Google Plans Gemini Intelligence Across Android Apps

Update, 11 November 2026: task automation reaches India

Six months after this announcement, Gemini Intelligence is on sale rather than planned, and its app-automation feature now lists India as a supported country. In a post on 10 November, Google said rollout began this summer on the Pixel 11 and on Samsung’s Galaxy Z Fold8 Ultra, Z Fold8, Z Flip8 and S26 series, that Pixel 10, Pixel 10 Pro XL and Pixel 10 Pro Fold get the latest features through that day’s Pixel Drop, and that task automation “is expanding to the UK, India, and Japan”.

Google’s help page for the feature sets out who can use it in India today:

The page does not name which Indian apps are covered. It describes the tasks as booking rides, flights and stays, buying tickets, buying and selling products, and ordering food and groceries, and adds that the travel, ticket and buy-or-sell examples start with the Galaxy Z Flip 8 and Z Fold 8.

Gemini Intelligence itself is defined by hardware. Google’s Android help page lists 12GB or more of RAM, a qualified chip and on-device AI processing among the requirements, and says individual features “may vary by country and language”. So owning a recent flagship does not guarantee every feature in India.

Who confirms the order or the payment

The help page is blunt about responsibility: “You’re responsible for Gemini’s actions during a task, including mistakes and unexpected results like purchases.” It tells users not to type sign-in details, payment details or other sensitive information into the chat, and to take control of the task and enter them in the app instead; for an Indian user paying through a UPI app, that includes the UPI PIN. Gemini is designed to pause and hand over for passwords, payment details and finalising a transaction, and to ask before sending messages or changing a delivery address.

One limit matters for the phones most likely to be in Indian hands: for now, a task cannot be taken over on Pixel 10 or Galaxy S26 devices, only stopped, from the chat box or the live notification. The page also warns about prompt injection, where hidden instructions in an app, website or email could push the agent to place an order or change an address without confirmation. Any payment in the final step still goes through the authentication your bank or UPI app requires; our guide to RBI 2FA rules explains which payments need two factors.

Per-app permission sits in the Gemini app under Settings, then Screen automation, where each app can be set to Always allow, Ask every time or Do not allow.

What else the November post added

Rambler, the voice-to-text cleanup tool described below, is now available on the Pixel 10 series and the new motorola signature 27, through Gboard in select countries and languages. Create My Widget, which builds a home-screen widget from a description, is rolling out over the next few weeks. Device Assistance, which lets you ask Gemini to free up space or check which apps can read your location, needs Android 17 or later on select devices. The motorola signature 27 opens for pre-order in the US on 19 November; Google gave no Indian date.

Last checked on 11 November 2026 against Google’s 10 November post and its Gemini and Android help pages. The report of 13 May follows unchanged below.

The original report, 13 May 2026

Google plans to bring a wider set of tasks powered by AI to Android through Gemini Intelligence, announced on 12 May. The features are intended to let supported phones carry work across applications, fill more complex forms and turn spoken notes into edited messages.

The company says the rollout will begin in waves on its latest Pixel phones and Samsung Galaxy devices during the summer. Other device categories are expected to follow later in the year. This is an announced rollout plan, so a user should not assume that the full set is already available on their Android phone.

Moving information between applications

Google’s examples include using a list visible on the screen to prepare a shopping cart, or finding information in Gmail and using it in another task. Progress would appear in notifications while the work runs in the background.

The distinction for a user is between having software explain a task and allowing it to carry out the steps. The latter requires a clear requested outcome and a review of what will be submitted. Google says the automation acts on the user’s command, stops when the task is complete and leaves the final confirmation to the user.

The announcement also describes Gemini in Chrome for Android, with a planned start in late June, and browser automation for activities such as bookings. Those dates describe Google’s intended availability at the time of the announcement, not a completed deployment.

Autofill and dictation draw on different information

Connecting Gemini to Autofill with Google is described as optional. The proposed connection would use relevant information from connected applications to complete fields, with the ability to turn it on or off in settings.

That makes the selection of connected information consequential. A form might need a delivery address without needing every detail in a person’s messages. Before submitting, the user still needs to check the actual values and the destination receiving them.

Google also introduces Rambler, which takes spoken thoughts and produces a more concise written message. The company says it can handle language switching, including English and Hindi, and that audio is used for live transcription without being saved. These are product claims in the launch announcement, rather than results from testing by this publication.

Editing speech creates a separate review question: whether the resulting message preserves the speaker’s meaning, particularly names, numbers and qualifications. Convenience depends on that accuracy as well as speed.

The broader workplace version of this shift is visible in Copilot Cowork’s early-access workflows. In both cases, the useful unit to assess is a complete, bounded task and the opportunities to review it.