📰 Key Takeaways

Airbnb rolled out new AI search features in this week’s fall update, a significant move for a company that’s always been cautious and slow about adopting AI. Co-founder and CEO Brian Chesky has previously said that pure chatbot interfaces aren’t a good fit for travel discovery, and in this interview he reaffirms that his stance hasn’t changed.

Chesky points to two big problems with chatbots. First, they’re a poor fit for browsing or shopping scenarios, since they can only offer a handful of options at a time, forcing users through multiple rounds of back-and-forth before getting results. He cites travel research showing that people actually get more enjoyment out of planning and anticipating a trip than from the trip itself, so you shouldn’t sacrifice exploration, browsing, and that “dreaming up a trip” experience. Second, chatbots are designed to serve a single user at a time and aren’t collaborative — but Airbnb’s use case is often a group of family or friends planning a trip together. So over the next three to six months, the company’s job is to explore interfaces that support “multiplayer” AI, letting a group of people (like a group planning a trip together) use it at the same time.

Chesky also doesn’t think the search interface Airbnb just launched is the final form — he expects something will eventually emerge that sits between a chatbot and the current interface. He mentions startups like Monogram and Wabi that are experimenting with “generative interfaces,” where the AI generates the screen in real time rather than designers locking in a fixed layout ahead of time. He thinks the future will be a hybrid of deterministic, pre-designed interfaces and generative ones living side by side.

On top of that, regarding the rise of consumer-facing AI agents like Instinct and Muse that act on a user’s behalf, Chesky says Airbnb needs to build out better technical infrastructure to make the platform more agent-friendly, and he believes these kinds of agents could become a traffic source down the line.

💬 JudyAI Lab’s Take

What stands out here is that Chesky — the CEO of a company that’s always been cautious about adopting AI — is launching a new AI search feature this fall, yet he’s doubling down on the idea that chatbot interfaces aren’t right for travel discovery.

This points to a design principle worth chewing on for any AI builder: the interface should serve the use case, not just default to whatever conversational-box format is trendy right now. Chesky notes that chatbots can only surface a handful of options at a time, forcing users through multiple rounds of dialogue just to get results — which kills the joy of browsing and “dreaming up” a trip. On top of that, chatbots are built for a single user, with no real collaborative layer, but Airbnb’s use case is often a group of friends or family planning together. His bet isn’t that one chatbot format eventually replaces everything — it’s a hybrid future where pre-designed interfaces coexist with the generative interfaces startups like Monogram and Wabi are experimenting with, while also making the platform more welcoming to consumer AI agents.

Next time you’re about to design an AI product interface, it’s worth asking: is your use case a single person making a decision, or is it more like trip planning — something that needs multiple people dreaming it up together?

📅 Source Info

🔗 Further Reading