Google is doubling down on agentic AI to improve Android users’ on-device experiences by embedding Gemini deeper into the products that people use every day.
The tech giant on Tuesday, May 12, introduced Gemini Intelligence, an AI agent-based operating layer designed to read a user’s screen, move across apps, and autonomously complete tasks that would otherwise require users to jump between multiple platforms.
The announcement was made at The Android Show 2026, a precursor to , where the company previewed the latest Android version, including AI-powered app automation, smarter version of Chrome on Android, new tools for creators, new security features, and a revamped Android Auto experience.
With Gemini Intelligence, Google is moving beyond traditional chatbot-style assistants toward autonomous AI agents. While AI agents are not new, their adoption is still in the nascent stages. Gemini Intelligence could potentially mark a turning point for the mainstream adoption of AI agents, especially in India where over 97 per cent of all smartphones are powered by Android.
Gemini Intelligence’s debut comes a month before Apple is expected to unveil a revamped Apple Intelligence, reportedly powered by Gemini itself, at WWDC 2026. It highlights Google’s massive advantage in consumer AI owing to its vast distribution network. However, rivals such as OpenAI are also reportedly eyeing smartphones with AI agents at the core.
“We’re transforming Android from an operating system into an intelligence system. This means deeply understanding the context, anticipating your needs, and getting things done on your behalf,” Sameer Samat, the president of the Android Ecosystem at Google, said during the keynote presentation on Tuesday.
At the launch, Google revealed little about Gemini Intelligence under the hood, including how it was developed or which large language models (LLMs) power its agentic AI capabilities. However, the tech giant mentioned that it spent months fine-tuning the multi-step task automation capabilities of Gemini Intelligence using popular food and rideshare apps on Galaxy S26 and Pixel 10 devices.
Task automation allows Gemini to use certain apps on a user’s behalf. For instance, users can ask Gemini to look at the guest list for a barbecue, build a menu, add ingredients to an Instacart list, and return for approval before checkout. Another example is asking Gemini to find a user’s class syllabus in Gmail then putting the books in their cart.
Google said task automation will be available for a wider range of apps “soon”. It will be rolled out in phases, starting with the latest Samsung Galaxy and Google Pixel phones this summer, before expanding across more Android devices, including watches, cars, glasses and laptops later this year.
Another addition to Gemini under Gemini Intelligence is multimodality. Instead of manually switching between apps and copy pasting text, users can, for instance, long press the power button on the grocery list saved on the Notes app, and Gemini will build a shopping cart with all of the items for delivery. Previously, Gemini could only use voice or text prompts to take actions.
Custom widgets: Users can create custom widgets just by describing what they want to see on their screen. For instance, users can ask Gemini to generate a home screen dashboard comprising weekly high-protein recipes for prepping meals, and cyclists can create a weather widget that only tracks wind speed and rain.
Autofill: Gemini Intelligence will pull relevant information from connected apps and automatically fill out tiny text fields in forms across Android apps, including Google Chrome. Google emphasised that this feature is opt-in and can be turned on or off in settings.
Rambler: This feature converts speech into text messages while removing pauses and filler words like ‘uhms’ and ‘ahs’ to make messages more concise and readable. Rambler will clearly show users when it is enabled, and the audio is only used to transcribe in real-time and is not stored or saved, as per Google. It also comes with support for Hindi and other languages.
Besides these features, Gemini Intelligence comes with an updated design language that builds upon Material 3 Expressive released last year. To address security concerns around AI agents going rogue, Google said that Gemini will always come back to the user before completing a transaction.
Google is betting on Gemini to cement its lead in consumer AI, given its control over platforms and products that people use every day, which allows it to embed its AI tools across the ecosystem at scale.
Take Android Auto. On Tuesday, Google announced a redesign of its in-car OS around Gemini, turning cars into another key surface for its AI assistant. Currently, Android Auto is in more than 250 million cars. Google said that the new version comes with its biggest maps update in a decade and Gemini-powered help with tasks like ordering dinner while driving. Among the first automakers to get a few of the new Android Auto features include BMW, Ford, Hyundai, Kia, Genesis, Mahindra, Mercedes-Benz, Renault, Škoda, Tata, and Volvo.
Google also said it is bringing an AI browsing agent to its popular Chrome web browser on all Android devices. The AI browsing agent can be used to research, summarise, and compare content across web pages as well as auto browse and autonomously complete tasks such as appointment bookings.
Meanwhile, Gemini Intelligence could add pressure on Apple to prove at WWDC 2026 (taking place from June 8 to June 12) that its own version of personal AI on the phone is further along. Notably, Google is part of the iPhone-maker’s AI reset. Gemini models will power Siri and a few other Apple Intelligence features as part of the multi-year deal between the two tech giants.
Apple’s iPhone is built around applications developed by third parties, and this model has played a huge role in making the iPhone what it is today. But it is facing potential disruption not just by Google but also by smaller players such as OpenAI. The Sam Altman-led startup is working on a mysterious new phone that would heavily rely on AI agents, according to a last month.
