Gemini for Google Home can now recognize objects on camera and trigger smart home routines automatically, for US subscribers on the $20/month Premium Advanced plan.
Google's Gemini AI can now watch your cameras and act on what it sees. On May 28, the company began rolling out a feature for Google Home that lets security cameras trigger smart home routines based on visual events: a yoga mat in hand dims the lights; a delivery at the door could unlock it.
The feature arrived alongside other Gemini for Home updates, including voice command improvements and stability fixes, as The Verge reported. Google calls the new capability a "starter," meaning a trigger users configure in plain English. Describe what you want the camera to watch for, select which cameras should monitor it, and the platform connects that visual cue to whatever smart home routine you have set up.
Requirements are narrow. The feature supports Nest cameras and a limited set of third-party cameras certified with "Gemini Built-In," and is only available to US users enrolled in the Google Home Public Preview program, in English. It also requires enabling a separate camera AI setting in the app and a subscription to Google Home Premium Advanced, priced at $20 per month or $200 per year.
That subscription barrier reflects a broader industry pattern. As Digital Watch Observatory has documented, artificial intelligence companies have been moving aggressively toward tiered subscription models to build predictable revenue without relying on advertising. For Google, Premium Advanced is the highest Home tier, which positions camera automation as a feature for committed subscribers rather than a default benefit for anyone who bought a Nest camera.
Gemini everywhere
Google has been systematically embedding Gemini across its hardware and software stack this month. Android Authority reported that the company repositioned the Ask Gemini prompt in Google Meet, moving it from a tucked-away icon to a visible panel in the bottom-left corner of the meeting window. In both cases the goal appears consistent: make AI less something users choose to activate and more something the product surfaces by default.
For Home specifically, this is a meaningful reframe of what a camera does. Historically, security cameras in smart home platforms provided live viewing and motion-triggered recording. The new approach treats the camera feed as a continuous context stream that can recognize objects and infer intent. Not motion, but meaning. A yoga mat signals one thing; a grocery bag signals another.
What Google has not specified publicly is where the computation happens. The announcement did not clarify whether footage is processed locally or sent to Google servers, or how long visual data is retained between trigger checks. For a feature built on the premise that cameras understand what they see, that is a notable gap in the disclosure.
Reading the competitive map
In competitive terms, camera-based automation puts Google in territory Amazon has approached with Alexa AI integrations, while Apple's HomeKit has remained largely passive on visual intelligence. The race to make home hardware an active artificial intelligence interface is accelerating across all three ecosystems.
The rollout is currently small by design: US only, English only, Public Preview only, Nest or certified hardware only, Premium Advanced subscription required. 9to5Mac noted this week that Anthropic is rolling out its latest Claude Opus 4.8 model with tiered performance settings, a common pattern as AI providers manage the compute costs of running capable models at scale. Google's path to general availability depends on how reliably Gemini's recognition holds up in the varied, unpredictable conditions of real homes.
Gemini for Home launched in early access in October 2025. A broader rollout is expected, with no date given.
Whether camera automations survive that expansion intact, and at what subscription tier, is the real measure of the feature's value. A yoga mat example makes for good launch copy. The harder question is what happens when the camera mistakes a rolled-up rain jacket for one.
FAQ
What cameras work with Gemini automations in Google Home?
Nest cameras are supported, along with select third-party cameras with "Gemini Built-In" certification. Google has not published a full list of compatible third-party devices.
How much does the required subscription cost?
Google Home Premium Advanced is $20 per month or $200 per year. The feature also requires enrollment in the Google Home Public Preview program.
Is the feature available outside the United States?
No. At launch it is restricted to the US and to English only, with no international timeline announced.
How does the natural language setup work?
Users describe the visual event they want to trigger a routine in plain English, then select which cameras should watch for it. AI descriptions must be enabled under "Gemini for Home camera features" in the app settings before the option appears.