AI Agents Threaten Booking, Expedia, And Airbnb’s Business Models

Travel planning giants are embracing AI tools as AI agents are entering this field and threatening to bypass online platforms.

The three major online travel platforms — Booking.com, Expedia, and Airbnb — are working to adapt to AI agents, which could bypass their role in the travel booking process, as reported in an analysis by the Financial Times (FT).

The $1.6 trillion global travel market faces transformation as OpenAI, Google, and Anthropic develop bots that operate as AI agents, designed to create personalized travel arrangements for users.

FT notes that Booking and Expedia have partnered with OpenAI to launch tools such as AI trip planners, while Airbnb has already introduced an AI-enabled customer service agent. Next year, Airbnb plans to add more “agentic” functions.

Glenn Fogel, CEO of Booking Holdings, said to FT: “We don’t have to do what OpenAI, Google, Grok or Meta are doing . . .[all of whom] are having to invest incredible amounts of money to build these models. Our belief is that as long as we . . . work closely with them that we will be able to participate in a way that provides a great return for our customers and our partners.”

Hotels see AI agents as beneficial because they reduce the 15–20% commission fees typically paid to online travel agencies. HOTREC, a European group, believes AI agents show strong potential to decrease OTA dependence but may create another dependency cycle.

Investors remain cautious. “There was a natural inclination and still is among investors that travel loses in an AI first world,” said Eric Sheridan, analyst at Goldman Sachs, as reported by the FT.

Airbnb chief Brian Chesky said its platform will become “more personalised and more agentic” in 2025: “It will not only tell you how to cancel your reservation, it will know which reservation you want to cancel,” as reported by the FT.

According to WIRED, new AI agents from OpenAI and Anthropic, including Operator and Computer Use, now generate customized itineraries based on user preferences

WIRED put the technology to the test by letting AI plan an entire vacation, from transport and accommodation to meals and activities. The experiment showed that while the tools are not flawless, the overall outcome was summed up with: “It wasn’t terrible.”

Still, AI travel agents have limitations. Researchers found OpenAI’s GPT-4 successfully handled complex travel planning just 0.6% of the time. “I am not foolish enough to say that I’m not worried about it,” admitted Fogel, as reported by the FT.

Indeed, AI systems fail to duplicate human emotional intelligence, and nuanced travel understanding which results in incorrect and absurd recommendations. In line with this Euro News gives the bizarre example of Microsoft AI guide’s suggestion of the Ottawa Food Bank as a top attraction.

Euro News points out that AI agents fail to handle complex itineraries, special dietary or mobility needs, and cannot negotiate or provide reassurance during travel disruptions.

Experts stress that AI should augment, not replace, human agents, as travellers still value personal connection, local knowledge, and tailored guidance over automated planning.