With the recent release of gemini 3we have cutting-edge reasoning to help you learn, build, and plan anything. But for AI to become a true “agent,” pursuing goals and solving real-world problems on your behalf, it requires more than mere intelligence. Must be able to work with tools and data reliably.
antropic model context protocol (MCP), often likened to “USB-C for AI,” has become a popular standard for connecting AI models to data and tools. MCP enables AI applications to perform complex, multi-step tasks required to solve real-world problems. However, implementing Google's existing community building servers often requires developers to identify, install, and manage individual local MCP servers or deploy open source solutions, which is burdensome for developers and often results in weak implementations.
Today we are announcing the release of a fully managed remote MCP server. Google's existing API infrastructure has been enhanced to support MCP, providing a unified layer across all Google and Google Cloud services. Developers can now easily point standard MCP clients such as AI agents and Gemini CLI to globally consistent, enterprise-ready endpoints for Google and Google Cloud services.
Importantly, we extend this functionality to the broader enterprise stack through Apigee, allowing organizations to leverage specialized APIs for specific data flows and business logic. Customers can now publish and manage APIs that developers have built themselves, as well as external third-party APIs, as tools discoverable by agents. Read more about Apigee's announcement here.
We are gradually releasing MCP support for all services, starting with the following services:
1. Google Maps: Rooting AI in the real world
map grounding light, Available through Google Maps Platformconnects AI agents to trusted geospatial data and provides access to up-to-date information about location, weather forecasts, and route details such as distance and travel time. This allows developers to build agents that can accurately answer real-world location and travel queries without hallucinations. For example, an AI assistant can use Grounding Lite to respond to questions like “How far is the nearest park from this rental property?”, “What should I pack for the Los Angeles weather this weekend?”, or “What are the best kid-friendly restaurants near my hotel?”
2. BigQuery: Reasoning about corporate data
of BigQuery MCP server This allows agents to natively interpret schemas and run queries against corporate data without the security risks and latency of moving data into the context window. Get direct access to BigQuery features like predictions while ensuring your data is kept in the right place and managed.
3. Google Compute Engine (GCE): Autonomous infrastructure management
By exposing functionality such as provisioning and resizing as discoverable tools, this server Enable agents to autonomously manage infrastructure workflows. Agents can handle everything from initial builds to day two production, including dynamically adapting to workload demands.
4. Google Kubernetes Engine (GKE): Autonomous container operation
of GKE MCP Server exposes a structured, discoverable interface that allows agents to reliably interact with both GKE and the Kubernetes API. This eliminates the need to parse complex CLI commands that contain fragile text output or strings. This unified surface allows agents to operate autonomously or with human-involved guardrails to diagnose problems, repair failures, and optimize costs.
Built-in security and Observability
We bring order to this ecosystem with a unified approach to discovery and governance. new Cloud API Registry and Apigee API hubdevelopers can find trusted MCP tools from Google and their own organizations. This combination of ease of discovery and strict controls allows administrators to manage access by: Google Cloud IAMplease rely on me audit log Ensure and leverage observability Google Cloud Model Armor Protect against advanced agent threats such as indirect prompt injection.
“Google's support for MCP across such a diverse range of products, coupled with our close collaboration on the specification, will help more developers build agential AI applications. As adoption increases across major platforms, we will move closer to agentic AI that works seamlessly across the tools and services people already use.”– David Soria Parra, MCP Co-Creator, Anthropic Technical Staff Member
Let's take a look at an example of these new MCP servers in action.
Imagine an agent helping you identify the ideal location for your retail business. use Agent Development Kit (ADK)), you can build natural language agents that support: gemini 3 Pro connects to BigQuery to forecast revenue-based sales data, while agents can cross-reference Google Maps to explore complementary businesses and validate delivery routes. All of this is done via a standard managed MCP server.
