Your Hotel’s New Front Door: What the New MCP Means for Hospitality

In 1873 there were no phones. Hotels took reservations by mail. A guest would write a letter, the hotel would write back and a few weeks later a room was booked.

Then came the telephone, and suddenly guests could call to book a room in minutes. Then came the fax machine, then the Internet and with it the OTAs. Each of these technologies changed the way travelers found and booked hotels. And each time the hotels that thrived were the ones that adapted faster.

We are now at the beginning of another one of these shifts. And this time, it has a name that sounds like it belongs to a tech lab rather than a hotel frontdesk: the Model Context Protocol mostly referred to as MCP.

Acronyms apart, this might be the best opportunity hotels have had in decades to reclaim something they have been losing to OTAs for years: the direct relationship with the guest.

Part I: What is an MCP and What Problem Does it Solve?

There’s something every hotelier knows too well. Your hotel runs on multiple systems: a PMS for operations, a Booking Engine for reservations, a CRM for guest relationships, Channel Manager for distribution and, most likely a Revenue Manager System for pricing. Each of these systems speaks its own language. Getting them to talk to each other requires custom integrations and technical teams to do the development work.

Now a new player enters the scene: AI assistants. Tools like ChatGPT, Google Gemini and Claude are quickly becoming the way people search for information including hotels. A 2026 survey by TakeUp found that 90% of US Travellers are already aware that AI tools can help them plan or book travel and, among those, 63% have tried to use AI for most or every trip.

The problem is these AI assistants are smart but they are also blind. They don’t know your hotel’s real-time rates. They don’t know if you have a sea-view suite available for next Saturday. They don’t know about your complimentary breakfast or your pet-friendly policy. Without access to your hotel’s data they can only offer generic answers that will probably end up redirecting the guest to an OTA that will give them the data.

This is the problem the MCP solves. MCP was introduced by Anthropic in November 2024 as an open standard. Think of it as a universal adapter, like a USB-C cable but for AI Systems.

Without USB-C every device had its own charger: phone, camera, laptop. USB-C standardized all of that into one connector. MCP does something similar: it creates a single standardized way for AI assistants to connect to your hotel’s systems and access real-time information.

The beauty of it is being a standard. Before MCP, if a hotel wanted ChatGPT to access its booking data, someone would have to build a custom connection just for ChatGPT. Then another one for Gemini. Another for Claude. That’s expensive and slow. With MCP, you build one connection and any AI assistant that supports the protocol can use it. One integration, many AIs.

As Google Cloud describes it, MCP allows AI tools to move beyond static knowledge and become dynamic agents that can retrieve current information and take action.

So How Does It Work Without the Jargon?

Imagine a guest sitting on their couch asking an AI assistant on their phone to find a pet-friendly boutique hotel in Lisbon for the next weekend under $200 a night.

Without MCP, the AI will search the web, find generic listings, maybe pull some information from an OTA and give an approximate answer. The guest would then have to go through multiple websites to verify the information, check prices and book.

With an MCP the AI directly queries your hotel’s system. It checks real-time availability, sees your actual rates, reads your pet policy and can even complete the booking right there in the conversation. No OTA in the middle, no commission. It’s just a direct booking from a conversation.

Mike Rawson, CIO of citizenM put it well: MCP shifts hotels “from coding to describing intent”, enabling anyone from general managers to front desk teams to shape the guest journey without waiting on developer backlogs.

Part II: Why this is a Turning Point for Hotels

OTAs currently hold about 55% of hotel booking market share. The commissions they charge typically range from 15% to 25% per booking. In a $200 room that’s $30 to $50 going to the OTA every single night.

Hotels have been fighting this battle for years, investing in their own websites and direct booking incentives. And it’s worked to some extent: hotel websites consistently deliver the highest value per booking (average of $516 according to Siteminder’s 2025 data), because direct guests tend to book longer stays and add extras.

But now there’s a new challenge: travelers are increasingly starting their search not on Google, not on Booking.com but inside AI assistants. According to Pew Research Center, 57% of U.S. adults already interact with AI at least several times per week. And in travel specifically, 86% of travellers who’ve used AI have done so to find or book accommodation.

The question is not if AI will change hotel distribution. It’s who will control the conversation when it does. And this is where MCP becomes genuinely exciting for hoteliers.

Without MCP, the AI ecosystem risks becoming just another intermediary layer and OTAs are already positioning themselves to dominate it. Booking.com and Expedia were among the first travel companies to integrate their data with ChatGPT’s new app ecosystem. If your hotel’s only presence in the AI world is through an OTA’s listing, you’re paying commission on a new channel, with less control than before.

This pattern isn’t new. OTAs didn’t win because they had better hotels. They won because they showed up first. When the Internet arrived, most hotels didn’t even have a website. OTAs filled that vacuum, and hotels have been paying for it ever since. MCP is the chance to not repeat that mistake.

With MCP, hotels can plug their own systems directly into AI platforms. Your rates, your availability, your descriptions and your photos all served directly from your data with no intermediaries.

Perhaps the most remarkable part of this story is who made the first hotel booking through MCP. It wasn’t a Marriott, it wasn’t a Hilton. It was Rose St. Gardens, a five-unit short-term rental on the island of Dominica, in the Caribbean. The booking was made directly through ChatGPT, with no OTA in the middle, using MCP to connect the guest’s request to the property’s real-time data.

This is the promise of MCP in a nutshell: it levels the playing field. A boutique hotel in the Azores can have the same AI visibility as a global chain, because MCP doesn’t care about your marketing budget. It cares about your data.

The properties that structure their information well, that make their uniqueness machine-readable will be the ones AI recommends. Not because they paid for a premium listing, but because they had the best answer to the traveler’s question.

Challenges Hotels Should Consider

None of this will happen by itself. There’s a path with real obstacles between where hotels are today and where MCP can take them.

The first is the competition for the channel itself. Booking and Expedia are already plugging into AI platforms and they have massive structured datasets and engineering teams to move fast. If hotels wait too long, there are risks that this is just another channel for OTAs to collect commissions.

Another big obstacle for hotels is themselves. Amenities described inconsistently, pet policies buried in a PDF, room descriptions that haven’t been updated in years, rates that live in one system while photos live in another. For an AI agent to recommend your hotel, it needs to be able to read your hotel. And right now, most properties are simply not readable.

An IDeaS survey found that only 54% of hoteliers use integrated tools. In other words, nearly half of the industry is running disconnected systems that can’t even talk to each other, let alone to an AI. Before thinking about MCP, many hotels will need to do something far less glamorous: clean up, structure and connect the information they already have. In some cases digitalize information that exists in the hotel but not on the systems themselves.

The real work is to make sure your hotel has a story that a machine can understand and tell it on your behalf.

Your Hotel’s new Front Door is Already Opening

It’s still early. MCP is not ubiquitous and the technology is still evolving (fast). But that’s precisely the point: the hotels that start preparing now will be the ones best prepared when AI driven bookings become mainstream. And based on data, that moment is approaching faster than most expect.

MCP is the next revolution in hospitality distribution and unlike the previous shifts it comes with a twist: it gives hotels a chance to reclaim the direct relationship with the guest in a world where AI is becoming the first point of contact.

The front door of your hotel used to be a physical entrance. Then it became a website, now it’s a conversation between a traveler and an AI that knows everything about your property because you told it.

The first booking was made by a five-unit rental in the Caribbean. The hotels that will win in the age of AI won’t be the biggest. They will be the ones with the best story to tell and the smartest way to tell it.