Google’s WebMCP Preview: AI Agents Can Now Interact With Websites
▼ Summary
– Google announced WebMCP, a new protocol that provides a standard way for AI agents to interact with websites through structured tools.
– It works by having websites publish a “Tool Contract” via a browser API, allowing AI to directly call defined functions instead of guessing actions.
– The protocol introduces two new APIs: a Declarative API for standard HTML form actions and an Imperative API for complex JavaScript interactions.
– Key use cases include helping AI agents handle travel bookings, customer support tickets, and ecommerce shopping with greater speed and accuracy.
– Industry experts consider this a significant shift for technical SEO and the future of agentic web experiences, with a preview currently available.
Google has unveiled an early preview of WebMCP, a new protocol designed to standardize how artificial intelligence agents interact with websites. This initiative aims to provide a structured framework, allowing AI to perform user tasks with greater speed, reliability, and precision by directly communicating with site functions. The protocol could fundamentally change how users complete complex actions online, from booking travel to managing customer service requests.
At its core, WebMCP allows websites to explicitly publish a structured “Tool Contract” that AI agents can understand. Instead of an AI system attempting to guess how to navigate a page or click buttons, the website provides a clear list of available tools and functions. For example, a site might define a function like `buyTicket(destination, date)`. An AI agent can then call this function directly through a new browser API called `navigator.modelContext`, leading to more accurate and efficient interactions.
The protocol introduces two primary APIs to bridge the gap between AI agents and web functionality. The Declarative API handles standard actions that can be defined directly within HTML forms, simplifying common tasks. For more intricate operations, the Imperative API manages complex, dynamic interactions that require JavaScript execution. Together, these APIs make websites “agent-ready,” offering a more reliable and performant foundation for automated workflows than the traditional method of manipulating a page’s Document Object Model (DOM).
Practical applications for this technology are significant. In travel, users could delegate the entire process of finding and booking flights to an AI agent, which would use structured data to search, filter results, and complete reservations accurately. For customer support, agents could automatically populate detailed technical tickets, ensuring all necessary information is captured without user effort. In ecommerce, shopping becomes more intuitive as AI helpers precisely locate products, configure options, and navigate checkout flows on behalf of the user.
Developers interested in exploring this potential can apply for the early preview of WebMCP through Google’s designated channels. The emergence of such agentic experiences is being watched closely as a potential future for search and online interaction. Some industry experts have noted this could represent the most substantial shift in technical search optimization since the introduction of structured data, marking it as a significant development worth monitoring and understanding for anyone involved in web development and digital strategy.
(Source: Search Engine Land)
