Hotels have already chosen their AI. Most of them just aren’t allowed to plug it in. That’s the real story in Apaleo’s latest numbers, and it should worry every hotel tech vendor still deciding which assistant its customers are “permitted” to use.
The industry keeps framing AI as a buying decision: which chatbot, which revenue tool, which shiny add-on from which supplier. I think that’s backwards. Your team already picked. The question now is whether your systems will let that choice reach your reservations, rates and guest data. If they won’t, you don’t have an AI problem. You have a lock-in problem wearing an AI badge.
What actually happened
According to Hospitality Net, Apaleo, the API-first property management system (PMS), says use of its Model Context Protocol (MCP) server grew 70% between July and September 2026. Growth sped up through the quarter: 23% from July to August, then 38% from August to September.
MCP is an open standard that gives an AI agent one consistent way to read live property data and act on it, instead of a custom integration for every tool. In plain terms, it lets a hotelier type what they want into the assistant they already use, and the agent makes the change in the PMS, with the user’s permissions and an audit trail.
The examples are refreshingly unglamorous. Smartments amended around 2,000 reservations in one action, work Apaleo says would have taken about 50 hours. Brera Serviced Apartments produced a forecast that previously tied up three people for two weeks. Appenzeller Hotels, a three-property Swiss group, runs a daily upgrade workflow in Claude that frees lower room categories for resale.
The number that matters isn’t 70%
Let’s be fair about the headline. This is a vendor reporting growth in its own product, and 70% of an undisclosed starting point could be a lot or a little. I wouldn’t build a strategy on it.
The more telling figures come from the h2c Hotel Chain Tech Report 2026, quoted in the same release. 91% of hotel chains now use AI. External tools such as ChatGPT account for 60% of day-to-day AI use, against just 27% for AI built into vendor systems. Yet only 15% of chains have a standardised AI data access layer, and 52% have no plans to build one.
Read that again. Staff have voted with their browsers. They use general-purpose assistants far more than the AI their suppliers bolted on. But more than half of chains have no plan to connect those assistants to the systems where the work actually happens. So people copy and paste between ChatGPT and the PMS, which is exactly the “adopted AI, kept the spreadsheets” pattern I wrote about in Hotels adopted AI. The spreadsheets didn’t leave.
Vendors choosing your AI is the new rate parity
Apaleo’s founder, Philip von Ditfurth, put it bluntly: most hoteliers have picked the AI they want, and too often the vendor decides whether they can connect it. Of course he would say that. Openness is his sales pitch. But he’s right.
For years, hotels accepted systems that held their own data hostage behind paid connectors and slow integrations. Doing the same with AI would be a bigger mistake, because the assistants are improving every few months. Locking yourself to whatever model your PMS supplier signed a deal with this year means you are stuck with yesterday’s tool while competitors swap to tomorrow’s.
If I ran tech procurement for a hotel group, I’d add one question to every renewal: can I connect any AI agent I choose, with proper permissions and logging, without paying you for the privilege? A vague answer tells you everything.
This is a marketing story too
Here’s why marketers should care. The same plumbing that lets your own staff point an agent at live rates is the plumbing that lets outside agents see and book you. Hilton is now claiming it’s plugged into every AI agent, and Wyndham talks about AI agents as a new type of guest. The big brands are building open pipes on purpose.
An independent hotel whose systems can’t talk to agents internally will struggle to be visible to them externally. As I argued in AI isn’t snubbing independent hotels. It just hasn’t heard of them., being invisible to machines is now a commercial problem, not a technical footnote.
The encouraging part of Apaleo’s report is that adoption isn’t limited to big groups with engineering teams. Family-run and independent hotels are building their own agents because chat interfaces removed the need to write code. That should end the excuse that AI is only for the chains.
Start small, but start with access
I like that operators are beginning with staff-facing jobs, where a human stays accountable, rather than unleashing agents on guests. Bulk reservation changes, forecasting and reporting are exactly the dull, expensive work AI should eat first.
But none of it happens without access. My take: in 2027, the hotels that win with AI won’t be the ones with the cleverest model. They’ll be the ones that refused to let a supplier choose it for them. As I said in Hotels have bought ChatGPT. They haven’t bought an AI strategy., the strategy starts with your data, and who controls the door to it.
So, is your PMS an open door or a locked one? And would you switch supplier over it?
Want more takes like this on AI, search and travel marketing? Subscribe to my LinkedIn newsletter for weekly analysis you can actually use.