Booking Holdings CEO Glenn Fogel says building an AI travel agent is not the hard part of the industry's next technology shift, according to remarks reported by Skift. For the operator of Booking.com, Priceline, Agoda and Kayak, the harder question is everything that surrounds the build: adoption, trust, execution, and protecting the booking funnel that generates the group's revenue.

The framing matters for anyone who sells travel. If the CEO of the largest online travel company by gross bookings believes the engineering challenge is solvable, then competitive advantage will not come from owning a model. It will come from how quickly a platform can move AI from demo to transactions — and how well it defends commission economics when the interface between supplier and traveler changes.

What does the claim actually signal?

Fogel's position separates two problems that the trade often conflates. The first is constructing an agentic system that can search, compare and complete a booking with minimal human input. The second is making travelers use it at scale, in categories where they have spent two decades learning to book through screens and filters.

The first problem, in his view, is tractable. The second is where incumbents, suppliers and intermediaries will win or lose share. That distinction has direct distribution consequences: if agent-led booking becomes the default, the surface where marketing spend, metasearch placement and ranking currently compete gets compressed into a conversation.

Why incumbents see execution as the moat

For Booking Holdings, the statement reads as a confidence signal rather than a concession. The group has spent years investing in its in-house tech stack, consolidating brands onto shared infrastructure, and layering AI into customer service, fraud detection and merchandising. A CEO who says the build is not the hard part is arguing that scale, supply inventory and transaction data — assets startups lack — decide the outcome.

That logic cuts against the wave of well-funded AI travel startups pitching that a clever agent can displace the intermediaries. Fogel's implicit counterargument: plenty of players will be able to build the agent; very few will have the volume, supplier relationships and service infrastructure to operate it profitably at global scale.

What is at stake for sellers of travel?

The shift from search-box to agent changes how travel is sold in several concrete ways:

• Distribution leverage. If AI agents intermediate the booking, platforms that control the agent control the shelf. Suppliers dependent on OTA channels could face steeper acquisition costs.
• Marketing economics. Paid search, metasearch bidding and rate parity displays were built for a browse-and-compare traveler. An agent-led traveler never sees that inventory of choices the same way.
• Commission structures. Whoever completes the transaction — the brand's agent, a device-level assistant, or a third-party agent — determines which entity captures the booking fee.
• Loyalty and retention. Agents that remember preferences across trips could lock travelers into one platform's ecosystem more effectively than loyalty programs ever did.

None of these outcomes is measured yet. Fogel's comments are a strategic position, not a results disclosure, and investors and trading partners should treat them as such — a statement of where the company believes difficulty lies, to be tested against future booking-channel data.

Does this weaken or strengthen the OTAs?

Fogel's message contains both a threat and a reassurance. The threat: the interface that made OTAs dominant — search, sort, book — is exactly the layer AI agents promise to bypass. The reassurance: the assets underneath that interface, from customer service operations to payments handling to supplier inventory, remain hard to replicate.

The CEO is effectively telling the market that Booking intends to own the agent rather than be disintermediated by it. Rivals including Expedia, Google and a long tail of AI-native entrants are running the same race with different starting positions.

What comes next

Watch for the proof points over coming quarters: share of bookings completed through AI-assisted flows, changes in customer service cost per booking, and any movement in marketing spend as a percentage of gross bookings. Fogel has set the terms of the debate — the build is easy, the adoption and economics are not — and the industry will now judge his thesis against Booking's transaction data as AI-driven booking flows roll out across its brands.