Brian Chesky places Airbnb's next AI search somewhere between a chatbot and the version that just shipped

The chatbot shopper, as Brian Chesky describes him, says "Book me a flight, I don't want to look at it." ^[3] Chesky says Airbnb's guest is a different shopper. "With Airbnb, they enjoy looking at the product," he told TechCrunch ^[3]. He supports that with a claim about travel research. "Travel studies show that people actually derive more pleasure in planning and anticipating travel than they do traveling. So you don't want to take the inspiration and the browsing and the library experience of dreaming away," he said ^[4].

Here is what Chesky says users do: they browse because they like browsing, and they plan with family and friends ^[4][7]. Here is what a chat window assumes they do, by his description: one person works through a few options per turn until a result arrives ^[2][7]. The interview does not include usage, booking or retention figures for the new search, so the only evidence so far for Airbnb's design is Chesky's account of his own customers ^[6].

Chesky calls the current search unfinished. "I think what you're seeing today is also not the endgame for e-commerce or for travel or shopping," he said ^[5]. He declined to call the destination a midpoint, then put it between a chatbot and "what you see in the first version we shipped" ^[5]. He also rejects the opposite outcome, where every app ends up as a data feed for one general assistant. "I don't believe in a world where the apps become data layers of a universal simple UI," he said ^[15]. "Our interfaces should be adaptive to the utility and the function that we're doing," he added ^[14].

Chesky wants Airbnb to be more agent-friendly. He said that will take software infrastructure on Airbnb's end ^[10]. He listed what a guest who arrives through an outside agent might still want from Airbnb itself: good browsing, messaging the host, comparing, verifying identity, a map, and adding other items ^[12]. "We want to make our app more agentic, and we want to build a lot more user interface controls," he said ^[11].

His two objections to chat make a grid that any team can place a task on ^[16]. One axis asks whether the shopper wants to look or wants it done. The other asks whether one person decides or several. The flight booker sits in the solo, get-it-done corner, where chat or an agent fits ^[3]. Chesky puts Airbnb in the solo, looking corner. AI can run the search there, but the screen has to stay a place to browse ^[3][4]. The group trip is the looking-together corner, and Airbnb has put it on its own list to explore ^[8]. A group that wants it done needs an approval step, so several people can sign off on one booking before anyone is charged.

The second test covers agents. Sort the steps on Chesky's list into the ones an outside agent can finish and the ones that have to happen on the company's own screen ^[12]. Each step in the second group needs one of the two routes he named: a handoff back to the app, or an SDK that carries the feature into the agent ^[12].