Google Home integrates MCP to support third-party AI agents

Google is launching Google Home MCP, a new integration utilizing the standardized Model Context Protocol. This update allows third-party AI agents—including Anthropic's Claude, Open Claw, Hermes, and Google's own Antigravity—to securely monitor and control devices within the Google Home ecosystem. The feature is initially rolling out in the United States to Google Home Premium Advanced subscribers, which costs $20 per month or $200 per year. To set up the integration, users must create a Google Cloud project and configure it to connect with the Home MCP.

With this integration, external AI agents can analyze device state history, perform cross-camera analysis, and build custom control dashboards. For instance, an agent can track laundry cycles, determine how long lights were left on, or broadcast audio messages through Google Home speakers. While Gemini for Home remains the primary interface for standard voice commands, MCP adds a deeper layer of control. Google has implemented rate limits and safety protections to prevent unauthorized actions, such as preventing agents from unlocking doors.

For smart home developers and AI practitioners, this move represents a significant shift in how Google positions its IoT ecosystem. By opening up its underlying data and control layer, Google is attempting to become the foundational infrastructure for smart homes, akin to how Amazon Web Services operates for cloud computing. Developers can now use agentic tools like Antigravity to build custom, context-aware smart home applications that leverage real-world device history. However, developers must weigh this opportunity against Google's history of abandoning smart home platforms, such as Android @ Home, Weave, Project Brillo, Works with Nest, and Google Assistant.