What to do now

Get your property data AI-ready. Before MCP matters, your hotel information needs to be structured, accurate, and machine-readable. Schema markup, clean HTML, comprehensive FAQ content, and current rates form the foundation everything else builds on.

Watch the MCP ecosystem develop. Cendyn AI Connect is in beta now. DirectBooker, LiteAPI, and Agentic Hospitality are building similar infrastructure. If your technology stack already connects to these platforms, you're closer to MCP readiness than you think.

Treat AI as a distribution channel. The hotels that win this transition will be the ones that recognize AI search as a booking channel on par with OTAs, metasearch, and direct web traffic, and allocate strategy accordingly.

Boost Background

Here's the scenario that keeps coming up in our hospitality conversations: a traveler opens ChatGPT and asks for a boutique hotel in Lisbon with a rooftop terrace under 200 euros. The AI recommends properties based on OTA listings, review sites, and whatever it can scrape from the web. The hotel that wins that recommendation probably didn't do anything to earn it. The OTA did. The hotel pays 15 to 25% commission on a booking it never controlled.

Model Context Protocol is about to change that equation. MCP is an open standard, introduced by Anthropic in late 2024 and adopted by OpenAI, Microsoft, and Google in early 2025, that lets AI models connect directly to external data sources and APIs. For hotels, it means pushing live availability, rates, and inventory straight into ChatGPT, Claude, and Gemini without an OTA intermediary. The AI queries your data in real time and presents your direct rates at the moment the traveler is ready to book.

What MCP actually does

Think of MCP as a universal connector between your hotel's systems and AI platforms. Instead of building custom integrations for each AI tool, a hotel builds one MCP server. Any AI model that supports the protocol can then access the tools and data it exposes: room availability, pricing, policies, amenity details, and eventually booking functionality.

The distinction from traditional AEO matters. Optimizing website content helps AI recommend your property in informational queries. MCP goes further by feeding real-time transactional data directly into the AI's response. When a traveler asks for pricing, the AI queries your live inventory through MCP and returns current rates, not cached numbers from a months-old crawl.

Cendyn launched AI Connect in late 2025 in partnership with DirectBooker, making it the first major hospitality platform to push hotel ARI (Availability, Rates, and Inventory) directly into ChatGPT, Claude, and Gemini through MCP. Perplexity is already ahead on the transaction side, with TripAdvisor and Selfbook partnerships enabling end-to-end hotel bookings inside the interface today. Google has committed to the same path with Trip Planning inside Gemini and AI Mode.

Why this matters for direct bookings

The OTA dependency problem runs deep. OTAs supply 55% of sources AI cites in hotel recommendations. Independent hotels send 61% of bookings through OTAs at 15 to 25% commission. Every intermediary booking costs margin and customer data.

MCP offers a structural alternative. When a hotel's direct rates appear natively in AI answers, the booking link goes to the hotel's own site. The traveler gets the best rate. The hotel keeps full margin and owns the guest relationship.

DirectBooker's architecture illustrates how this works. Hotels connect existing data feeds (through DerbySoft, SynXis, or Cendyn's DMP) to DirectBooker's MCP server. When an AI receives a query, it routes a structured request to the server, which returns property data, pricing, and direct booking links. The traveler books directly.

The agentic future is closer than most hoteliers think

MCP is the foundation, but the real shift is agentic AI: systems that don't just recommend but act on behalf of the traveler. IDC predicts that by 2030, 30% of travel bookings will be executed by AI agents.

The next evolution won't be human-to-hotel. It will be agent-to-agent. A traveler's AI assistant will negotiate booking details with the hotel's AI system: querying availability, applying loyalty preferences, comparing value, and completing the reservation. Hotels with machine-readable inventory accessible through MCP will be in the decision set. Hotels without it will be invisible.

Google's recent introduction of UCP (Universal Commerce Protocol), compatible with MCP and major payment processors like Visa and Mastercard, signals that commerce infrastructure for this channel is being built now. By end of 2026, multiple AI platforms will support end-to-end hotel bookings without leaving the assistant interface.