Google launches early access for AI agents to Google Home devices in the US
Google has begun early access to the Model Context Protocol (MCP) server for the Google Home ecosystem in the United States. It enables AI agents that support MCP to securely interact with connected smart home devices and access event history.
As TechCrunch reports, the new capability will work with Claude, Hermes, OpenClaw, ChatGPT and Google Antigravity, among others. Users will be able to give agents natural-language commands to view camera summaries, monitor smart home activity, control connected devices and create their own smart home dashboards.
Which devices are supported
The system will support devices from the Google Home ecosystem, including Google Nest doorbells and thermostats. It will also work with Works with Google Home or Matter devices, such as smart light bulbs.
More current news is available on the UA.News Telegram channel Telegram.
How to set up access
To connect, a user needs to create a project in Google Cloud and configure it to use Home MCP. After that, the MCP configuration data must be provided to the selected AI agent and it must be asked to perform the setup. The agent will prompt the user to sign in to their account and grant permissions. Setup instructions are also expected to be posted in the Google Home Developer Center.
Access will be rolled out from the launch over the coming weeks for US subscribers to Google Home Premium Advanced, which costs $20 per month. This plan includes, among other features, longer video event history, descriptive notifications and detailed alerts, video archive search and daily summaries. Google did not comment on whether or when Home MCP will become available for other plans or in other markets.
