What if you could ask your phone to book a ride or repeat your last order while you continued using the device? Google is preparing a Gemini feature capable of carrying out multi-step daily tasks directly on Android.
The tool will soon arrive as a beta feature for Gemini on the Pixel 10, Pixel 10 Pro, and Samsung Galaxy S26 series. Its initial availability will be limited to the United States and South Korea.
Gemini can handle repetitive tasks
The idea is simple: press and hold the power button and ask Gemini to do something for you. For example, it could help you book a ride home or reorder your last meal on DoorDash.
Unlike a conventional response, Gemini won’t just tell you how to do it. The assistant will work in the background, following the necessary steps inside the relevant app while you continue using your phone.
Does that sound like automation with too much power? Google says the feature was designed with specific limits to keep control in the user’s hands.
Security, control, and real-time tracking
Automation begins only after a direct instruction. It also stops when the task is complete instead of continuing to take actions on its own.
You’ll also be able to monitor Gemini’s progress through live notifications. From there, you can review what it’s doing, intervene in the process, or stop it if something doesn’t seem right.
The goal isn’t to give Gemini unlimited access, but to let it complete specific tasks within a controlled environment.
To carry out each action, Gemini will open the necessary app in a secure virtual window inside the phone. This limits the assistant’s access to certain apps and prevents it from freely reviewing the rest of the device.
An initial version with limited scope
During the beta, the feature will be available only for select apps in the food, grocery shopping, and ride-hailing categories. Google has not yet provided a complete list of supported services.
This detail matters. This is not yet an assistant capable of controlling every app on your phone, but an initial test to see how multi-step tasks work in everyday situations.
Imagine, for example, that you’re leaving work and need to order a ride and some food. Instead of jumping between apps, filling out forms, and repeating instructions, you could delegate part of that process to Gemini and monitor it through notifications.
The feature also raises practical questions: What happens if the price changes? If the restaurant doesn’t accept orders? Or if the saved address is no longer correct? That’s precisely why the ability to review and stop the task will be just as important as the automation itself.
Gemini is beginning to move from providing informational answers to taking concrete actions on your phone. The challenge won’t simply be doing more, but proving that it can do those things predictably, transparently, and under the user’s supervision.
Original source
https://blog.google/innovation-and-ai/products/gemini-app/android-multi-step-tasks