Google's Gemini AI will soon control Android apps through screen automation, allowing users to order food and book rides without touching their phones. The feature, codenamed "bonobo," appeared in Google app beta version 17.4 this week.
Gemini's new "Get tasks done with Gemini" capability uses screen automation to interact with app interfaces directly.
Users can instruct the AI to perform tasks like ordering from Uber Eats or booking Uber rides, with Gemini launching apps and navigating menus autonomously.
The feature requires Android 16 QPR3 and will initially work with a limited number of apps. Google plans to release it as an experimental Labs feature first, with support expanding gradually to third-party applications.
Google warns that Gemini can make mistakes and users remain responsible for its actions. The company advises close supervision and allows manual takeover at any point during automated tasks.
Privacy protections include warnings against sharing login credentials or payment information with Gemini. When activity tracking is enabled, Google may review screenshots taken during automation to improve services.
Samsung Galaxy device users will likely receive the feature through Google's integration partnership, according to industry analysis. The collaboration ensures new Gemini capabilities typically extend to Samsung's Android ecosystem.
The screen automation builds on Android 16 QPR3 groundwork and follows Chrome's recent Auto Browse feature powered by Gemini 3.
Both developments represent Google's push toward agentic AI that requires minimal user intervention, similar to Google Maps' Gemini integration for community edits.
Manual smartphone interaction continues decreasing as AI handles routine workflows. Gemini's evolution from basic text tasks to app control marks a significant shift in mobile device operation patterns.
