The Limitations of Chatbots in Travel and Commerce

Airbnb co-founder and chief executive Brian Chesky recently discussed the state of consumer technology, reiterating his view that traditional chatbot interfaces remain unsuited for e-commerce and travel discovery. According to Chesky, chatbots fall short because they typically present only a few options at a time, forcing users to navigate multiple turns before reaching a satisfactory result.

While Airbnb recently introduced a new search function as part of its fall update following earlier tests designed to ship features faster, Chesky emphasizes that this iteration is not the endgame. Travel research indicates that consumers derive significant pleasure from planning, browsing, and dreaming about upcoming trips, meaning that platforms should preserve the library and inspiration experience rather than condensing everything into a single rapid command.

Exploring Multiplayer AI and Generative Interfaces

Another major drawback of standard chatbots is their single-user design, which fails to support collaborative planning among family and friends. Over the next three to six months, Airbnb plans to explore multiplayer AI interfaces that allow multiple people to use the technology simultaneously while planning group trips.

At the same time, startups across the technology landscape are experimenting with generative interfaces—screens created dynamically by artificial intelligence on the fly rather than predefined by a designer. Chesky anticipates that future applications will feature a balanced combination of deterministic and generative user interfaces.

Preparing Airbnb for Consumer AI Agents

As automated tools designed to execute tasks on behalf of users—such as Instinct and Muse—continue to evolve, companies are working to make their platforms more agent-friendly. Chesky notes that consumer agents will require robust software infrastructure on Airbnb's end to serve effectively as lead generators.

Rather than eliminating traditional apps entirely, Chesky believes that application interfaces should remain adaptive to specific utility and functions. He envisions the Airbnb application eventually evolving to incorporate specialized agents across exploration and customer service, while utilizing standard protocols like MCP to achieve broader interoperability with other agents.

The Case for an AI-Native Operating System

Despite ongoing industry enthusiasm, Chesky argues that the broader technology sector faces a more fundamental obstacle: the lack of an operating system built natively for artificial intelligence. Current systems run AI applications on top of existing platforms like iOS, macOS, and Windows without operating at the kernel level where core components can natively interact.

He points out that attempts to build app stores without comprehensive software developer kits and foundational operating system support have struggled to achieve broad functionality. According to Chesky, current consumer AI tools have not yet fully succeeded, and achieving a breakthrough will require deeper architectural changes.

The Future Role of Platform Dominance

Addressing common misconceptions in Silicon Valley, Chesky disputes the notions that user interfaces will completely disappear in favor of pure speech or that software will become entirely generative. He maintains that human designers can prompt and structure software more effectively than average consumers.

Ultimately, as long as Apple and Google maintain dominance over mobile platforms, traditional applications will likely remain central fixtures. Chesky concludes that platform holders must build new foundational environments to truly facilitate the transition from standard apps to fully interoperable AI agents.

Sources

• TechCrunchBrian Chesky interview: AI agents need their own operating system