Google Lets AI Agents Control Google Home Devices in Early Access

Google has begun rolling out early access that lets compatible AI agents control Google Home devices and work with household activity records. The release makes an agent chat a potential control surface for lights, thermostats, cameras and other connected devices, after a user authorizes access.

The change comes through a Model Context Protocol, or MCP, server for the Google Home ecosystem. MCP is a way for an AI agent to connect with outside tools and services. Google’s compatibility list includes Claude, Hermes, OpenClaw, ChatGPT and Google Antigravity.

The connection goes beyond basic device commands. Users can issue natural-language instructions to review camera summaries, monitor smart-home activity, check event history and build custom dashboards. Supported hardware includes Google Nest doorbells and thermostats, plus Works with Google Home and Matter devices such as light bulbs.

It is not yet available to every Google Home customer. Access will roll out over the coming weeks to U.S. subscribers of Google Home Premium Advanced, Google’s $20-per-month tier. The plan also includes longer event-based video history, descriptive notifications, detailed alerts, video-history search and daily summaries.

That sequence makes the initial release a multi-step configuration process rather than a one-click consumer feature. It also puts the permission decision directly in front of the user: the agent needs Google authorization before it can interact with the connected home.

Google is soliciting early-adopter feedback through its Smart Home for Developers Community. The company has not said whether or when Home MCP will expand to other subscription tiers or markets. The rollout remains a narrow test of whether people will connect an AI agent to their homes’ controls and records.