Artificial intelligence promises to reinvent travel. But behind the chatbot capable of making reservations, offering advice, and handling the unexpected, a long-standing question remains: Who is the user still willing to trust when everyone can recommend the same hotel?
There has never been a shortage of travel options
A delayed flight, a missed connection, a reservation that’s no longer responding: a trip rarely falls apart all at once. It unravels step by step. It is precisely within this fragile dynamic that Booking.com intends to build its next phase of growth.
Glenn Fogel does not view AI as merely an additional interface feature. He places it at the heart of a broader initiative: the “connected trip”—the promise of a journey in which search, booking, assistance, and incident resolution would finally form a single, seamless experience.
The idea isn’t new to him. It’s almost a revival of the old-fashioned travel agent: someone who knows your preferences, anticipates your constraints, and steps in when the original plan falls apart. The difference lies in the scale. Booking.com doesn’t want to recreate a travel counter; it wants to put this service right in the pockets of hundreds of millions of users.
An AI agent alone isn’t enough to make a platform
It would be tempting to reduce this shift to the emergence of a chatbot. But that would be missing the point. In the travel industry, making a recommendation is only the tip of the iceberg of a much more complex chain: actual availability, payment, cancellation policies, local regulations, customer service, disputes, and liability.
That’s where the competitive advantage lies. General-purpose models are already capable of recommending a neighborhood, suggesting a restaurant, and planning an itinerary. But they still can’t, on their own, handle the operational complexity of travel.
Booking.com has an asset that’s harder to replicate than an interface: habit. According to Glenn Fogel, 65% of Booking.com users go directly to the platform, without going through Google or any other intermediary. This figure speaks volumes. It doesn’t just measure marketing performance; it measures a kind of reflex.
In a world saturated with automated recommendations, acting on instinct has become a rarity.
Details
According to Glenn Fogel, 65% of Booking.com’s traffic comes directly from users. In the world of AI agents, this statistic takes on strategic importance: it means that most users aren’t looking for an answer first, but rather a trusted place to finalize a decision.
The New Rarity: Not Delegating Everything
The other important point concerns the concept of agency—that decision-making power that users are not always willing to relinquish. Fogel acknowledges that certain simple trips could be fully automated: a business trip from Paris to London, with a known schedule, a meeting location, and an organized transfer.
But leisure travel is proving to be more resilient. A family, multiple stops, conflicting preferences, trade-offs between budget, comfort, and personal preferences: the algorithm can reduce friction, but it cannot eliminate choice.
This is where AI applied to travel ties into a broader issue. Technical sophistication isn’t about making decisions for the user, but about knowing when to step back. The ideal agent isn’t the one who takes center stage. It’s the one that understands when the decision must return to human hands.
Regulate without stifling
The conversation then shifts to Europe, and the discussion takes on a more political tone. Booking.com was founded in the Netherlands in the late 1990s, in an environment that still allowed for rapid experimentation. Fogel believes that this leeway has been reduced due to a proliferation of regulations—at both the national and European levels—and constraints that are sometimes redundant.
His argument is not a call for the elimination of rules. Rather, he questions their actual effectiveness. In a company shaped by A/B testing, the idea seems almost logical: why not measure more systematically the effectiveness of a regulation, its cost, its side effects, and its impact on objectives?
This observation goes beyond the specific case of Booking.com. It touches on a deeper European concern: How can we build technology leaders if the most talented people, capital, and ambitious founders are leaving to seek—elsewhere—a pace of execution that has become rare?
What the models don’t have yet
The most straightforward question remains that of competition from large language models. What happens if Claude, ChatGPT, or Gemini become excellent travel advisors?
Fogel’s answer can be summed up in one line: Booking.com has to earn that reservation every day.
This isn’t a defensive stance. It’s a clear-eyed acknowledgment. Generative AI makes part of the available intelligence more commonplace. It makes recommendations more fluid, more conversational, and more accessible. But it doesn’t make responsibility any less important.
A model can make suggestions. A platform must respond.
It is this nuance that will shape part of the future of travel—not the beauty of an AI-generated response, but the ability to turn that response into a reliable experience: booking, paying, making changes, getting refunds, and resolving issues.
Augmented travel, then, will not be just a matter of prediction. It will be a matter of staying the course. Keeping a promise. Maintaining a relationship. Staying the course when the journey, as always, begins to go off the rails.
Glenn Fogel isn’t promising a world without frustration five years from now. He knows all too well that no one can control the weather. But he’s betting on a gradual reduction in friction, more refined assistance, and less intrusive personalization.
Perhaps the real question isn’t whether AI will become a good travel agent. It will. The question is which digital companies will still have enough credibility for us to trust them with more than just an idea: a departure, an arrival, and sometimes a moment in life.
Cette publication est également disponible en :
