Airbnb CEO on Why AI Will Create a New Era of Consumer Products

Good. How are you guys doing?

Yeah, I mean things are really good. We've accelerated growth for the first time since the pandemic. We grew 10% last year in revenue and this quarter we announced that revenue was 18%. So from 10 to 18% which is a pretty big acceleration. And marketplaces are really really hard, right? Like it's kind of like gravity. Once a marketplace at our size doing around hundred billion in gross bookings a year starts to come down, it's really really hard to tip that curve. So this has been a pretty big feat for the team to be able to do it.

You know, it's kind of funny. A number of years ago, I noticed that as we got bigger, we started losing a little bit of that startup intensity, that startup energy. And I asked myself how can we get that energy back? And we basically took a very small team, we named the team Project Hawaii. The name doesn't really matter but I took a very small team of people and we said we're going to focus on a very very small service area and we decided to focus on conversion rate, the guest journey. And I basically tried to work with a team like as if we're at Rouse Street. Rouse Street was the apartment that Joe and I started and I said we're going to act like a startup. And the team just grinded really really hard. We weren't working like a big company. We were a very small team grinding really hard focusing on obsessing over the customer experience, really looking at the data and we really got a lot of points on the board. Then we really started taking these pods and we really started working with the teams trying to coach them how we worked in the early days and I just think the pace increased, the intensity increased. We really tried to bring in world-class people onto the team. They got very very focused. We tried to get all the management and bureaucracy out of their way and it's been a couple years and in fact we were doing this last year but one of the things is as you know financial results are lagging indicators especially when you're a big company. So it takes sometimes a while to get the financial numbers to reflect what's happening inside the company. But I feel like we're a startup again more than ever. We feel much younger and smaller than earlier. And I think with AI that would be the other thing is 60% of our code is now written by AI which is twice our benchmark of our competitors and peers. And it's really really helping us. AI I think is a huge boon to us. I don't know if it's helped the OTAs but it's helped us with customer service. The cost per customer service ticket is down 40%, people who contact me the AI solves the problem for them and we brought it through the entire journey. So everything is really accelerating.

Yeah. So it's a great thing. I mean really what I'm talking about has played out over like really five years and the term founder mode what it really meant that Paul Graham wrote was about me skipping layers of management and going into the details with teams. And instead of trying, I mean here's my advice. If you're a CEO or a leader of a company that's big don't try to change the whole company. Try to change a corner of the company. It's kind of like don't renovate the whole house. Pick like one room and make it perfect and then go room to room. So, I really told teams like, hey, I actually didn't replace the team. I took the team that was already working. I handpicked some people on the team, but I really just taught them the pace and I would review work very regularly. So, I'd sometimes review the work weekly or even daily to just teach them a level of intensity, a level of perfection. And it was unfamiliar and uncomfortable. And I will tell you that not everyone liked it and some people didn't stick around the company because they didn't like that way of working. But those who stuck around and the vast majority did, they realize, wow, actually when the CEO is involved, it's actually easier. There's less bureaucracy. I try to make the work better and I try to clear all obstacles. And so I basically did that group by group. And the great thing is I was in massive number of details reviewing everything for years. But eventually it's like muscle memory. It's like an instructor or a coach. You have to teach them. But once they learn and they have the muscle memory like a golf swing, you can step back and now it's muscle memory. So now I don't have to be in all these details all these teams. But I would just say a couple things of people listening like great leadership is presence not absence. I think a generation of management consulting or management school management teaching taught us that CEOs should trust their people and get out of the way. And I don't think trust and getting out of the way are the same thing. Again, I think leadership is presence. And you should actually be partnering with your people. You should be on the field with them. If you're a cavalry general, you should be on a horse. You should be on the battlefield. You're not overseas somewhere else just writing out blueprints. You got to be on the field with them. And you got to be leading from example. I think leadership from the front, not from the back. And so these are just some of the things we do. And I think AI is going to create the equivalent of an AI founder mode, which is now you got to be even more hands-on. And I don't think in the era of AI there should be any pure people managers because you're so close to the details, to the data that everyone has the opportunity to be hands-on and it's hard to imagine only managing people and not agents. So I think this hands-on approach is for everyone inside of a company.

Yeah, and it's slightly scary because I've told, I had a meeting with our team recently. I said that 26-year-old me and Joe and Nate could f us up if we wanted to. And so I told them this is what I would do. That's because you understand this market better than anyone else. You know the key drivers. Like I don't think any 26-year-old off the street... Hey, I'm paranoid. I think that if every me sits still, I think a different group of 20 or 30 year olds could also disrupt us. And I think that's true of every one of us. And maybe, by the way, maybe that's not true. And maybe that's not reassuring because we do have a brand that's a noun verb. We've got a global network effect. I think it's actually hard to build a network. That may not be possible but the software and the app, I told our team like we can't sit still like our app is beautiful it's really nice but we got to be in a world of AI native and here's a key point I would make I do not think a chatbot is the right interface for travel or e-commerce that might be a radical statement. ChatGPT launched third party apps last year in March they shut them down I don't think a chatbot's the right interface it's got four or five problems the first problem is it's text-based, photos are an afterthought. Second is there's no direct manipulation. You have to type every single prompt, which is fine for a conversation, but you can't add filters. You can't click around. The third problem is it's hard to compare. A lot of e-commerce and travel is comparison shopping. If you have thousands of options, the AI has to know exactly what you want to be able to show you one or two things, but you usually want to see more choices and you get lost. And most AI is single player. It's not collaborative. Let alone the fact that Airbnb it requires people to have an account that 85% of people send a message. So, what I'm trying to do is...

Yes, I agree. That's the future. And that's not a chatbot what you described. That's not a chatbot. It's going to be a completely different interface. It's going to be, well, I guess you'll have to wait and see, but I think the future are not apps. The future are agents, but I don't think they're going to be text-forward. I think they're going to be really rich user interface. And so I think the current chatbot paradigm, what you're describing, it can do it. I just don't think it's the best way to do what you just described. I think there's a much more immersive way to do that.

I really appreciate them picking on someone else for once. But, yeah. No, absolutely. I think there were entire, actually when ChatGPT launched the third party apps, their stock probably went down like 7%. And by the way, I thought it was a really good idea for them to do third party apps. I think it could have been successful, but they would have needed a richer SDK for it to work. And it would have, like the app store, Apple's app store was good for Airbnb and it was good for every company because they had a really rich user interface. Maybe this is the point I'm making that imagine using iMessage on your iPhone to do everything when in fact every other app has a unique interface. So what I think is e-commerce you want a very rich user interface. It would be agentic. You would be able to have a conversation with it. You can talk to it. It could talk back to it. But I think the point is it has to be more visual. I think a text-based interface is for some solutions and a chatbot that's visual. Yeah, that would work. I'm just saying today's chat bots aren't the right solution. But the answer to your question, I had a couple new employees like one of our new team members is a guy named Ahmed who is the CTO of Airbnb now. He led the Llama models at Meta. And one of the comments he made to me was he said, 'Wow, yeah, Airbnb is so much more than the app.' And in fact, the app that you see is like 20% of Airbnb. I mean, we have typically four or five million people staying in Airbnb every night in more than 100 countries around the world. So, there's a lot more to do around payments, around customer service, adjudicating everything. We have a $3 million guarantee against theft or property damage for a million homes a night. That's $3 trillion. There's just so many types of things around managing five and a half million hosts. We have a host app. So, there's a lot of things that are beyond the guest app. I actually think the guest app would be pretty easy to copy. And I think in an age of AI, you can make a better app than ours and we want to make that app before anyone else does and we want to be agentic. But I think the key is most of it is not the app that you use. It's mostly the off-on experience. It's the operation.

Yeah. Well, before you say it, one of the things is the last time I was on, I did make a comment and I'll make it again. Almost every AI company is an enterprise company. And the last stat I think I said on TBPN was like 3 months ago and the stats haven't changed. I think it was 175 companies in YC batch and I think 16 were consumer. That trend hasn't changed. Almost everyone is going into coding. Almost everyone's going to enterprise. I think that's great. My whole point though is that the consumer experience hasn't fundamentally changed that much beyond a chatbot. And I think the consumer is a massive opportunity for AI and I think it's going to need to be a richer user interface. I think people that want to do more things, they want it to be, I mean the modern AI today is text, photos, and buttons. Mostly and some videos. And I think there's a more breakthrough visual paradigm that can be much more immersive. And with the new image and video generation models, you could do something so much more immersive. So the exact example you gave of trying to book a hotel, yeah, you can do it on chatbot, but there's probably some more breakthrough way to imagine that, to visualize it, to see the neighborhood, to see the map, to be able to talk to it, to understand where it is, to be able to compare photos, different hotels. There's something, I think, richer on the horizon.

Yeah, maybe the simple framework is think about a travel journey. Step one is destination discovery. Where should I travel? Step two is flights. How do I get there? Step three is where do I stay? Airbnb or hotel. Step four is what do I do when I get there? Restaurants, activities. And step five is typically logistics, car rental, services. And then step six is you're in the city and then you might want to do things spontaneously. I think LLM or let's call it the LLM technology applied to, I'm arguing a slightly richer user interface than typical chatbot will be revolutionary for step one, step two for destination discovery and flights also because flights are not very hard to build there's just three global distribution systems anyone can pipe in an API and have a flight booking app pretty quickly and so I think the LLMs are really really good at destination discovery if you want to say like hey I want to go somewhere that's like Paris but it's a little more affordable, it's good to go in August. I like Opera. It's going to give you a very rich suggestion of places to go. Now, I think the current chatbot doesn't have really rich maps. I think it could do a lot more visually. I think eventually chatbots or with this new interface could be very video-based. So, imagine a chatbot that was actually video-based or very photo-based rather than a little less text based. I think that would probably be what you'd want for travel. I think what I'm describing would disrupt travel more than the current chat bots which are more acting like Google sending referral traffic and they actually the referral traffic from these chat bots are converting higher than Google and so actually the chat bots are actually additive to travel companies. To really disrupt the travel journey it would have to be much richer.

That's right.

No. Hipmunk.

Well this is such a good question. I think there's like three or four factors going on. The first factor is I saw a tweet recently that I think there's like 60 new AI labs being formed and I've met a number of these researchers and a lot of them are interested in doing the same things. And I think what I'm noticing is most these teams are purely AI people. They don't have product people or designers on their team. And they're picking things like science. They're picking things like coding. They're picking things like we want to create a different type of model. This is great, but I don't know if all 60 companies should be doing the same problem. And so what you're seeing is you don't see a big focus on people wanting to do consumer. I think some people think that the AGI can just figure out consumer. I think that's maybe overly simplistic thinking. I do think you have to have a point of view about consumer. So I think that's the first point. I think the second point is Silicon Valley's become more vibe and trend based than when I came to Silicon Valley 2007 although it was back then. And one of the things I think when people see enterprise companies doing well they go into it and I think there's this natural flywheel. The other thing is I know in Y Combinator for example we teach the YC companies to use the other companies as their customers. Now back in the day we would get them to use our product as consumers. But I think everyone's figured out actually it's way more efficient to get them as customers. Five codegen startups all being like you code with me you code. Yes. Exactly. And so I think enterprise makes a lot of sense until everyone does enterprise and then suddenly everyone's saying consumer is hard but you know what's also hard like competing with 10 other companies in enterprise doing your idea too. That's also hard. I think additionally the image and video generation models are having a, I think image generation is almost like a Claude Code breakthrough. It's so good now and the video models that I think we're about to enter this new era where suddenly going beyond text is possible and I think you're going to even be able to see real time AI generated interface in the next year or two. So my prediction is this has been the era of enterprise even ChatGPT's breakout success I would predict that most of the revenue is going to be going to Codex that's my, I might be wrong but just given how much money is going to Claude and I think in the next two years you're going to see a massive revolution in consumer and I think you're going to need a couple companies to lead the way. I think we need reference points. I mean even me saying the chatbot's not the interface people probably asking well what is and someone needs to be the one to do that. I mean we're going to try to do our part. I'm not sure Airbnb will be the company to blaze the trail, but I think someone has to. I think in another era, it would have been Apple, like the Steve Jobs era, Apple would have done that. Maybe Apple will do that, but they would have been typically the company that would have done that.