AI-native distribution isn't a project you run

Becoming AI-native inside the hotel is a choice you make on your own clock. Becoming bookable by a machine is a deadline someone else sets.

The line you hear now from investors and consultants goes like this. Hospitality has spent two years putting AI on top of how it already works. The next step is to stop doing that and build processes that assume AI from the start. The companies that pull it off pull ahead. Most can't, the argument goes, because they don't have the engineers.

That last part is built for a software company, not a hotel. Hotels don't hire engineers and build their own systems. They buy them — from a PMS vendor, an RMS vendor, a channel manager — and they've done it that way for twenty years. So the framing misses how a hotel actually adopts anything. The real question is simpler: will a vendor sell me this, and when. And that question splits in a way the slogan hides.

There isn't one job called "go AI-native." There are two, and they don't run on the same clock.

Two clocks

Inside the hotel, going AI-native is something you choose, and you choose it the way you've chosen every tool before it: you wait for your PMS or RMS vendor to ship an AI-native version, you buy it, you switch it on. You can rebuild forecasting around an agent or leave it in a spreadsheet. You can let a system set prices on its own or keep a person in the seat. If the vendor is late, you wait. The pace is the vendor's, and nothing breaks tomorrow if it slips.

Distribution is the other job, and you don't hold the clock.

The guest's AI is already shopping. Someone opens an assistant, describes a trip, and the assistant goes looking for a place to put them. It doesn't read your website the way a person does. It asks a question and wants a clean answer back. If your hotel can't answer in the form the machine expects, the machine moves to one that can. You didn't agree to this and you can't postpone it. The demand side switched first, and it set the date.

So "become AI-native" means two things in the same building. On the inside it's an option. On distribution it's a deadline.

What AI-native distribution actually looks like

It helps to see it concretely, because the phrase is easy to say and hard to picture.

Start with today. You write a description of the hotel for a person to read. You push that description and your rates out to the channels. Then you wait — to be ranked, listed, or scraped. The whole posture is push it out and hope it lands well.

Now the machine version.

The description stops being a brochure. It becomes a set of facts a machine can ask about: does it have a pool, is parking free, will two adults and a child fit in this room on these dates. Each fact sits where an assistant can find it and check it in a second.

You stop pushing and waiting. You open a live connection the assistant pulls from — price, availability, and terms, returned the moment it asks. It's the move behind HomeToGo wiring a direct line into ChatGPT, a pipe an assistant can query for real-time availability rather than a listing it has to go find. Pull, not push.

Your rate is no longer a single number you set in the morning. Now it's an answer your own system gives in the moment — for this guest, these dates, this much notice, here is the price and here are the conditions. Close to how you'd answer a company asking for a corporate rate, except it happens instantly and nobody types it.

And the booking finishes against you. The assistant doesn't just point the guest your way. It places the booking, and your system takes it, confirms it, and fulfills it.

A hotel still running the old posture — write the brochure, push it, wait to be picked up — is bringing a brochure to a database query. The two don't meet.

Merchant of record stops being a footnote

When an assistant books your hotel and your system takes it, who is the merchant of record? Who sold the room, charged the card, and owns it when the trip goes wrong?

Google has already answered. In its lodging documentation, the hotel remains the merchant of record, and Google frames that as a benefit: you keep the guest, the data, and the relationship, the same as a booking made on your own site. The docs go further — your own terms and conditions cover the booking, just as if the guest had reserved with you directly.

That control has a second face. If you sold the room, you also own it when something breaks: the chargeback, the dispute, the refund. Google's docs don't spell that out — it follows from being the seller.

For most hotels that's a new daily task. Today an OTA or a card processor often sits between you and the mess. Under the merchant model many hotels run through Booking.com, Booking.com is the merchant of record, not the hotel. An agent-placed booking on your own connection moves that back onto you. The person who used to watch rates and channels now writes the rules the system follows when it answers, and works the exceptions when a machine-placed booking breaks. A different job than that role used to do.

Who set the date

Here's the uncomfortable part for the inside-the-hotel framing. You can take your time rebuilding the back office. You wait for the vendor, you buy when it ships, and a slow rollout costs you without sinking you.

The distribution side gives you no such room. The guest's assistant is shopping now. It's asking questions now. The hotels that can answer in the machine's own form will be the ones it can book. The rest stay visible to people and quiet to machines — listed, scrapeable, and increasingly skipped.

Going AI-native inside the building is a plan you write. Being bookable by a machine is a date already on the calendar. One of those you control.

by Markus Busch, Editor and Publisher of Hospitality.today

Read also: Hotels in the cart / The channel is still forming / Why AI needs to sit beside your PMS, not inside it / Nobody gets bypassed

Enjoying this analysis? Hospitality.today delivers daily insights on hotel distribution, AI trends, and travel commerce — straight to your inbox. Subscribe for free at Hospitality.today →