Booking.com AI answers hotel questions before guests book

By Marcus Bennett

Booking.com put an AI property-question tool into its app in October 2024, initially serving travelers in the U.S., U.K., Australia, New Zealand and Singapore. It generates immediate pre-booking answers from a property listing, traveler reviews and photos.

The tool handles questions that once reached a hotel through email, phone or the booking site’s message button. A useful answer often won the reservation, and no hotel employee now writes the pre-booking response.

Neither Booking.com nor Expedia says whether a hotel can inspect guest questions or correct an AI answer. Both companies also leave key commercial metrics unpublished. Neither has disclosed how many travelers use the tools or how often an AI answer precedes a booking.

What does the AI answer before a reservation?

Guests still ask whether parking is onsite, whether a street-facing room stays quiet, whether a crib fits and whether breakfast is included at the quoted rate. Reception once supplied those answers, and a good answer could win the business.

• The listing and photos: The hotel controls facilities, room types, policies, descriptions and images uploaded through the Extranet.
• The reviews: Guests supply the text, including observations from last month and two years ago.
• The synthesis: Booking.com combines those inputs into a short response without asking hotel staff to write it.

Booking.com described the mechanism: “The AI pulls the relevant information from the property listing, traveler reviews and photos, and presents a short answer.”

Missing fields become hard limits. If the listing does not say the garage sits under the building, the AI cannot supply that detail unless a review does. The same constraint applies to cribs, connecting rooms and check-in after midnight.

Three guests who complained about a nightclub last year can still shape an answer after the venue closes this spring. Booking.com does not describe a freshness check for review content.

Booking.com also said in October 2024 that it was testing review summaries. Those summaries compress all reviews into a few lines and could later focus on parking or wheelchair access, but the company said deployment required more testing.

At launch, Booking.com said property pages would follow in the near term. It also promised local-language versions in Spain, Italy, Germany, France, Poland and the Netherlands. Two years later, the company had not published current feature coverage or confirmed those extensions.

How does Expedia’s version compare?

At its May partner conference, Expedia presented AI Property Expert and AI Property Compare without launching either product. The company gave no release date, no explanation of source data and no view of what hotels can monitor.

Expedia said the tools are “meant to help people get answers faster and evaluate hotel options with more confidence.” Its advice to hotels was concrete: stronger connectivity, richer content and consistent rates help properties earn attention earlier.

If the comparison tool reaches production, a traveler could ask which of two hotels is quieter or nearer the beach and receive one answer covering both. That places property content and review history directly in the pre-booking decision.

Where does Google fit?

Google offers a starker control problem. Its developer documentation describes AI place summaries written from user reviews alone and labels them “Summarized with Gemini” in the apps that show them.

The documentation gives users one remedy: a button to flag inappropriate content for removal. It provides no field for a business to correct the summary. A hotel can publish better material, but it cannot directly amend Google’s synthesis.

Does the system replace hotel staff?

Booking.com introduced Auto-Reply for hotels in October 2025. The hotel defines its reply topics, and the system answers common and unique guest questions instantly. Booking.com says the tool is available worldwide.

Auto-Reply operates on Booking.com’s guest messaging, which opens with the reservation. It therefore reaches a mostly post-booking guest, unlike the AI that answers listing questions before a purchase.

At the operating level, the AI replaces some repetitive pre-booking response work that reception previously handled. Neither platform claims headcount reductions or labor savings, however, and the pre-booking announcements disclose no hotel fees.

Without published usage data, operators cannot tell whether the channel deflects calls, lifts conversion or simply repeats content already on the page.

What should hotels check this week?

Operators should run a listing audit now. Open the Booking.com app on a phone, find the property and ask the five questions reception hears most. Record the answers, then compare each with the listing.

Where an answer is wrong or missing, the listing usually caused the failure, and the listing remains the hotel’s most direct control. Accurate room fields, policies and photographs now carry more weight because the platform can quote them before staff speaks.

Until Booking.com and Expedia disclose source rules, hotel visibility and usage, operators should treat AI answers as an extension of listing accuracy. Expect the pre-booking answer to become another comparison gate, with less room for undocumented details.