Your AI agents can now control your Google Home devices
Google is opening early access to a new Model Context Protocol (MCP) server that lets compatible AI agents interact with devices connected to Google Home. Starting Wednesday, subscribers to the Google Home Premium Advanced plan in the U.S. will begin receiving access, allowing agents such as Claude, ChatGPT, Hermes, OpenClaw and Google Antigravity to control smart home devices and retrieve activity information through natural-language requests. as reported by Techcrunch The move extends AI agents beyond chat and software tasks into the physical home. Depending on the connected devices and permissions granted, users can ask an agent to operate smart home equipment, review camera summaries, inspect home activity and create custom dashboards.
What the Google Home MCP server does
MCP is a protocol that allows compatible AI agents to connect with external services and tools. In this case, Google’s server provides an interface to the Google Home ecosystem, including Google Nest doorbells and thermostats as well as “Works with Google Home” and Matter devices such as light bulbs. The policy regulation story also surfaces in EU Regulators Demand Apple and Google..., adding another angle.
That means a user could interact with supported devices using ordinary instructions rather than navigating multiple smart home apps. The system also gives agents access to event history, enabling requests related to camera summaries and other smart home activity.
Google is positioning the feature for consumers experimenting with agents that can handle everyday tasks. The company already supports MCP across areas including Google Cloud, data platforms, developer tools and Google Workspace, but the Google Home integration is aimed more directly at household use.
How to connect an agent
Setting up the connection requires several steps. Users must first create a Google Cloud project and configure it to use Home MCP. They then provide the MCP configuration details to an AI agent that supports the protocol and ask the agent to complete the setup. The ai models story also surfaces in OpenAI Unveils GPT-Red an Automated Model..., adding another angle.
The agent will request that the user sign in and grant the necessary permissions. Google says it will also provide a setup guide through the Google Home Developer Center. Once connected, the agent can work with the user’s eligible Google Home devices and available event information within the permissions provided.
Early access is limited
The rollout will begin with people subscribed to Google Home Premium Advanced in the U.S. and continue over the coming weeks. The plan costs $20 per month and includes features such as longer event-based video history, descriptive notifications, detailed alerts, video-history search tools and daily summaries. To see anthropic in practice, This AI Agent Runs Your Tasks... walks through a concrete example.
Google has not said whether Home MCP will become available to other subscription tiers or expand to additional markets. The company is instead using the early-access period to collect feedback from adopters through its Smart Home for Developers Community.
The limited launch leaves the broader direction of the feature unresolved. For now, access is restricted to a specific paid plan and market, but the underlying approach gives AI agents a new way to connect natural-language interactions with devices, cameras and activity data inside the home.