Google wants Gemini to work while you sleep: Check biggest new features announced at I/O 2026

Google took the wraps off some of its biggest AI updates at I/O 2026, including the Gemini 3.5 Flash model, Gemini Omni, Gemini Spark, and the new Antigravity platform. But perhaps the biggest change the company showcased this year was transforming Gemini from a chatbot experience into an agentic one, not just with Gemini Spark but also by adding dedicated AI agents inside the app.

Here's everything Google announced for Gemini this year.

1) Gemini Spark:

Google announced a new 24/7 personal AI agent named Gemini Spark, designed to work in the background and actively complete tasks on behalf of users.

The new platform is powered by the Gemini 3.5 Flash model, which Google also unveiled at I/O 2026, while taking advantage of Antigravity 2.0. Spark can continue working even when users lock their phones or close their laptops. The agentic AI tool is also deeply integrated with Google Workspace apps like Gmail, Docs, and Slides, allowing it to take more complex actions across a user's digital life.

With Gemini Spark, users can:

Set recurring triggers: Instruct the agent to automatically parse monthly credit card statements and flag hidden subscription fees.

Teach custom skills: Have Spark monitor inboxes for emails from a child's school, extract critical deadlines, and compile a daily digest.

Execute complete workflows: Users can ask the agent to synthesise raw meeting notes from emails and chats, draft a polished Google Doc, and write a companion email to kick off a project.

Google says Spark requires explicit permission from users before making high-stakes decisions like spending money or sending emails.

2) Daily Brief:

Google is also introducing a new AI agent inside the Gemini app called "Daily Brief". The agent is designed to give users a "personalised morning digest that's designed to be your first stop every day."

The new agent is an opt-in experience, meaning users will have to explicitly enable it before using it. The feature connects with various apps in the background and compiles information into a skimmable, prioritised briefing based on users' goals while also suggesting immediate next steps.

3) Gemini app redesigned:

Google has also redesigned the Gemini interface across both the app and web versions of the platform. The new design, called "Neural Expressive", brings fluid animations, vibrant colours, refreshed typography, and haptic feedback.

Google has also baked Gemini Live more directly into the Gemini app to make switching from typing to voice interactions easier. Meanwhile, the company showcased regional dialect support, including some Indian languages, which will soon be available inside Gemini Live.

4) Gemini Omni:

Google is also adding its latest "world model", Gemini Omni, to the Gemini app for AI video generation and editing. The company says Omni can take combinations of text, images, and videos as inputs and transform them into cinematic-quality videos.

Users can upload footage directly from their camera roll and use simple text prompts to apply cinematic zooms, swap backgrounds, or use built-in templates with a single tap. They can even create custom AI avatars of themselves to place inside generated videos.

The first version in this series, Gemini Omni Flash, is rolling out now to Google AI Plus, Pro, and Ultra subscribers.

5) Gemini app for macOS:

Google recently rolled out its Gemini app for macOS and at I/O 2026, the company announced that the app will get Gemini Spark support later this summer. This would allow the AI assistant to manage local files and automate workflows natively on users' machines.