On 11 November 2026, at the Skift Commerce Summit in New York, MCR Hotels chairman and CEO Tyler Morse said he wants to stop personal AI agents from booking rooms at his hotels. "Everyone is going to ban the agents because it's in their interest to do so," he said, according to Skift. His reason was commercial, not legal: travel merchandising is built to trigger impulse purchases, and software cannot be talked into a splurge.

The same day, United Airlines said it is restricting unauthorised third party AI agents, on its own platforms and through travel agencies and aggregators, citing regulatory and consumer safety concerns (Skift). And Citi's Abhinav Anand told the same summit that liability for an agent's mistake should fall "to the agent themselves, the company that is actually coming with the agent" (Skift).

Three different industries, one question: who decides which machines get to book, and on what terms? Here is my view on what this means for hotel tech vendors.

My position

Hotels will not ban AI agents. They will want to control them. The hotel tech vendors that sell control, not access and not refusal, will win the next round of booking and guest platform deals.

A ban is a statement of frustration. A policy is something a hotel can buy. Vendors should be selling the second.

Why a hotel ban does not work

Morse himself admitted the practical problem. As Terence Ronson reports in Hotel News Resource, Morse followed his prediction with: "I can't figure out how to do it." Ronson's argument is that no hotel can ban its way out of distribution: if the front door is shut, the agent comes round the side through an OTA. He notes that in Singapore and Japan, Agoda is already bookable through Meta's Muse.

United can attempt a ban because it controls its own contract of carriage. According to Deep Arrival, citing View from the Wing, United revised that contract to prohibit AI agents from making, changing or servicing tickets and can cancel reservations made that way. United also said agent bookings are still rare and that it can identify where bookings come from. Most independent and mid sized hotels have no equivalent lever, and a large share of their rooms are sold by intermediaries whose own AI deals they do not control.

What hotels are actually worried about

Strip the headlines down and the concerns fall into five groups. Each is a product and positioning opportunity.

1. Upsell: make offers machine readable

Morse's objection is that agents are rational buyers. Ronson's response is to rebuild upsell for a rational buyer: structured, rule based offers an agent can accept on the guest's behalf. That is a specific brief for upsell, booking engine and guest messaging vendors. An offer that exists only as a pop up banner will not survive an agent. An offer with a clear price, condition and inclusion can be read and accepted. Our post on AI answer visibility covers the content side of the same problem.

2. Source visibility: help hotels see agent traffic

Skift reviewed robots.txt files for 50 major travel sites and found only 10 explicitly addressed AI crawlers; the rest lumped them in with every other bot or left them to defaults (Skift). The same piece cites Adobe data showing AI visits to travel sites surged 119% year over year in July. Hotels cannot set a policy for traffic they cannot see. Vendors that report agent driven visits and bookings separately give the hotel the first thing it needs before deciding anything.

3. Liability: put it in the contract and the API

Citi's principle that the agent's company should carry the liability is a starting point, not a settled rule. For a vendor, the practical step is to make cancellation, refund and modification rules explicit in what you expose to third parties, so an agent cannot claim it was not told. Payment and distribution models matter too: in the Duffel and Katanox model reported by Skift, the hotel remains the merchant of record, which changes who handles disputes. We covered the security side buyers will raise in the hotel security questionnaire.

4. Double commission: show the real channel cost

Ronson tells hotels to model double commission in their 2027 budgets, because an agent booking through an OTA could carry both the OTA's cut and the platform's. Revenue management, channel management and business intelligence vendors can show net revenue by source, including agent paths, so hotels can see whether this is happening rather than guessing.

5. Visibility: sell readiness, not fear

Being invisible to AI is a real risk, and vendors selling websites, content and data should address it. Sell it as a checklist, not a threat. The Google AI Mode booking rules give a concrete framework to anchor the conversation.

The data says agents are coming slowly

Vendors should not oversell urgency. Agoda's 2027 Asia Travel Pulse, based on more than 3,300 travellers in 11 markets, found 53% now use AI to plan trips, up from 33% the year before (Agoda). But only 7% would use AI to make a booking, 6% would fully trust an agent to book for them, and 89% want to keep some control over the booking. Of the 38% who would let an agent book in some form, most want conditions, such as reviewing before confirmation.

That points to a near term product: agent assisted booking where the guest confirms, rather than fully autonomous booking. Vendors who build for that middle state will be ready earlier than those waiting for full autonomy.

How to use this in sales conversations now

• Ask prospects whether they have an AI agent policy; most will not, which opens a useful discussion

• Show how your product tags and reports agent driven traffic or bookings, or say honestly that it is on the roadmap

• Demonstrate one structured offer an agent could read and accept

• Explain who is merchant of record on each path your product supports

• Avoid promising volumes; cite the Agoda numbers so the hotel trusts your judgement

AI agent policy is now a GTM topic, not just a product one. It belongs in your messaging, your sales enablement and your roadmap story, as part of the system set out in our hotel tech GTM strategy guide. For the OTA side of the same shift, see what Booking Holdings said about AI.

If your sales team does not yet have a clear answer when a hotelier asks about AI agents, a free GTM leak audit will show where that gap is costing you deals.

Frequently Asked Questions