A group of major tech companies, including Google and Microsoft, have introduced a new web API called WebMCP. This JavaScript-based interface allows web apps to give AI agents the ability to interact directly with web pages. The goal is to let AI tools work alongside humans within the same web interface more seamlessly.

What is WebMCP and How Does It Work?

WebMCP is designed as a way for developers to expose specific parts of their web applications as “tools.” These tools are JavaScript functions that come with descriptions and structured schemas, making it easier for AI agents, browsers, and assistive tech to invoke them. This setup turns web pages into Model Context Protocol (MCP) servers, where client-side scripts handle these tools instead of relying solely on server-side processing.

By doing this, WebMCP enables collaborative workflows where users and AI agents can work together within the same interface. For example, an AI assistant could help fill out forms, browse products, or book flights directly on a website, all while the user watches and interacts. The API aims to make these interactions faster, more reliable, and more precise than traditional methods.

Early Preview and Use Cases

Google has released an early preview of WebMCP, and it’s available for testing. Andre Cipriani Bandarra, a developer relations engineer at Google, explained in a February blog post that the API is meant to standardize how web tools are exposed to AI agents. This means developers can create structured, natural language-friendly functions that AI can easily understand and invoke.

The WebMCP community group published a draft report on February 12 outlining the API’s details. The report describes WebMCP as a way to bridge web apps and AI, with use cases spanning customer support, e-commerce, and travel. For instance, AI agents could help users submit support tickets, find products, or book flights—all within the same web interface.

Two main APIs are proposed within WebMCP: a declarative API for simple actions that can be embedded directly into HTML forms, and an imperative API for more complex, dynamic interactions requiring JavaScript. These APIs are meant to make websites more “agent-friendly” and improve the performance and reliability of AI-driven workflows.

While WebMCP is not yet a formal W3C standard, it’s a step toward creating more integrated and intelligent web experiences. Developers and companies interested in AI-powered web tools can explore the early preview and contribute feedback to help shape its future development.

Overall, WebMCP aims to open new possibilities for how websites and AI agents work together, making digital interactions more efficient and user-friendly. As the technology matures, we can expect to see smarter customer service bots, more intuitive online shopping, and streamlined booking processes—all happening within the same web pages we use every day.

Inspired by

• https://www.infoworld.com/article/4133366/webmcp-api-extends-web-apps-to-ai-agents.html

Sources

• chrome.com
• chrome.com
• github.io
• github.io
• github.io