Google appears to be developing a new Gemini feature that could allow the AI assistant to perform tasks directly within apps on Android smartphones. New evidence found in the latest Google app beta suggests that Gemini may soon gain screen automation abilities enabling it to carry out actions such as placing online orders or booking rides on behalf of users.
The upcoming capability is part of Google’s push to enhance Gemini’s agentic features allowing the AI to move beyond conversations and take practical actions. Code strings discovered during an APK teardown of the Google app beta reference a Labs feature described as “Get tasks done with Gemini.” This feature is said to use screen automation in selected apps meaning it will initially work only with certain supported applications.
According to the findings the feature carries an internal codename “bonobo” and is expected to require Android 16 QPR3 to function. Google plans to clearly inform users that Gemini can make mistakes and that users remain responsible for actions carried out by the AI. Users will also have the option to pause the automation at any point and manually complete a task themselves.
Privacy and security appear to be key considerations for Google. Users will be warned not to share payment details within Gemini chats and advised to avoid using screen automation for emergencies or sensitive activities. The company also notes that screenshots may be reviewed by trained staff to improve services when activity tracking is enabled.
Google has been steadily expanding Gemini’s capabilities across its ecosystem. Recent updates brought deeper Gemini integration into Google Maps enabling hands free navigation support for walkers and cyclists along with limited task completion features such as sending messages. The upcoming screen automation feature could mark a significant step toward making Gemini a more proactive digital assistant on Android devices.