Google Connects AI Agents to Real-World Data

Google is introducing managed Model Context Protocol (MCP) servers designed to streamline the integration of AI agents with its cloud services. This new offering aims to resolve the past difficulties developers faced in connecting AI chat interfaces to external tools and data. The MCP servers will initially support services like Google Maps, BigQuery, Compute Engine, and Kubernetes Engine, allowing AI agents to interact with real-world, up-to-date information more effectively.

Previously, developers relied on patching together connectors, a fragile and difficult-to-scale approach. Google's new solution enables developers to connect agents by simply pasting a URL to a managed endpoint, significantly reducing setup time. This initiative, following the launch of the Gemini 3 model, focuses on pairing enhanced AI reasoning with dependable access to real-world tools and data, making Google's services 'agent-ready by design'.

While the MCP servers are currently in public preview, they are available at no extra cost to enterprise customers. Google plans to expand MCP support to a wider range of services, including storage, databases, logging, and monitoring, in the coming months. The company is also leveraging its Apigee API management product to extend these capabilities, allowing existing API guardrails to apply to AI agents, thus enhancing security and governance.