Getting AI agents to work seamlessly with tools and data beyond chat interfaces is no easy feat. To tackle this, Google has unveiled a fully managed, remote MCP server. This innovation is designed to streamline the connection process between AI agents and Google's ecosystem, including its cloud services. This strategic move comes hot on the heels of the Gemini 3 model's debut, with the goal of blending robust reasoning abilities with practical tools and real-world data.

During its initial rollout, Google has launched MCP servers tailored for services like Maps and BigQuery, significantly cutting down on the integration hassles for developers. As an open-source framework, MCP has already gained widespread acceptance. Currently, Google's MCP servers are in public preview mode, offering complimentary access to paying enterprise clients. A full-scale launch is anticipated early next year.

Google has underscored that this endeavor not only bridges the gap between services and AI agents but also incorporates its API management tool, Apigee. Apigee has the capability to transform standard APIs into MCP servers. Security is a top priority, with MCP servers fortified by permission controls and firewalls. Looking ahead, Google intends to broaden MCP support across additional service domains in the months to come.