AI agents are taking the world by storm, promising to revolutionize how we plan trips, answer business queries, and tackle complex problems. However, integrating these agents with external tools and data beyond their chat interfaces has proven to be a significant challenge. Developers have had to resort to cobbling together various connectors, a fragile and cumbersome approach that is difficult to scale and manage.
Enter Google, aiming to simplify this process with its new fully managed, remote MCP servers. These servers are designed to seamlessly integrate Google's array of services, such as Maps and BigQuery, with AI agents, making it easier for developers to harness the power of these tools.
Google's latest move follows the launch of its Gemini 3 model, which aims to combine advanced reasoning capabilities with reliable connections to real-world data and tools.
"We're making Google agent-ready by design," says Steren Giannini, Product Management Director at Google Cloud.
Instead of spending days setting up complex connectors, developers can now simply paste in a URL to a managed endpoint, according to Giannini.
Initially, Google is launching MCP servers for Maps, BigQuery, Compute Engine, and Kubernetes Engine. This could mean an analytics assistant directly querying BigQuery or an ops agent interacting with infrastructure services.
In the case of Maps, Giannini explains, without the MCP server, developers would rely on the model's built-in knowledge. However, with the MCP server, the agent gains access to up-to-date, real-world location information, enhancing its trip planning capabilities.
The MCP servers will eventually be available across all Google tools, but for now, they are offered under public preview, meaning they are not yet fully covered by Google Cloud's terms of service. However, enterprise customers who already pay for Google services can access them at no additional cost.
"We expect to bring them to general availability very soon," Giannini adds, anticipating a steady stream of new MCP servers in the coming weeks.
MCP, or Model Context Protocol, was developed by Anthropic as an open-source standard to connect AI systems with data and tools. It has gained widespread adoption in the agent tooling community, and Anthropic recently donated MCP to a Linux Foundation fund dedicated to open-sourcing and standardizing AI agent infrastructure.
"The beauty of MCP is its universality," Giannini explains. "Since it's a standard, any client can connect to Google's servers. I'm excited to see the range of clients that will emerge."
MCP clients can be thought of as the AI apps at the other end of the connection, communicating with MCP servers and utilizing the tools they offer. For Google, this includes Gemini CLI and AI Studio, but Giannini has also successfully tested it with Anthropic's Claude and OpenAI's ChatGPT.
Google emphasizes that this initiative is not just about connecting agents to its services. The real enterprise value lies in Apigee, Google's API management product, which many companies already use to manage API keys, set quotas, and monitor traffic.
"Apigee can essentially translate a standard API into an MCP server," Giannini explains. "It transforms endpoints like a product catalog API into tools that agents can discover and use, with existing security and governance controls in place."
In essence, the same API guardrails that protect human-built apps can now be applied to AI agents as well.
Google's new MCP servers are protected by Google Cloud IAM, a permission mechanism that explicitly defines an agent's access and capabilities. Additionally, Google Cloud Model Armor acts as a dedicated firewall for agentic workloads, defending against advanced threats like prompt injection and data exfiltration. Administrators also have access to audit logging for enhanced observability.
Google plans to expand MCP support beyond the initial set of servers, with upcoming support for services in areas such as storage, databases, logging, monitoring, and security.
"We've built the infrastructure so developers don't have to," Giannini concludes.
This move by Google is a significant step towards simplifying the integration of AI agents with real-world tools and data, opening up new possibilities for developers and enterprises alike.
What are your thoughts on Google's MCP servers and their potential impact on the AI landscape? Do you think this will revolutionize how we interact with AI agents, or is there a potential downside to consider? We'd love to hear your opinions in the comments!
