Brian Chesky interview: AI agents need their own operating system

Airbnb has introduced an AI-powered search in its fall product update, but CEO Brian Chesky said the company remains cautious about treating chatbots as the default travel interface. Chesky reiterated that conversational bots tend to present a few options at a time and require multiple turns to reach results, making them ill-suited for browsing and discovery — activities many users enjoy when planning travel. Chesky said Airbnb’s search UI will continue to evolve and that future interfaces may sit between today’s conventional app screens and pure chatbots. He also flagged a need for collaborative, or "multiplayer," AI experiences that let multiple people plan a trip together, since many travel decisions are made in groups and current chatbots are primarily single-user interfaces. Preparing for consumer AI agents is a strategic priority for Airbnb, Chesky said. He expects agents — autonomous tools that perform tasks on a user’s behalf — to act as lead generators, but to work well they will need richer software hooks on Airbnb’s side. That could include an expanded SDK and more granular UI controls so agents can hand off to the app for browsing, messaging hosts, comparing options, verifying identity, or using maps without breaking the user experience. Looking beyond Airbnb’s own product roadmap, Chesky argued the industry lacks an operating system built specifically for AI. He described current efforts as apps built on legacy OSes rather than AI-native platforms, and said a full developer toolkit and kernel-level integration are needed to enable agent-to-agent interoperability. Chesky suggested that agents may one day operate like "apps with faces," interacting across platforms when standards or agreements allow, and that Airbnb intends to develop multiple agents (explore, customer service, etc.) that could eventually interoperate with external agents via protocols such as MCP.