Airbnb recently introduced an AI-powered search feature in its fall update, marking a cautious but notable embrace of artificial intelligence by the company. CEO Brian Chesky explained that while chatbots remain inadequate for travel browsing due to their linear and limited interaction style, Airbnb’s current AI approach is only a step toward more sophisticated interfaces. Chesky envisions a hybrid model that balances quick, agent-driven actions with the enjoyable, exploratory browsing users expect from planning travel, emphasizing the importance of preserving the “dreaming” and collaborative aspects of the experience.

Chesky also highlighted the challenge of building AI interfaces that support multiple people simultaneously, which is crucial for travel planning shared among friends or families. Unlike single-user chatbots, Airbnb is exploring “multiplayer” AI agents that several users can interact with at once. The company aims to develop a variety of AI tools within its app, including agents capable of handling search, customer service, and other functionalities while maintaining robust user controls for browsing, messaging hosts, and comparing options.

Looking ahead, Chesky stressed the need for a new kind of AI operating system, one that operates at a fundamental, kernel-level scale rather than relying on current general-purpose operating systems like iOS or Windows. He argued that real progress in consumer AI requires making all apps into interoperable agents capable of working with each other, supported by a comprehensive software developer kit. According to Chesky, current AI efforts, such as ChatGPT’s early attempts at an app store, lack this depth and thus have not fully delivered on the promise of seamless consumer AI.

Finally, Chesky sees voice technology as an increasingly important interface method and plans to introduce voice agents within Airbnb’s app to enhance search and customer support. He also reflected on how AI is already transforming internal operations by reducing the need for meetings, allowing quicker access to information. Despite the excitement around AI, he remains pragmatic about its limitations and believes that fully AI-native platforms will likely depend on major players like Apple or Google creating new foundational systems for agents rather than apps dominating the landscape going forward.