Google is launching managed Model Context Protocol (MCP) servers, streamlining how AI agents interact with its services. This initiative makes Google's services 'agent-ready by design,' starting with Maps and BigQuery. The move aims to simplify integrations, allowing AI agents to use real-world tools more effectively.
These fully managed, remote MCP servers provide a unified layer across Google and Google Cloud services. Developers can now connect AI agents to a globally consistent endpoint using standard MCP clients like Gemini CLI and AI Studio. Google is extending this capability through Apigee, enabling organisations to expose their APIs as discoverable tools for agents. This integration supports custom business logic and governed data flows, with discovery and governance relying on Cloud API Registry and Apigee API Hub. Access is controlled by Cloud IAM and monitored via Cloud audit logging, with Model Armor adding protection against agent-specific threats.
The new Maps server offers access to Grounding Lite data, enabling agents to answer location, routing, and weather queries. The BigQuery server allows agents to read schemas and run queries directly on enterprise datasets. Compute Engine exposes provisioning and scaling tasks through MCP for automated infrastructure workflows, while the GKE server offers structured access to Kubernetes resources.
