Skift has turned its research desk onto the question every distributor and supplier is currently wrestling with: how is agentic AI changing travel booking?

The answer comes through Ask Skift, the publisher's AI-powered query service that draws on its editorial archive. A reader posed the question directly, and Skift's response pulls together the outlet's reporting on the shift from chatbots that recommend to agents that transact.

The distinction matters for revenue. Generative AI tools deployed over the past two years — Expedia's in-app assistant, Booking.com's chatbot, Kayak's Ask AI — largely sit inside existing funnels. They answer questions, surface options, and hand users back to a checkout page the operator controls. Agentic AI is different. It is built to complete the booking itself: interpret a request, compare supply, make the purchase, and manage the trip afterward, with minimal human intervention.

That difference reshapes distribution economics. If a software agent selects the flight, the hotel, and the payment method, then the interface where travel is sold stops being a branded app or website and becomes whatever surface the agent runs on. Suppliers lose visual merchandising. Intermediaries lose the cross-sell screen. Loyalty programs lose the login moment that anchors repeat booking.

The question of who owns the customer relationship is therefore the commercial core of the agentic-AI debate. Skift's coverage has tracked how airlines, hotel groups, and online travel agencies are each positioning for a booking environment in which the buyer may never see their brand during the transaction.

Ask Skift is itself part of the story. The service applies large-language-model technology to Skift's own reporting, letting readers interrogate a decade of trade journalism in natural language. Media companies, like travel sellers, are testing whether AI becomes a threat to their distribution or a new interface for it.

For sellers of travel, the practical questions Skift's answer points toward are concrete. Who pays the commission when an autonomous agent books a room — the hotel, the agent's developer, or the underlying distribution platform? Can suppliers make inventory machine-readable and direct-bookable fast enough to avoid being disintermediated a second time, as they were by metasearch and online travel agencies? And will travelers actually delegate spending authority to software, or keep agents in an advisory role?

The technology is arriving faster than the commercial rules. Payment infrastructure, liability for errors, refund handling, and loyalty accrual all remain unsettled for transactions completed by non-human buyers. Skift's reporting on the topic suggests the industry is in an experimentation phase rather than a measured-results phase — pilots and partnerships outnumber disclosed booking volumes attributable to autonomous agents.

That gap between capability and verified impact is the honest read for trade audiences. Agentic AI has moved from conference keynotes into product roadmaps, but no major operator has yet published bookings or revenue figures attributable to fully autonomous transactions at scale.

Skift's Ask Skift response signals that the question has matured from speculation into a researchable trade topic — one the publisher expects to keep answering as pilot programs convert, or fail to convert, into distribution that moves share.