At Google's "Android Show: I/O Edition," the company announced multiple new "Intelligence" features for Gemini, marking its transformation from a chat tool into an omnibus assistant capable of cross-app operations. Gemini now supports multi-step task chains: users can speak commands by pressing the power button, allowing AI to read screen context and execute tasks automatically—for example, extracting shopping lists from memos, switching to shopping apps to add items to carts, and completing payments only after user confirmation. The "Auto Browse" feature launched in January has also officially arrived on Android, enabling online searches and appointment scheduling on behalf of users. This summer, Chrome will integrate Gemini to support web page summaries and real-time Q&A. Additionally, Gemini will be integrated into the Gboard keyboard, introducing the "Rambler" voice input function that recognizes user tone, auto-removes filler words, and formats outputs accordingly. The form-filling capability relies on "Personal Intelligence" learning user data before generating entries automatically, requiring explicit user activation. More intriguingly, the "vibe-code" widget generation allows users to describe needs via natural language—such as requesting "three weekly meal prep recipes rich in protein"—enabling Gemini to create corresponding widgets in real time. These functions will debut this summer exclusively on Samsung Galaxy S26 devices and Pixel models, with other Android devices expected to follow later this year. All designs fully adopt the Material 3 Expressive visual language.