Google has unveiled Gemini Intelligence, a new system designed to automate tedious tasks on Android phones. Announced during the Android Show: I/O Edition, the system functions as a computer use agent capable of navigating and completing multi-step tasks across popular apps. The company spent five months fine-tuning the technology.

The agent can read context from emails, images, and your phone's screen. Google demonstrated it reading a class syllabus in Gmail and automatically adding required books to a shopping cart. Users might also point the camera at a travel brochure and ask Gemini to find similar tours on Expedia.

Privacy and user control are baked in. Gemini Intelligence won't start any task until explicitly instructed, and any purchase requires manual confirmation. A progress bar lets users stop the agent at any time, while familiar permissions menus control data access.

The feature will first arrive on recent Pixel and Samsung Galaxy phones. Whether people embrace it remains an open question — most apps are already easy to use, and if Gemini makes mistakes like other computer use agents, users likely won't adopt it beyond initial experimentation.