Ai agents are sold as solutions for travel, answering business questions, and solving problems of all kinds, but they work with devices and data outside of the interface. Developers need to plug in multiple connectors and get them to work, but it’s a fragile approach that’s difficult to scale and create generational headaches.
Google claims that it is trying to solve that by launching a complete MCC server that will create Google and cloud services – like maps and large to plug into.
The move follows the launch of the Google’s Latest Gemini 3 Modeland these companies are looking for strong reasons with more reliable connections to real-world tools and data.
“We’re making people Google-Ready agents by design,” Steren Giannini, director of product management at Google Cloud, told TechCrunch.
Instead of spending a week or two setting up a connector, developers can now paste in URLs to endpoints, Giannini said.
At launch, Google started with MCO servers for maps, maps, computing engines, and kumberets engines. In practice, this may look like a live query analytics writer, or an OPS agent interacting with infrastructure services.
In the case of maps, Giannini said, without MCP, developers would rely on the science built into the model. “But by giving the agents (…) a tool like the MCP Google Maps server, then they talk about real, up-to-date location information for places or travel plans,” he said.
TechCrunch events
San Francisco
I’m fat
October 13-15, 2026
While the MCP server will be offered to all Google devices, it is initially being launched in public preview, meaning it is not yet closed by the terms of the Google Cloud Service. However, it is offered to companies that have paid for Google services at an additional cost.
“We hope to bring it to the public for general availability in the new year,” said Giannini, adding that he has more MCP servers for weekly tricks.
MCP, which is a model context protocol, was developed by Anthropic about a year ago as an Open-Source standard for connecting AI systems with data and tools. The protocol has been widely adopted around the world Al agent, and anthropotropic earlier this week Donated MCP to the New Found Linux Fund Dedicated to open-source and AI agent infrastructure.
“The beauty of MCP is that, because it’s a standard, if Google provides a server, it can connect to the client,” Giannini said. “I’m looking forward to seeing how many more clients will show up.”
One can think of the MCP client as an AI application at the other end of the cable that talks to the MCP server and calls the tools it offers. For Google, it includes Gemini Cli and Ai Studio. Giannini said he also tried Chatgpt Claude Anthropik as a client, and “he just works.”
Google insists this is not about connecting agents to services. The larger Play Enterprise is Apigee, an API management productwhich many companies already use to issue API keys, set quotas, and monitor traffic.
Giannini says apige may be able to “translate standard APIs into MCP servers, into API product markup points into API Product Catalog APIs into discoverable Agents, existing government security and control layered on top.
In other words, the same Guardrails companies use in human applications can now be applied to AI agents as well.
Google’s new MCP server is protected by a permission mechanism called Google Cloud Iam, which can protect what agents do with the server. They are also protected by Google’s model armor, which Giannini describes as a firewall dedicated to the channeling of agents who discuss religious injection and data exfiltration. Administrators can also rely on log logging for additional observers.
Google plans to expand MCP support beyond the initial set. In a few months, the company will roll out support for services in areas like storage, databases, logging and monitoring, and security.
“We build the plumbing so the developer doesn’t have to,” Giannini said.