Google Home Adds MCP Support for Third-Party AI Agents

Google has launched early access to Home MCP, an integration that lets third-party AI agents control devices, monitor states, and analyze event history in the Google Home ecosystem.

Google Home now supports the Model Context Protocol for agentic AI control.
The company rolled out the feature on September 16, 2026. It enables any AI agent that supports MCP, including Google Antigravity, Claude, Hermes, and OpenClaw, to securely interact with all devices and historical data in a user's Google Home setup. Taylor Lehman, group product manager at Google Home & Nest, described the integration in a blog post as one that connects chosen AI agents to real-world events.

Users gain capabilities such as cross-camera analysis to review what occurred when children returned from school, querying device state history to determine weekly laundry loads or light usage duration, and having agents send voice messages via Google Home speakers upon task completion. Agents can also generate custom dashboards for device control using natural language instructions.

Setup requires a Google Cloud project and a paid subscription.
Access begins rolling out immediately and continues over the coming weeks exclusively to Google Home Premium Advanced subscribers in the United States. The tier costs $20 per month or $200 per year and already includes extended video history, detailed alerts, video search tools, and daily summaries. Users must create a Google Cloud project, configure it for Home MCP, provide details to their selected agent, and complete an authentication flow that includes signing in and granting permissions. A setup guide appears in the Google Home Developer Center.

The integration works across the full range of Google Home devices.
Supported hardware includes Google Nest doorbells, thermostats, cameras, and any Matter or "Works with Google Home" products such as light bulbs, locks, and appliances. Home MCP is designed to let agents discover homes and devices, read state and history, and run supported device actions through the Google Home ecosystem. Google Home smart speakers continue to use Gemini for primary control, but MCP-enabled agents can deliver audio notifications through them.

Security and privacy concerns accompany the expanded agent access.
Granting AI agents control over physical devices such as smart locks, HVAC systems, and appliances raises questions about safety and data protection. Google notes that users can revoke an agent's Home MCP access at any time through the Google Home app or My Accounts page. The company frames the protocol as an open standard that provides agents with physical context to analyze activity, summarize footage, manage devices, and build personalized interfaces.

The rollout marks a step toward more flexible, agent-driven smart home management while maintaining Gemini as the default interface for the Home app and speakers. Additional regions and broader availability details have not yet been disclosed.