Google Wants Gemini to Order Your DoorDash — What Could Possibly Go Wrong?

Smartphone with holographic task bubbles showing food delivery and rideshare automation

Google just announced that Gemini on Android will soon be able to handle multi-step tasks for you — ordering food, booking rides, and getting groceries — all while you keep scrolling. Launching as a beta on Pixel 10 and Samsung Galaxy S26 devices in the U.S. and Korea. Sounds magical. Let's look closer.

What Google Announced

The new feature lets you long-press the power button and ask Gemini to do things like "book me a ride home" or "reorder my last DoorDash meal." Gemini then runs the relevant app in a secure virtual window on your phone, automating the clicks and taps needed to complete the task while you continue using your phone normally.

Google emphasizes three safety pillars: you control when it starts and stops, you can monitor progress via notifications, and Gemini only accesses the specific app it needs — not your entire device. The beta launches with "select apps in the food, grocery and rideshare categories."

The Skeptic's Take

Let's be honest about what this actually is: Google is automating button presses inside third-party apps. It's screen scraping with extra steps and a marketing budget. The fact that they felt the need to explicitly state "you can stop the task" and "it can only access limited apps" tells you everything about the trust gap they're trying to bridge.

There's also the elephant in the room: this only works with "select apps" at launch. So Gemini can order your Chipotle but can't book your dentist appointment, pay your electricity bill, or do any of the genuinely tedious multi-step tasks that would actually change your life. It's solving the easiest version of a hard problem and calling it a revolution.

The Real Question

The fundamental issue isn't whether Gemini can tap buttons inside DoorDash — it's whether you trust an AI to spend your money. Every "multi-step task" here involves a financial transaction. Wrong address on a ride? Wrong size on a grocery order? The stakes aren't trivial, and the error recovery story is conspicuously absent from Google's announcement.

There's also the competitive angle. Apple hasn't shipped anything like this yet, which either means Google is ahead of the curve or rushing to market with something that isn't ready. Given Google's track record with assistant features (RIP Google Assistant routines, Duplex, and a dozen other abandoned experiments), skepticism seems warranted.

The Bottom Line

The concept is genuinely exciting — a phone AI that can actually do things, not just answer questions. But limiting it to food delivery and rideshare apps at launch, on only the newest phones, in only two countries, feels less like a paradigm shift and more like a carefully controlled demo. Wake us up when Gemini can handle the DMV.