Airbnb CEO says AI needs more than chatbots
Brian Chesky says travel planning needs richer, shared interfaces as Airbnb adds AI search, voice tools, and support for future agents.
Airbnb has rolled out a new AI-powered search feature as part of its fall update, but CEO Brian Chesky says the company is still far from done with AI. In a wide-ranging interview, he argued that chatbot-style interfaces are a poor fit for travel discovery, shopping, and collaborative planning. His comments also point to a broader debate in the AI race: whether consumer apps will shrink into chat windows or evolve into something much more layered. For Airbnb, the answer matters now, not later.
The discussion reaches beyond one company. If AI becomes a main way people search, compare, and book, the design of apps, the role of voice, and the relationship between platforms and users could all shift. Chesky’s core point was straightforward: the current wave of consumer AI still feels unfinished.
Airbnb’s new AI search is only an early step
Airbnb’s fall update includes a new AI-powered search, but Chesky stressed that the version available now is not the final form of the product. He said the company has been deliberate about AI and expects the interface to keep changing.
He pushed back on the idea that a chatbot alone is the right tool for travel or shopping. In his view, chatbots show only a few options at a time and force users through multiple turns before they reach a result. That may work for some tasks. It works less well when people want to browse places to stay or compare options in a meaningful way.
Chesky also said Airbnb does not want to remove the parts of travel planning that people actually enjoy. Many users like browsing, dreaming, and comparing before they book, he said, and the planning itself can be part of the appeal. According to IT-PUB News, that is why the company does not want to replace discovery with a purely transactional interface.
Chesky wants AI that works for groups, not just individuals
One of Chesky’s main criticisms of today’s chatbot model is that it is built around a single user. He said that makes it a weak fit for Airbnb, where trips are often planned with family or friends.
Over the next three to six months, the company plans to explore what he called “multiplayer” AI — interfaces several people can use at once while planning together. He did not present that as a finished feature, but as a direction Airbnb wants to test.
Chesky also said future interfaces will likely mix two kinds of design: deterministic, meaning pre-designed and predictable, and generative, meaning screens that AI creates on the fly. He pointed to startups such as Monogram and Wabi as examples of companies experimenting with that model.
The larger idea is that AI should do more than answer questions. It should support how people actually make decisions together. For Airbnb, that matters because booking a trip is often a group process, not a solo one.
Airbnb is preparing for AI agents in and around the app
Chesky also discussed AI agents — tools that can carry out tasks on a person’s behalf. He said they will require strong software infrastructure from Airbnb, and that the company wants to become more agent-friendly.
He said agents could even act as lead generators. At the same time, he made clear that Airbnb does not want its app reduced to a thin data layer behind a universal interface. Different services need different designs, he argued, because people do different things in each one.
That is why Airbnb wants to build more user interface controls rather than disappear into a generic AI layer. Chesky listed examples of functions people may still want inside the app: browsing, messaging a host, comparing listings, verifying identity, using a map, or adding other items. If an external agent is involved, he said, it may need to hand off to the app. The other option is a richer software development kit that would bring more of the app experience to the user through the agent.
Chesky also expects agent-to-agent interactions to become more common. Some of those exchanges will happen inside Airbnb, he said, while others may involve agents from other apps. In his view, agents are like “apps with faces,” and if they can communicate with one another, apps could become more interoperable than they are today.
Voice is part of Airbnb’s next AI push
Voice is another part of Airbnb’s AI plan. Chesky said he believes people will eventually speak to computers more than they type, and that voice will play an important role in Airbnb search and customer service.
The company plans to roll out voice agents this fall to help users. That does not mean text is going away, but it does show Airbnb is betting that some tasks will feel easier through speech than through typing or tapping through menus.
Chesky also said AI is already helping Airbnb ship features faster. On a personal level, he said it has cut the number of meetings he needs, because he can ask AI for information directly instead of relying on conversations to gather it.
Chesky says consumer AI still has not been solved
Despite the momentum around chatbots and agents, Chesky said he does not think consumer AI has been cracked yet. He said his own experience using Airbnb through AI tools such as Muse and Instinct has not worked well, adding that hotels do not work especially well either, while Airbnbs work even worse.
His broader argument is that consumer AI needs a lower-level foundation before it can really work. In his view, apps need to become agents, agents need to be interoperable, and user interfaces need to be far richer than a simple text box.
He also challenged two ideas he sees as common in Silicon Valley: first, that the interface will disappear and people will only talk or type; second, that software will become fully generative. Chesky said designers can prompt better software than ordinary consumers can, which in his view means design will still matter even in an AI-heavy future.
He ended by saying that as long as the iPhone remains dominant and Apple and Google control the main platforms, apps will probably stay central. A real shift from apps to agents, he said, would require a new platform from Apple, Google, or another company. Until then, he expects most services to keep their apps and build AI features into them rather than replace them entirely.
