Roadmap

Where Genesis OS
is headed

Six phases from a working Docker MVP to a full AI-native operating system that runs on your hardware, your terms, everywhere.

Current Status

Phases 0 – 3 Complete

Docker shell, persistent memory, voice, and agentic AI are all live.

✓ Phases 0–3 Done
Phases 0–3 — Docker MVP · Memory · Voice · Agentic Tools100%
OS desktop shell (glassmorphism)
AI chat with streaming
File Manager, Terminal, Editor, PDF, Browser
Docker Compose single-command boot
ChromaDB persistent vector memory
File watcher + auto-indexing
Voice — Whisper STT + Kokoro TTS
Agentic tool registry + reasoning loop
Approval system for destructive actions
Ollama auto-pull on first boot

Six phases to
the complete OS

Phase 0
Docker MVP — Ship a Working OS
● Active

A fully working AI operating system accessible via docker compose up. World-class OS desktop UI, AI chat with streaming, 7 built-in apps, and a model that runs on CPU with no GPU required.

🖥️ OS Desktop Shell 💬 Streaming AI Chat 📁 File Manager ⬛ Terminal PTY 📄 PDF + Office Viewer 📝 Monaco Editor 🌐 AI Browser 🐳 Docker Compose
Phase 1
Enhanced Memory & Persistent Identity
✓ Done

Deep memory — not just conversation history. Genesis will watch your filesystem, automatically index documents, and build a rich long-term understanding of your projects, preferences, and identity. The ChromaDB sidecar serves as the backbone.

🧠 ChromaDB semantic memory 📂 chokidar file watcher 🗜️ Context compression 👤 Identity profile 🔍 Semantic file search 📚 Auto-indexing
Phase 2
Voice — Speak Naturally, Hear Genesis
✓ Done

The full voice stack: faster-whisper for transcription, Piper TTS for natural speech synthesis, and openWakeWord for "Hey Genesis" hotword detection. Speak your commands. Hear your answers. All processed locally.

🎤 faster-whisper STT 🔊 Piper TTS 👂 "Hey Genesis" wake word 🌊 Web Audio API 🔇 Local VAD 📸 Screenshot → multimodal
Phase 3
Agentic Tools — Act on the World
✓ Done

The full agent tool registry goes live. Genesis gains the ability to autonomously browse the web, write and edit files, execute terminal commands, send emails, and schedule tasks — all with the approval system keeping humans in control.

🔧 Tool registry + plugins 🌐 search_web (local SearxNG) 📧 send_email (local SMTP) 📅 node-cron scheduling 🔁 Agentic reasoning loop 🛡️ Approval system
Phase 4
Multi-Device Deployment
○ Planned

Genesis expands beyond Docker. A polished Electron desktop app, a Debian live-build bootable OS image you can write to a USB drive, a WebGPU browser demo using Transformers.js, and LAN sync between devices.

🖥️ Electron desktop app 💿 Bootable Debian OS image 🌐 WebGPU browser demo 📱 PWA (mobile) 🔄 LAN device sync 🔐 Caddy + Authelia
Phase 5
Robotics & Physical World
○ Planned

Genesis leaves the screen entirely. A Pi-optimised OS image, GPIO control tools, picamera2 integration with Gemma 4's multimodal vision, and ROS2/pigpio bridges for robotics. Your AI now has hands.

🍓 Raspberry Pi image 🔌 GPIO tools 📷 picamera2 vision 🤖 ROS2 bridge 💡 pigpio automation 🌡️ Sensor integration

What we're building
toward

🌐

Universal Presence

Genesis OS runs consistently across desktop, laptop, phone, Raspberry Pi, home server, and browser — with memory that syncs privately across all of them over your local network.

🔌

Open Plugin Ecosystem

Any developer can publish Genesis plugins — new tools, new UI apps, new voice commands. The plugin API is intentionally identical to built-in tools, giving community plugins first-class status.

🧩

Model Marketplace

One-click model switching — swap between Gemma, Qwen, Mistral, LLaMA, phi-4, and future models without touching config files. Specialised models for code, vision, and voice.

🤝

Multi-Agent Collaboration

Multiple Genesis instances on a LAN can collaborate on tasks — one handles document management, another monitors servers, another handles communications. All memory private, all coordination local.

👁️

Ambient Awareness

Genesis watches your filesystem, calendar, email (locally), and sensor data to proactively surface information — not just respond to prompts. A truly ambient presence.

🌱

Always Free & Open

Genesis OS will never have a SaaS tier, premium features, or cloud lock-in. MIT license. Forever. The AI revolution should be universally accessible, not gated behind subscriptions.

Build Genesis OS
with us

Genesis OS is in active development. Every contribution moves us closer to the complete vision.

🐛

Report Bugs

Found something broken? Open an issue on GitHub with reproduction steps and we'll fix it fast.

Open an Issue →

Request Features

Have an idea that would make Genesis OS better? Start a discussion on GitHub and we'll consider it for the roadmap.

Start a Discussion →
🔧

Submit PRs

New tools, UI improvements, bug fixes, documentation, new app integrations — all PRs welcome. Fork, branch, and send it.

Fork the Repo →

Building a plugin takes minutes

packages/tools/plugins/my-tool.js
// Every Genesis OS plugin exports this exact shape
module.exports = {
  name: 'weather',
  description: 'Get current weather for a location (local weather API)',
  requiresApproval: false,
  schema: {
    type: 'object',
    properties: {
      location: { type: 'string', description: 'City name or postcode' }
    },
    required: ['location']
  },
  async execute({ location }, context) {
    // Your logic here — context has { projectRoot, userId, bus }
    const data = await fetch(`http://localhost:7703/weather?q=${location}`);
    return await data.json();
  }
};

Phases 0–3 complete. Phase 4 is next.

Docker shell, memory, voice, and agentic AI are all live. Now expanding to native Linux, Electron, and multi-device.