The Skift Takeaway

THE ARGUMENT

AI has gotten good at inspiration — destinations, itineraries, hotel suggestions. Khan said the harder problem starts the moment a traveler wants to book, and that the execution layer is where AI-driven travel wins or loses. He tied that to a shift in how people search: away from deterministic “A to B” queries toward attribute-based ones like “somewhere warm in July under 1,000 bucks,” which creates uncertainty suppliers must resolve. Travelport’s answer is TripServices, a rebuilt cloud-native API suite meant to push indicative pricing higher up the funnel and let AI agents transact safely. The pitch sits inside a build-versus-buy framing the moderator raised directly.

THE EVIDENCE

• The way people search has shifted over the last 18 months or so, with more travelers starting at an LLM rather than traditional search forms.
• “The higher up the funnel you give a price, the more chance there is of someone converting,” Khan said.
• With TripServices, a single hotel API call returns all properties, live availability, and images, allowing partners to surface corporate-negotiated or most profitable rates first.
• “The reason we’ve put so much orchestration within [our TripServices tool] is that when we interact with the AI agents or platforms, we want to make it as simple as possible. We don’t want the AI agents to have to go and do a lot of the legwork,” Khan said.
• A new TripServices user completed a build on Travelport’s flights platform in three weeks, which Khan said is “a lot shorter” than past GDS API work.
• Travelport partnered with Anthropic and Cognizant to improve code velocity through AI and to learn what travel problems to solve.