Gemini | I/O 2026 Keynote

**TL;DR:** At I/O 2026, Google announced a complete redesign of the Gemini app (neural expressive), the new Gemini Omni multimodal creation model, and proactive agent features like Daily brief and Gemini Spark, along with voice-driven multi-document processing coming to the macOS version. ## Redesigned Gemini Experience: Neural Expressive Over the past summer, Gemini's monthly active users surpassed 900 million, spanning 230+ countries and regions and supporting 70+ languages. The first major update at this I/O is a **ground-up complete redesign of the Gemini app**, using a design language called "neural expressive" featuring fluid animations, vibrant colors, new typography, and haptic feedback. Design isn't just about looks — it's about how you use it. The new version makes it easier to generate images, videos, and music, with built-in templates for easy remixing. The Gemini Live experience has also been overhauled: it can now be opened inline instantly, and more regional dialect options (like Liverpool accent, different variants of Portuguese) will roll out in the coming weeks. The most fundamental change is how model responses are presented: instead of showing blocks of text, Gemini now arranges responses for you in real-time (similar to generative UI in Search). As you scroll, you might see interactive images custom-generated by Gemini, a quick-scannable timeline, or embedded videos. This experience is rolling out globally starting today on Android, iOS, and Web. ## Gemini Omni: Multimodal Creation Enters a New Phase The second update is **Gemini Omni**, available starting today for paid subscribers (Google AI Plus, Pro, Ultra), directly integrated into the Gemini app. It supports any combination of text, images, and video input to turn ideas into reality. The presenter showed the example of Sasha: she uploads raw video, adds reference visual assets, can change the video style, and even switch the camera angle to a 360-degree shot. Gemini understands the physics of motion and intelligently overlays effects onto the real world while preserving the soul of the shot. The presenter called it the "Nano banana moment for video" — creating, remixing, and editing video has never been this easy. ## Agents: From Answering Questions to Working Proactively for You The third update focuses on how agents integrate into Gemini. With the model upgrade to Gemini 3.5 Flash (3.5 Pro coming soon), agents can not only answer questions but also work proactively for you. Two out-of-the-box agents were announced: ### Daily Brief A personalized summary designed to be the user's first stop every morning. It integrates inbox, calendar, tasks, and more to surface the most important items (e.g., remind you to return a book), organize by topic, and suggest next steps. Users can take follow-up actions inline. Rolling out today in the US to Google AI Plus, Pro, and Ultra subscribers. ### Gemini Spark (Custom Workflows) Power users can create their own workflows using Gemini Spark. It syncs across devices, and certain tasks require user confirmation — it won't execute unwanted actions. For example, create a "school year planning checklist" document leveraging all of Google Docs' formatting features, showing each child's list, dates, activities, and color coding. Spark will launch for trusted testers this week, and as a Beta in the US next week for Google AI Ultra subscribers. It will also come to Gemini Workspace and Gemini Enterprise. The future roadmap includes MCP integration, allowing Spark to handle tasks more proactively — for instance, if it detects the user is responsible for snacks at their child's T-ball game on Saturday, it can proactively set up an Instacart order and remember to select nut-free snacks. Multiple partners will integrate with Spark throughout the summer. ## macOS App: Voice-Driven Multi-Document Processing As a bonus, the team demonstrated new capabilities of the macOS version of Gemini. The native app launched last month (built from scratch using anti-gravity technology, delivering over 100 features in under 100 days) now supports: selecting multiple documents in Finder (PDFs, invoice images, etc.), long-pressing the function key, and using voice commands to have Gemini process them. For example, the user says "email to the shelter with the dog's details, allergies, vaccinations, organized into a table, friendly tone" — Gemini simultaneously understands the voice command and the selected files, automatically corrects speech errors (e.g., changing Thursday to Friday), and generates a table inline. These new voice capabilities, along with Gemini Spark, will come to the Mac app this summer. ## Summary The three core updates announced today (neural expressive design, Gemini Omni, agents) plus the macOS surprise bring Gemini closer to the vision of a "universal assistant" — personalized, proactive, powerful, and suitable for students, busy parents, and small business owners. **Source:** Gemini | I/O 2026 Keynote – YouTube (https://www.youtube.com/watch?v=7Z5Vy9JBANs)