Google lets AI agents control Google Home devices

Google has introduced early access to its Model Context Protocol (MCP) server for the Google Home ecosystem. This integration allows any AI agent that supports the open standard—including Anthropic's Claude, OpenAI's ChatGPT, Hermes, OpenClaw, and Google Antigravity—to securely interact with connected smart home devices and retrieve their event histories.

To establish the connection, users must create a Google Cloud project and configure it to use Home MCP. After sharing these configuration details with their chosen AI agent, the agent will prompt them to log in and grant the necessary permissions. The system supports the entire Google Home ecosystem, which includes Google Nest doorbells, Nest thermostats, and any "Works with Google Home" or Matter-compatible devices, such as smart light bulbs.

The feature is rolling out to U.S. subscribers of the Google Home Premium Advanced tier, which costs $20 per month. This premium subscription already provides features like extended event-based video history, descriptive notifications, video search tools, and daily summaries. Google has not yet commented on when or if MCP support will expand to other subscription tiers or international markets.

For developers and smart home enthusiasts, this update bridges the gap between large language models and physical automation. Users can now leverage natural language instructions to review camera summaries, monitor activity, control appliances, and build custom smart home dashboards. By leveraging MCP, Google is expanding its existing support for the protocol beyond Google Cloud, developer tools, and Google Workspace, directly targeting consumers who want to automate daily tasks with AI.