OMNI¶

Speak anywhere. Act everywhere.¶

One AI brain. Every device. Infinite capabilities.

What is Omni?¶

Omni is a multi-client AI agent hub that lets you speak to one intelligent agent from any device — web dashboard, mobile, Chrome extension, desktop, or smart glasses — and have it act across all of them simultaneously.

Built for the Gemini Live Agent Challenge hackathon, Omni demonstrates how a single AI brain powered by Google Gemini Live API and Google ADK can orchestrate actions across every device you own.

One Voice, Every Device¶

Web, mobile, Chrome extension, desktop tray app, ESP32 glasses — speak from anywhere.

MCP Plugin Store¶

Install new agent capabilities in one click, like an app store for AI skills.

GenUI¶

Agent renders live charts, tables, code blocks, and cards on your dashboard while speaking.

Agent Personas¶

Switch between specialized AI personalities — analyst, coder, researcher — with distinct voices and skills.

Cloud Desktop¶

Full Linux desktop in the cloud — agents can launch apps, write code, run scripts.

Cross-Client Actions¶

Say "save this to my dashboard" from your phone — it appears on your desktop instantly.

Architecture at a Glance¶

graph TB
    subgraph Clients
        A[Web Dashboard]
        B[Desktop Client]
        C[Chrome Extension]
        D[Smart Glasses]
        E[CLI]
    end

    subgraph Backend
        F[FastAPI + WebSocket]
        G[Google ADK Agents]
        H[MCP Plugin Registry]
        I[E2B Sandbox]
    end

    subgraph Google Cloud
        J[Gemini Live API]
        K[Vertex AI]
        L[Firebase Auth]
        M[Firestore]
        N[Cloud Run]
    end

    A & B & C & D & E --> F
    F --> G --> J & K
    F --> H
    F --> I
    F --> L & M
    F -.-> N

Quick Links¶

Topic	Description
Installation	Set up the backend, dashboard, and desktop client
Architecture	Understand the system design
Plugin Development	Build custom MCP plugins
API Reference	REST & WebSocket API docs
Deployment	Deploy to GCP with Terraform
Contributing	How to write and contribute docs