Everyone is being sold AI that "sells for you." Travel's own consumer data draws a much narrower line than that — and it is worth knowing exactly where it falls.
The pitch going around right now is an AI that sells trips for you. The consumer research does not back that, and it fails in an interesting way: not a flat "travelers hate AI", but adoption that concentrates in planning and thins out at the purchase.
Where travelers actually use AI
PwC's Holiday Outlook 2026 measures where in the journey it actually shows up: 60% of consumers who plan to travel say they'll use AI somewhere in the booking journey, most often to research or compare hotel and flight prices (41%), get recommendations (32%), or build an itinerary (22%). Read that list again — every named use is information gathering. Last year's edition asked the purchase question directly, and overall, 68% said they were likely to use AI for comparing flights, and a lower 57% for booking travel.
Deloitte has the planning end growing fast — nearly a quarter of travelers reported using gen AI tools for trip planning in late 2025, three times as many as in 2022. At the purchase end it stays thin. Skift, reporting on an Expedia Group survey, found that only 8% of travelers are comfortable letting AI handle bookings. A separate trust report this July found only 24% of travelers trust AI-generated recommendations, even as about half are comfortable using AI to compare travel options.
Machines inform, humans commit
So the useful split is not "AI for cheap trips, humans for luxury." My read of those numbers is simpler and more operational: the machine handles information transfer; the human handles commitment. The networks building this at scale already draw the line in the same place — Selectour, France's largest network of independent travel agencies, launched AI agents designed to help consumers discover and plan trips, then hand off to human advisors who finalize arrangements.
Which means the automation worth buying this year is front-door automation: catching the inquiry, collecting the facts, getting it to a person fast. A $9,000 anniversary trip is not draft work, and "probably right" is not a standard you can put your name on.
My read of those numbers is simpler and more operational: the machine handles information transfer; the human handles commitment.
Collect the facts before the call, not during it
List the plain facts you need on every single inquiry — destination or shortlist, dates or window, who is travelling and any needs, budget band. Those are information transfer, so put them in your form, auto-reply or chat and let them arrive already answered. Then spend the live conversation on the part a form cannot do: why this trip, why this year, and what would make it a disappointment.
This is roughly the shape of the system we built for a growing travel agency: 5× monthly sales — 8 to 40+ in 90 days. The case is here if you want to see how it was put together: https://brandrstudio.com/our-work/travel-agency-ai-sales See the case →
Get this every week. One issue, one idea, sources you can check. No pitch.