agentic AI – Google unveiled new Gemini Intelligence features for Android, including agentic app-to-app tasks, auto web browsing, form filling, smarter Chrome, and vibe-coded widgets.
Google is pushing deeper into “agentic” AI on Android, and the changes unveiled around Gemini Intelligence at its Android Show: I/O Edition are aimed at making the assistant do more of the work for you, not just answer questions.
A core part of the update focuses on letting Gemini complete multistep tasks across multiple apps.. The idea is that you can press the phone’s power button and describe what you want to get done. while the assistant uses what’s currently on your screen as the context to guide the process.. Google also said Gemini will stop for your final confirmation before it completes checkout. which is designed to keep the user in control when actions affect purchases.
Google previously introduced agentic capabilities to Gemini at the Samsung Galaxy S26 launch. describing features like ordering food or booking a ride.. That earlier push also hinted at more complex flows. such as booking something specific like a front-row bike spot for a spin class. pulling a syllabus from Gmail. and then searching for related books.. The new Android update extends that direction by positioning Gemini as able to carry out a chain of actions spanning different apps.
The company’s example makes the mechanics clear: Gemini can take text from a notes app. then use that information to add items to a cart inside a shopping app.. In other words. the assistant isn’t just parsing what you ask. it’s coordinating steps where information has to move from one place to another. with your screen acting as a live reference point.
Web abilities are also expanding. Google referenced a feature first introduced in January that let Gemini browse the web and carry out tasks like booking an appointment, initially as an experimental rollout. The report stated that this auto-browse capability is now moving to Android as well.
On top of that, Google said Gemini is coming to Chrome on Android in late June.. The goal is similar to what Gemini in Chrome already supports on desktop: summarizing content or helping users ask questions about what’s shown on a web page.. For many people, that means fewer copy-and-paste loops and less switching between tabs to understand long articles or dense pages.
Form-filling is getting its own upgrade, too.. Google said Gemini will be able to fill out forms on your behalf after learning relevant details through Personal Intelligence.. This feature is opt-in. and Google added that you can disable it anytime through settings. addressing one of the biggest friction points for users when AI systems start handling personal data.
Speech input is also being refined at the keyboard level.. Gemini is coming to Android’s Gboard. and Google is using the assistant’s multimodal capabilities with a feature called Rambler.. The report described Rambler as similar to AI dictation tools found in other apps: you speak. it transcribes in your own tone. and then it formats the result by removing filler words.
Google also wants Android users to experience “vibe-coding” through widgets. a move that reflects how quickly natural-language creation has been spreading across consumer devices.. The company is introducing a way to build Android widgets by describing what you want in plain language—for example. generating a meal-planning widget by asking for “three high-protein meal prep recipes every week.”
While “vibe-coded” widgets are positioned as a new offering for Android. Google emphasized that widget creation via Gemini isn’t entirely new.. The report noted that Gemini had been able to help create widgets before. and it pointed to a similar tool released by the hardware startup Nothing last year. suggesting this style of interface is becoming a competitive feature rather than a novelty.
Design is part of the plan as well. Google said Gemini Intelligence features will follow its Material 3 expressive design language in the way they present and behave across the user experience.
Timing-wise. Google set expectations around device availability: the report stated that these AI-powered features will first land on the latest Samsung Galaxy and Google Pixel devices this summer. with broader rollout to other Android devices later this year.. For Android users. that means the next wave of features may arrive in stages. starting with flagship hardware before widening across the ecosystem.
For Google’s broader bet. the through-line is clear: the company appears to be moving Gemini from a “help me” assistant toward a “do it with me” agent that can operate across apps. interpret what’s on your screen. and carry tasks closer to completion—while still reserving moments for confirmation when it matters. like checkout.
Google Gemini Intelligence agentic AI Android widgets Gboard dictation Chrome AI summarization web browsing assistant