roommaster plugs live PMS data into ChatGPT with an AI Connector

roommaster has shipped an AI Connector that links live PMS data to ChatGPT, Claude, Microsoft Copilot and Google Gemini. The mechanism is the interesting part: it is built on the Model Context Protocol, the open standard that lets an AI assistant read from and act on an external system without a bespoke integration for each model. In plain terms, a receptionist or revenue manager can ask a general-purpose assistant a question about their own property and get an answer pulled from the system that actually runs the hotel, in natural language.

That sounds mundane until you look at how hotel data normally reaches an AI. Most of the time it does not. A hotelier opens a chat window, exports a report, pastes numbers, and the model reasons on a stale snapshot. Meanwhile the guest-facing side of the industry has already moved: our own coverage of Hilton plugging into Google, ChatGPT and Claude showed the big brands wiring their inventory and content into assistants, and we flagged in why a hotel can rank on Google yet vanish in ChatGPT that visibility now runs through the model, not the SERP. The gap between that and the average independent property's PMS has been wide. A connector on an open protocol narrows it for anyone running roommaster.

The value sits in the query layer. Occupancy for next Tuesday, ADR against the same week last year, which rate codes are underperforming, how many arrivals still need a card on file. Those are the questions staff ask ten times a day, and today they cost clicks and tab-switching. An assistant that reads the PMS directly removes the export step, and because it runs on MCP rather than a single vendor's API, the same connection should travel across the four major models instead of locking the hotel to one. That portability is the detail worth noting, and it is where a PMS provider earns its keep beyond storing reservations.

One caution, and it is about plumbing rather than hype. Live data in a chat window means permissions matter: who can ask what, and what the assistant can write back, not just read. Any hotel rolling this out should map access by role before it goes wide. Watch whether roommaster publishes the permission model and whether other PMS vendors follow with MCP connectors of their own. If they do, the question stops being whether an AI can see your hotel's data and becomes which assistant your team prefers to ask.