Airbnb CEO Brian Chesky argues that the current landscape of AI applications is insufficient for true agent interoperability, calling for a dedicated operating system architecture built specifically for AI agents. Speaking during the rollout of Airbnb’s new AI-powered search, Chesky outlined his vision for how agents should function within existing digital ecosystems and why he believes consumer AI remains unsolved.
What Happened
Airbnb recently launched its new AI-powered search as part of its fall update, marking a significant shift for a company that has historically been deliberate about AI adoption. In an interview, Chesky clarified that while the company is integrating more AI, he remains skeptical of chatbots as the primary interface for travel discovery. He stated that chatbots are poorly suited for browsing or shopping because they present limited options and require multiple turns to reach a result. Instead, he envisions an interface that sits between a traditional chatbot and a standard browsing experience, preserving the 'inspiration and browsing' aspects of travel planning.
Chesky identified a structural flaw in current AI agents: the lack of a robust software infrastructure. He explained that for agents to be effective, they need to interact with apps at a deeper level than current integrations allow. He cited his own experience using Airbnb through agents like Instinct and Muse, noting that the experience 'doesn’t work very well.' According to Chesky, this failure stems from the absence of a true 'operating system' for AI. He argued that current AI apps are built on top of iOS, macOS, or Windows, rather than having an AI-native kernel where components and agents can interact seamlessly.
Why It Matters
Chesky’s comments highlight a growing tension in the tech industry regarding the architecture of AI agents. He posits that the current race to create a 'quarterback' AI agent is flawed because no major player has developed a full software developer kit (SDK) or operating system for AI. He referenced Sam Altman’s early attempt to create an app store within ChatGPT, noting that Chesky had previously advised Altman that without a proper SDK and OS structure, functionality would be limited. This perspective suggests that the current wave of agent-based applications may be hitting a ceiling due to infrastructural gaps rather than model capability.
For developers and the industry, Chesky’s thesis implies that interoperability is the next frontier. He described agents as 'apps with faces' and suggested that if agents can talk to each other, apps can become interoperable. However, he emphasized that this requires a shift from deterministic interfaces to adaptive ones. He noted that while startups like Monogram and Wabi are experimenting with generative interfaces, a mix of deterministic and generative elements will likely persist. The burden of creating this new platform, he argued, falls on major ecosystem players like Apple or Google. Without their involvement, apps will remain containers with agents inside them, rather than becoming robust, interoperable agents themselves.
Chesky also addressed the role of voice and multiplayer AI. Airbnb plans to explore 'multiplayer' interfaces over the next three to six months, allowing groups to plan trips collaboratively using AI. Additionally, the company intends to roll out voice agents this fall for search and customer service. Chesky believes that while people will eventually speak to computers more than they type, the current misconception in Silicon Valley that interfaces will disappear entirely is incorrect. He argued that designers can prompt better software than consumers, suggesting that human-designed, adaptive interfaces will remain crucial.
The Bottom Line
Airbnb’s CEO views the current state of consumer AI as unresolved, primarily due to a lack of foundational operating system support for agents. By advocating for kernel-level AI integration and robust SDKs, Chesky suggests that the industry must move beyond simple chatbot wrappers to build true interoperable agent ecosystems. For now, Airbnb is focusing on enhancing its own app’s agentic capabilities and preparing for voice interactions, while waiting for platform giants to provide the necessary infrastructure for a full shift from apps to agents.