Apaleo, the API-first property management platform for hospitality businesses, has reported a 70% increase in the use of its Model Context Protocol (MCP) server between July and September 2026, as hotels move AI agents out of testing and into live operations. Growth accelerated through the quarter, rising 23% from July to August and 38% from August to September.
Apaleo launched agentic hospitality with the first MCP server in the hotel PMS category in September 2025, at a point when the industry was still some way from putting agents to work. Adoption has risen steadily since, with the sharpest growth in the last quarter.
The same growth has spread across the customer base as opposed to concentrated in large, tech-forward groups. Apaleo sees comparable adoption across segments and verticals, from independent and family-run hotels without an in-house engineering team through to enterprise operators managing rooms and apartments at portfolio scale.
According to the h2c Hotel Chain Tech Report 2026, 91% of hotel chains now use AI, up from 78% a year earlier, while only 15% have implemented a standardised AI data access layer, such as MCP, and 52% have no plans to do so. Integration with existing systems remains a barrier for 38% of respondents.
“A year ago, hoteliers were asking whether AI agents could do real work. Now they are asking how quickly they can hand more over. What changed is access.
“Once an operator can point an agent at live reservations, rates and guest data, the work gets done. We are seeing PMS migrations that used to take weeks run in hours, and forecasting work that occupied three people for a fortnight come back the same day.
“Most hoteliers have already picked the AI they want to use. The question is whether their systems let them connect it, and too often the vendor decides that for them.”
Philip von Ditfurth, Founder, Apaleo
The Model Context Protocol is an open standard built for AI agents. It gives an agent one consistent way to read live property data and act on it, covering reservations, rates, availability and guest profiles, instead of a custom integration for every AI tool a hotel wants to use. Apaleo’s MCP server extends that standard across its whole platform, so an agent works through a single connection, with the permissions of the user it acts for and a full audit trail.
That governance is becoming more relevant as agents take on more work. The same h2c research found reliance on AI climbing from 4.7 to 5.4 out of ten over the past year while trust held flat at 6.5. Chat interfaces have made the MCP server reachable for people who do not write code. Hoteliers describe what they want in plain language, and the agent carries out the change in the system, which is why operators without technical teams are building and running their own agents.
The choice of model matters as most property management systems determine which AI assistant the operator can enable. Apaleo’s customers connect to whichever model they prefer and can swap it as the tools improve. h2c found that external AI tools, such as ChatGPT, account for 60% of day-to-day AI use among hotel chains, against 27% for AI embedded in vendor systems, so most hoteliers have already chosen the assistant they want to work with.
Apaleo also sees a pattern once operators start. The first use case is usually small and contained, and teams quickly expand it as they work out what else can be handed over. Adoption so far is concentrated in staff-facing work, where accountability sits with the team rather than the guest.
Appenzeller Hotels, a three-property group in Switzerland, runs a daily revenue workflow in Claude that identifies which bookings can be upgraded to free up lower room categories for resale, saving about two hours per week alongside booking reconciliation and reporting.
Brera Serviced Apartments added its own MCP alongside Apaleo’s and built a data lake bringing revenue and cost data together. Its IT team used last year’s balance sheet to generate a forecast covering both halves of this year, then passed it to the controller for budgeting, work that previously occupied three people for two weeks.
McDreams Hotels uses agents to automate repetitive processes and to answer day-to-day questions that would previously have required a report, with staff querying their own data several times a day.
“We can automate repetitive processes, get relevant information faster, and use AI to help with questions that would otherwise take hours to answer.
“We’re still on a journey, but the potential is not just about saving time. It’s about making better use of the data we already have.”
Christoph Klein, Chief Marketing Officer, McDreams Hotels
LikeStay uses the MCP server alongside AI to turn existing price lists into rate plans, analyse reservation patterns and speed up PMS migrations, moving data from other systems onto its Apaleo-based stack in a few hours. At the enterprise end, Smartments used the MCP server to amend around 2,000 reservations in a single action, work that would otherwise have taken about 50 hours.
