A proposed standard driven by Google and Microsoft has the potential to help travel and hospitality websites better prepare for artificial intelligence (AI) agents, according to industry stakeholders.
WebMCP is available for prototyping in an early preview, Chrome for Developers shared, listing travel as a likely use case.
For travel companies, tangible results could include reduced customer service issues, higher conversion rates, higher revenue and more personalized experiences, according to Robert Cole, senior research analyst covering lodging and leisure travel for Phocuswright.
At its core, WebMCP gives websites a path to becoming “agent-ready,” he said of the proposed World Wide Web Consortium standard.
“WebMCP is a great new technology,” Cole said. “It will be highly disruptive, helping those that adopt it properly and leaving those who ignore it behind.”
How WebMCP works
The emerging standard allows AI agents to perform actions on behalf of site owners with improved accuracy, speed and reliability by exposing structured tools. With WebMCP in place, site owners can tell agents how and where to interact with their website.
WebMCP offers a path, while model context protocol (MCP) serves as a translation layer between data sources and large language models (LLMs) powering AI agents and other tools.
“This direct communication channel eliminates ambiguity and allows for faster, more robust agent workflows,” wrote André Cipriani Bandarra, a staff developer relations engineer for Google.
It will be highly disruptive, helping those that adopt it properly and leaving those who ignore it behind.
Robert Cole, Phocuswright
The standard includes two new APIs that give browser agents the ability to act on behalf of a user, according to Bandarra. Declarative API allows for the performance of standard actions defined in HTML. Imperative API allows for more complex interactions requiring JavaScript execution.
“These APIs serve as a bridge, making your website ‘agent-ready’ and enabling more reliable and performant agent workflows compared to raw DOM actuation,” he wrote.
WebMCP's role in travel
In addition to helping in areas such as e-commerce and customer support, Bandarra suggested WebMCP could create a “confident agent” to help with complex travel.
AI agents haven’t had access to real-time information on availability and pricing unless companies have their own MCP integrations in place, according to Rick Egan, co-founder and head of digital for Agent.m. That has posed a limitation for consumers and travel companies.
WebMCP makes it easier for companies to expose their data and bring real-time information to consumers through LLM interfaces, Egan said.
The connection should save AI agents and LLMs from having to search or make guesses: “All the complex product details, options, rules and fees can be better organized and more efficiently communicated,” Cole said.
WebMCP also acts as a stepping stone toward what will eventually be a native AI integration model, according to Pablo Delgado, managing partner and CEO for America of Mirai.
Travel websites already expose complex actions such as search, filter and book, but those were designed for human interaction, Delgado said.
Moving those capabilities to structured APIs or native MCP servers is time consuming, and WebMCP can serve as a kind of antidote.
“WebMCP offers a practical, short-term way to make existing web flows operable by AI agents without redesigning the entire backend architecture,” he said. “In that sense, it accelerates adoption by leveraging the public web layer that already exists.”
But it doesn’t stop there.
Deeper transformation could occur when bookings aren’t dependent on a browser session, Delgado said.
“Long term, structured APIs and protocols like MCP—enabling direct system-to-system interaction—will be more efficient, reliable and scalable,” Delgado continued, noting WebMCP helps things move faster. “Native capability exposure will define the end state.”
The emerging standard allows collaborative planning between travelers and their AI assistants. It bypasses confusion and frustration that stems from conflicting information from different sources such as websites and APIs, Cole said.
Travel brands should be thinking about WebMCP as a new distribution channel for direct connection with AI agents, according to Egan.
“Similar to the early days of SEO, the more data you can expose to AI agents, the quicker you will become the trusted data source,” Egan said. “As consumers continue to trust LLMs more and look to use agents to take on more tasks for them, the companies that are early adopters are set up best to win going forward.”
As consumers continue to trust LLMs more and look to use agents to take on more tasks for them, the companies that are early adopters are set up best to win going forward.
Rick Egan, Agent.m
WebMCP's impact on industry competition
Enabling AI agents to book directly can benefit certain players in spaces like the hotel sector, according to Nick Slavin, CEO and co-founder of Curacity.
“Mid-scale and budget hotels compete on price and location, so enabling AI agents to book them directly has a straightforward upside,” Slavin said. “They can pay OTAs less to capture existing demand and leverage AI search to monetize the new demand LLMs create and shape during the discovery process.”
It’s more complicated for players such as luxury and lifestyle hotels that are selling aspiration and experience, he cautioned.
Slavin predicts the industry will bifurcate. He envisions commodity hotels using WebMCP to cut distribution expenses. On the flipside, he expects high-end brands will hold onto human engagement, justifying premiums and driving revenue.
“Giving AI agents direct booking access through WebMCP separates high-value guests from the content that justifies the high-consideration purchase,” Slavin said. “The brand imagery, the storytelling, the ability to explore room categories and experiences, it’s not just marketing, it’s what makes the premium defensible.”
WebMCP also has potential to open up industry competition.
“For very small players with limited resources, this can help level the playing field when competing against major brands, OTAs and platforms with robust IT budgets, partner agreements and advertising spend,” Cole said.
WebMCP's limitations, future indications
While WebMCP is “an exciting breakthrough,” as Egan put it, it’s early days for the standard.
“The foundational infrastructure of the agentic web is still being created,” Cole said. “While WebMCP sounds awesome—and it is—security and privacy best practices must be table stakes before any WebMCP—or MCP—server gets deployed.”
And it isn't likely to reduce reliance on other APIs, either, according to Cole, Delgado and Egan.
API connections will still be doing a lot of the heavy lifting. Cole said he sees WebMCP as a streamlined way for AI agents to access API resources when connecting through a web browser. While website design is about user interface, this is about machine interface for the website, he said.
“Inventory control, pricing logic and transaction integrity cannot rely solely on browser-layer automation,” Delgado said.
“This is the classic ‘right tool for the right job’ issue,” Cole said. “If a human and their AI agent assistant are working together when browsing a website, WebMCP is the perfect tool. If the AI agent assistant is working independently, using an MCP server connection to a backend API would probably be a better, more efficient choice.”
Bottom line, Delgado said the innovation of WebMCP reinforces the importance of “clean system architecture.” He anticipates protocols such as MCP will become more important in standardizing how trusted, structured capabilities are exposed to AI systems.
Right now, companies should consider WebMCP as a short-term enabler rather than a final structure.
“WebMCP allows structured browser automation, which is a pragmatic way to make existing booking flows AI-operable without major backend redesign,” Delgado said.
He advised companies to “think beyond the browser”: In the immediate future, structured browser automation can come through WebMCP. Longer term, direct capability exposure will come through APIs and protocols such as MCP.
The winners in the long game will be companies that expose their inventory, booking logic and pricing natively to AI ecosystems over those that use interface automation, Delgado said.
And, of course, like with any new technology, WebMCP comes with risks.
“WebMCP clearly accelerates AI adoption by making existing web flows operable by agents,” Delgado said. “But that acceleration comes with trade-offs.”
He listed some including a potential reduced urgency to build native capability layers, the fragility of browser-based interaction versus native integration, dependency on the browser layer and security and privacy concerns.
“WebMCP is a useful step forward,” Delgado said. “But it should be viewed as a transitional layer—not necessarily the optimal long-term foundation.”