Google Home gains AI agent control via new MCP server
Google has launched early access to a Model Context Protocol server for Google Home, enabling AI agents like Claude and ChatGPT to manage connected devices and review camera summaries using natural language. Users need a Google Cloud project and the premium Home subscription. The rollout begins today for U.S. subscribers of Google Home Premium Advanced.
The new MCP server bridges Google Home’s ecosystem with third-party AI agents, requiring users to set up a Google Cloud project and grant permissions through their chosen assistant. Supported devices span Nest hardware and Matter-compatible accessories, while the rollout is limited to U.S. subscribers of the $20-per-month Premium Advanced tier, which already includes enhanced video history and descriptive alerts. Google is gathering early feedback through its developer community, though no timeline exists for broader availability.
This move extends Google’s existing MCP adoption across its cloud, developer, and Workspace products into the consumer smart home space. By opening device control to agents like Claude and ChatGPT, Google positions its ecosystem as a neutral platform for agent-driven automation, rather than locking users into its own Assistant. The premium-tier restriction suggests a deliberate focus on power users willing to pay for advanced features.
This integration could shift how households interact with smart devices, making natural-language control more accessible beyond dedicated apps or voice assistants. Users with premium subscriptions may gain convenience, but the requirement for a Google Cloud project adds technical friction, potentially limiting adoption to enthusiasts. Broader rollout could democratize agent-based home management, yet privacy concerns around granting third-party AI access to camera feeds and event history may temper enthusiasm. The impact likely hinges on trust and ease of setup.