Google will allow to create and manage multiple AI agents right in Search.

This may be powerful,

For ongoing tasks, like planning a wedding or managing a move, Search can code entire custom experiences with Google @Antigravity. Think: mini apps, dashboards or trackers you can come back to again and again.

Google claims to bring generative UI to everyone, free of charge, with Antigravity and the agentic coding capabilities of Gemini 3.5 Flash.

Google has announced Introducing Gemini Spark

Google announces Android Halo coming later this year.

Google announces Antigravity ’s agentic surfaces and features are now available.

Google claims Gemini 3.5 Flash as their strongest agentic and coding model yet. Generally available, starting today.

Text that says Gemini 3.5 Flash set against a deep blue background.

Gemini 3.5 models revealed.

Google announces Gemini Omni. Read more here.

Follow our Google I/O 2026 live blog for more breaking Gemini announcements, Android updates, Veo 4 reveals, and all major AI news directly from Google’s keynote.

Catch our full Google I/O 2026 coverage with interesting leaks, rumors and announcements by clicking here.

Just ahead of Google I/O 2026, a massive leak has revealed that Google is preparing a radically upgraded Gemini desktop app focused heavily on autonomous AI agents, local workflows, coding assistance, and multimodal computing.

The leaked details suggest Google is transforming Gemini from a simple chatbot into a full desktop AI operating layer capable of understanding apps, files, windows, codebases, and even user intent in real time.

Google I/O 2026 Live Blog: Gemini Announcements Expected Soon

Google is expected to officially unveil major Gemini upgrades during the Google I/O 2026 keynote, including deeper Gemini AI integrations across Android, ChromeOS, Workspace, and desktop environments.

Our live blog will track all major Google I/O AI announcements including Gemini Spark, Gemini Omni, Veo 4 integration, Android AI features, and Google’s next-generation AI agent ecosystem.

Gemini Desktop App Is Becoming a Full AI Agent Platform

According to the leak, Google’s new Gemini desktop app is no longer designed around simple conversations.

Instead, the app now appears to function as a true AI agent platform with:

  • Local system awareness
  • Workflow automation
  • File understanding
  • Coding assistance
  • Screen context analysis
  • Persistent AI overlays
  • Voice-powered multitasking

The leak suggests Google wants Gemini to become a universal AI assistant that actively interacts with your PC rather than simply responding to prompts.

This would position Gemini much closer to OpenAI’s operator-style agents and Microsoft’s Copilot vision for Windows.

Gemini App Split Into Chat and Spark Workspaces

One of the biggest changes revealed in the leak is a new dual-workspace interface.

The Gemini desktop app reportedly now includes:

  • Standard Chat mode
  • Dedicated Spark workspace

The normal Chat mode behaves similarly to the current Gemini experience focused on conversations, reasoning, and multimodal interactions.

Spark mode, however, appears to be the real breakthrough.

What Gemini Spark Can Do

The leaked Spark environment is reportedly built for local agentic workflows and advanced desktop tasks.

Gemini Spark may support:

  • Connecting to local folders
  • Reading and analyzing files
  • Understanding code projects
  • Running scripts locally
  • Organizing folders automatically
  • Managing workflows
  • Syncing directly with Google Drive

This could allow Gemini to function more like a proactive desktop assistant rather than a cloud chatbot.

Developers may be able to point Gemini Spark toward a coding project and let the AI analyze, modify, debug, and organize files autonomously.

“Stream to Cursor” Could Be Google’s Most Important AI Feature Yet

Another major leak references a feature called “Stream to Cursor.”

This appears similar to Google’s earlier “Magic Pointer” concepts where AI understands the context of whatever application, button, file, or interface element the user is hovering over.

If accurate, Stream to Cursor could allow Gemini to:

  • Understand apps in real time
  • Analyze UI elements dynamically
  • Read active windows instantly
  • Provide contextual help automatically
  • Trigger AI actions based on cursor location

Essentially, Gemini may gain the ability to visually interpret the desktop continuously.

This could become one of Google’s biggest competitive responses to Microsoft Recall, OpenAI Operator, and Apple’s rumored contextual AI system.

Floating Gemini Overlay Adds Real-Time Screen and Camera Sharing

The leak also reveals a floating Gemini overlay integrated directly into the desktop experience.

This overlay reportedly allows users to instantly:

  • Share full screen context
  • Share specific app windows
  • Share camera input
  • Switch between Gemini models rapidly
  • Launch voice interactions
  • Trigger AI workflows without leaving current apps

Google is reportedly allowing fast model switching between:

  • Gemini 3 Flash
  • Gemini 3.1 Pro
  • Other experimental models

That could make Gemini significantly more flexible for coding, research, creative work, and multitasking.

Google Is Preparing Local “Skills” Support

One of the most powerful leaked features is support for local “Skills.”

This system reportedly allows users to attach:

  • Custom scripts
  • Capability folders
  • Specialized automation tools
  • Workflow packages

directly into Gemini’s agent environment.

In practice, this could turn Gemini into a modular AI platform where users install additional capabilities much like plugins or automation packs.

For developers and power users, this may become one of the most important Gemini upgrades yet.

Gemini Omni May Actually Be “Veo4 Omni”

The leak also contains references to “Gemini Omni” internally being labeled as “Veo4 Omni.”

This strongly hints that Google may deeply integrate Veo 4 video generation capabilities directly inside Gemini’s desktop ecosystem.

If true, users may eventually be able to:

  • Generate videos locally
  • Edit video workflows with AI
  • Combine multimodal creation tools
  • Use Veo directly inside Gemini overlays

This would align with Google’s broader push toward fully multimodal AI experiences that combine text, voice, image, video, and real-time context.

Gemini Live Is Becoming a Persistent AI Voice Layer

Google also appears to be working on a persistent Gemini Live overlay.

While still reportedly unfinished internally, the feature may provide:

  • Continuous voice interaction
  • Always-available AI assistance
  • Hands-free workflow management
  • Real-time contextual conversations

This would move Gemini closer to becoming a true operating-system-level assistant rather than a standalone application.

Google Is Clearly Building an AI Operating System

Taken together, these leaks suggest Google’s ambitions are much larger than simply competing with ChatGPT.

The company appears to be building:

  • An AI-native desktop layer
  • Autonomous local agents
  • Context-aware computing
  • Multimodal desktop intelligence
  • Persistent AI workflow automation

The leaked Gemini desktop app sounds less like a chatbot and more like an entirely new AI computing platform.

Google I/O 2026 Could Become Google’s Biggest AI Event Yet

Google I/O 2026 is shaping up to be one of Google’s most important AI-focused events ever.

Expected announcements include:

  • Gemini Spark
  • Gemini Omni
  • Veo 4 integration
  • Android AI upgrades
  • Chrome AI features
  • Agentic AI workflows
  • Gemini Live improvements
  • Advanced desktop integrations

If these leaks prove accurate, Google may finally be preparing its strongest answer yet to Microsoft Copilot, OpenAI Operator, and Apple Intelligence.

And this time, Gemini may not just assist users — it may actively work alongside them across the entire desktop experience.

Follow our Google I/O 2026 live blog for more breaking Gemini announcements, Android updates, Veo 4 reveals, and all major AI news directly from Google’s keynote.

Interested in reading more about Google Gemini news. Read our full Google Gemini coverage by clicking here.

Please follow us on our Facebook page and X account for all latest and breaking Google, Android and Nokia related news.

Add NPowerUser (https://nokiapoweruser.com) as a preferred source on Google News
Add NPowerUser as a preferred source on Google News

RELATED ARTICLESMORE FROM AUTHOR