• Powering the next generation of agents with Google Cloud databases Vice President, AI & Databases Sr. • Product Manager, AI for Databases Our most intelligent model available yet for complex tasks on Gemini Enterprise and Vertex AI For developers building AI applications, including custom agents and chatbots, the open-sourceModel Context Protocol (MCP)standard enables your innovations to access data and tools consistently and securely. • At the end of 2025,we introduced managed and remote MCP supportfor services like Google Maps andBigQuery, establishing a standard method for AI to connect with tools, and effectively creating a universal interface for applications. • Today, we are expanding this offering to include PostgreSQL withAlloyDB,SpannerandCloud SQL, as well asFirestoreandBigtablefor high-performance NoSQL workloads, and introducing a newDeveloper Knowledge MCP server, which presents an API to connect IDEs to Google’s documentation. • These servers run in Google Cloud, providing a secure interface for Gemini and other MCP-compliant clients to easily interact with data and infrastructure. • With the launch of Gemini 3, developers gained advanced reasoning capabilities to plan, build, and solve complex problems.

Article Summaries:

  • Google Cloud announced an expansion of its Model Context Protocol (MCP) servers to support additional database services-AlloyDB for PostgreSQL, Spanner, Cloud SQL, Firestore, and Bigtable-alongside a new Developer Knowledge MCP server. The move allows AI agents built on Gemini and Vertex AI to securely and seamlessly query and manage these databases without deploying infrastructure, providing enterprise‑grade auditing, observability, and governance. Building on the 2025 launch of managed MCP for Google Maps and BigQuery, Google is positioning itself as a unified interface for AI to interact with data and tools. Developers can now configure MCP endpoints in agent settings to access operational data and automate workflows across relational, graph, and NoSQL workloads.

Sources: