Gemini is about to start checking off your to-do list by actually using your apps

For a long time, the true AI Assistant dream has been a bit of a tease. We can ask Gemini to write an email or summarize a doc, but when it comes to actually doing something—like booking a ride or ordering dinner—we’ve still had to do the heavy lifting of opening apps and clicking through menus ourselves.

That is officially starting to change. Google just announced an early preview of a feature that allows Gemini to handle multi-step, real-world tasks directly on your Android phone.

Real Automation for Daily Tasks

Starting soon as a beta for the Pixel 10, Pixel 10 Pro, and newly-announced Samsung Galaxy S26 series, you’ll be able to long-press your power button and give Gemini a complex command like, “Book me a ride home” or “Reorder my last meal on DoorDash.”

Instead of just giving you a link or a suggestion, Gemini will actually work in the background to execute those steps for you. It’s starting with select apps in the food, grocery, and rideshare categories, and initially, it’ll be available for users in the U.S. and Korea.

How It works and stays private

I know what you’re likely thinking: Do I really want an AI clicking around inside my apps? Google seems to have anticipated the privacy concerns here with a pretty clever technical approach:

• Virtual Windows: Gemini doesn’t just get free rein of your phone. It runs the necessary app in a secure, virtual window. This limits the AI’s access to only the app it needs to complete the task, keeping the rest of your device off-limits.
• Live Transparency: You aren’t left wondering what the AI is doing. You can monitor the progress live via notifications. If something looks wrong, you can jump in or kill the task instantly.
• User Command Only: These automations only trigger when you specifically ask for them and stop the moment the task is finished.

Why This Matters

This is a massive leap forward for the Pixel 10 and the latest Galaxy flagships. It’s the first real look at “Agentic AI”—where the assistant becomes a doer rather than just a talker. Being able to offload those repetitive, “muscle memory” tasks while you keep using your phone for other things is exactly where the mobile experience needs to go.

It’s early days, and being a beta, I’m sure there will be some kinks to iron out. But the prospect of never having to manually navigate a food delivery app again? Sign me up.