Overview
Google’s Gemini AI is moving beyond answering questions to actually performing actions on a user’s behalf. The upcoming “Get tasks done with Gemini” feature—codenamed bonobo—will let users ask Gemini to book rides, order food, and more, all through screen‑automation on Android.
How It Works
When a user issues a command (e.g., “Book an Uber to the office”), Gemini launches the target app, fills in the required fields, and confirms the action. The process runs as a Labs feature in the latest Google app beta (v17.4) and requires Android 16 QPR3, which introduced the underlying screen‑automation framework.
Supported Apps & Limitations
- Initially limited to a small set of “certain apps” – likely first‑party Google services and popular third‑party apps such as Uber and Uber Eats.
- App UI changes can break automation, so Google is starting with a controlled list.
- Potential tier restriction: early access may be limited to Pro and Ultra users.
Privacy & Safety Measures
- All screenshots captured during automation are reviewed by trained reviewers and may be used to improve Google services if the “Keep Activity” setting is enabled.
- Users are warned never to enter payment details directly into Gemini chats.
- Google advises disabling screen automation during emergencies.
- Users retain full control – they can stop or take over the task at any moment.
Availability & Future Outlook
The feature is expected to launch in beta alongside Android 16 QPR3’s public release in March. As the AI matures, broader app support and deeper agentic capabilities are likely to follow.
What Users Should Do
Try the beta, share feedback, and stay tuned for official announcements. Remember to keep interactions respectful and follow community guidelines when discussing the feature.