Google Opens Google Home to Claude and Other AI Agents — Here’s What They Can Control

Google Home now lets compatible AI agents like Claude Cowork access authorized smart-home data and controls through a new Model Context Protocol server. This expands the role of AI agents beyond answering questions to performing approved actions. The system provides five core capabilities: discovering homes, finding connected resources, checking device states, reviewing historical events, and executing supported actions.

Users can connect agents via an OAuth authorization flow, with Google listing Antigravity and OpenClaw as compatible clients. Agents can query device status, check if a home is secure, turn off outdoor lights, and inquire about past events. However, some physical actions are restricted, such as unlocking doors, and users can revoke agent access at any time.

This integration reflects Google's growing presence in household AI alongside its $99 Gemini-powered Home Speaker. As AI agents gain more credentials and authority, careful access controls become crucial to prevent security gaps.