- New Feature: Google launched Gemini screen automation on the Galaxy S26, letting the AI assistant control apps like Uber Eats and Starbucks via voice commands.
- Usage Limits: Free users are capped at five daily requests, while Google AI Ultra subscribers at $249.99/month get 120.
- Early Constraints: The feature launches with only six supported apps in the US and South Korea, with English as the sole language.
Ordering lunch through Gemini now takes a voice command instead of a dozen taps – but Galaxy S26 owners will hit a wall fast. Google rolled out screen automation on March 12 with a free tier capped at just five requests per day, while the Google AI Ultra plan unlocks 120 daily requests at $249.99/month.
A four-tier daily limit structure governs how often users can call on the feature. Google AI Plus subscribers paying $7.99/month receive 12 requests per day, while Google AI Pro subscribers at $19.99/month get 20. According to Google’s official support page, these caps apply specifically to screen automation and are separate from other Gemini usage limits.
Tiered Usage Limits and Pricing
A 24x gap between free and Ultra positions screen automation as a feature non-paying users can sample but not rely on. A single morning routine of ordering coffee, booking a ride, and adding groceries to a cart would consume three of the five available free requests, leaving little room for additional use.
Mindy Brooks, VP of Product Management at Google, wrote that the company has been evolving Android from an operating system into “an intelligent system that learns and works for you.”
Google has previously metered Gemini capabilities by tier. Screen automation now adds another metered capability to that pricing framework, reinforcing a pattern where each new AI feature serves as a subscription upgrade incentive.
Taken together, the tier structure suggests Google is treating screen automation less as a standalone product and more as a conversion funnel, giving free users enough of a taste to demonstrate value while reserving practical daily use for paid subscribers.
How Screen Automation Works
Beyond the pricing model, the technical implementation reveals how Google designed the feature to operate. Under the hood, apps run in a virtual window on the phone, with cloud processing directing where to scroll, tap, and type. Supported apps at launch include Lyft, Uber, GrubHub, DoorDash, Uber Eats, and Starbucks, covering food, grocery, and rideshare categories.
Hands-on testing confirms the feature delivers on its promise. In one test, Gemini completed an Uber Eats order for food, building the cart without user intervention. In another, it navigated the Starbucks app to complete an order, selecting the size, customizing the drink, and adding it to the cart.
Despite these results, the system does not finalize checkout on its own. It sends a notification and hands control back to the user for final confirmation, a deliberate safety measure that prevents unintended purchases. Brooks noted that Gemini runs in the background during automation, leaving the phone “still yours to use.”
With only six supported apps and mandatory checkout confirmation, the initial rollout prioritizes safety over breadth. For the pricing tiers to justify their cost, the supported app catalog will need to expand well beyond food and rideshare.
Availability and Early Limitations
Screen automation is available on the Galaxy S26 in the US and South Korea, with Pixel 10 support not yet live. Running the feature requires One UI 8.5 with the February 2026 security patch. Only personal Google accounts are supported, with work and school accounts excluded and English as the sole language at launch.
Google has called the feature an early preview beta on select devices. During testing, a notable bug surfaced: in one instance, the automation locked the phone in a fullscreen view that required a forced reboot to exit. Screen automation is also different from the Gemini Agent capability, a cloud-based feature exclusive to AI Ultra subscribers at 200 requests per day.
By restricting the launch to two countries, six apps, and one language, Google is running what amounts to a controlled experiment before broader deployment. For users outside these boundaries, screen automation remains out of reach regardless of subscription tier.
Google has signaled expansion is coming. Instacart support is confirmed for a future update, and the AppFunctions framework is designed to open the system to additional third-party developers. Users can also set persistent default preferences – such as always choosing the cheapest Uber fare or only ordering gluten-free items – through Gemini’s Personal Intelligence settings. Whether five free daily requests prove sufficient to drive adoption or simply funnel users toward paid plans will depend on how quickly that app roster grows.
