management function and an engineering function in the past 6 to 7 months than the past 30 years of my career.

Yeah.

It's incredible how much has changed and navigating that, knowing what worked in the past, but translating that to modern times. I'm excited to do that. And then obviously helping companies figure out their own paths, whether that is M&A. There was a period earlier this year where I had a couple different friends who were going through M&A processes, and I was talking to them multiple times a week, helping them just think through the options, how to set up right for every meeting, how to do this. And then lastly is how to tell your story. And one thing I've gotten this incredible masterclass, getting to work at Apple, is been right in the heart of how do we tell a compelling story to regular people? How do we take this world of AI and intelligence and distill it into products that we think people would want to use and bring into their lives every day. And I'm super proud of what the team put together as this most recent WWDC and how we've told the story of Apple Intelligence and Siri AI. And I can't wait to bring more of those lessons to companies as well.

Incredible. How can people get ahold of you?

They'll be able to email me at jelman@a16z.com. Find me through all the social medias. Find me through a friend. I'm really excited to help.

I love it. Josh, welcome to the team.

Thank you. Excited to be here.