orbitos-ai

🚀 OrbitOS AI

Your AI Chief of Staff for Modern Teams.

OrbitOS AI unifies meetings, tasks, documents, and workflows into one workspace, and proactively summarizes, prioritizes, and automates work using AI. Built for the OpenAI Build Week 2026.

📖 Short Description

OrbitOS AI is an intelligent workspace designed to eliminate busywork. By leveraging large language models, it connects your meetings directly into manageable, actionable engineering tasks and workflow items. Make your team more productive by turning unstructured discussions into trackable executions instantly.

🏗 Why OrbitOS AI?

OrbitOS AI is not just a meeting summarizer.

It acts as an AI Project Manager that transforms discussions into structured execution plans. It decomposes chaotic meetings into granular engineering tasks, assigns employees, flags risks, and extracts essential decisions autonomously.

📸 Screenshots

(Placeholder for dashboard screenshot)

Dashboard Screenshot

(Placeholder for meeting assistant output)

Meeting Assistant Screenshot

🎥 Demo

🎥 Watch Demo Video

✨ Key Features

All the following capabilities are actively implemented and functioning:

🏗 Architecture

The platform uses a decoupled frontend React/Vite application talking to a Node/Express backend. All integrations point to a Supabase PostgreSQL instance featuring Row Level Security and managed Auth.

Security principles embedded in the architecture:

💻 Tech Stack

🔄 AI Workflow

The AI engine extracts the payload from active meetings. It sequentially passes the transcript to LLMs to generate summaries, action items, risks, and decisions. Users can selectively assign AI-generated rich action items to specific team members before seamlessly converting them into structured project tasks within their team spaces.

🚀 Installation

1. Create a Supabase project

Create a free project at supabase.com, then run supabase/migrations/0001_init.sql against it (SQL Editor → paste → run) to configure the full schema, RLS policies, and triggers.

🏃‍♂️ Running Locally

1. Backend

Navigate to the backend folder, set up your keys, and run the server:

cd backend
cp .env.example .env
npm install
npm run dev        # http://localhost:4000

Note: Ensure you fill in SUPABASE_URL, SUPABASE_SERVICE_ROLE_KEY, SUPABASE_JWT_SECRET, and your selected AI Provider Key.

2. Frontend

Navigate to the frontend directory, configure your anon variables, and start the app:

cd frontend
cp .env.example .env
npm install
npm run dev         # http://localhost:5173

Note: The Vite dev server proxies /api/* to http://localhost:4000, so both must be running for the app to work end to end.

🔐 Environment Variables

The backend supports AI_PROVIDER=groq and AI_PROVIDER=openai. Local development and testing default to Groq. Keep real API keys only in .env; do not commit secrets.

Backend (backend/.env):

AI_PROVIDER=groq
GROQ_API_KEY=your-groq-api-key
GROQ_MODEL=llama-3.1-8b-instant

To switch to OpenAI, change only the provider value and provide an OpenAI key:

AI_PROVIDER=openai
OPENAI_API_KEY=your-openai-api-key
OPENAI_MODEL=gpt-4o-mini

Frontend (frontend/.env):

VITE_SUPABASE_URL=your-supabase-url
VITE_SUPABASE_ANON_KEY=your-anon-key

📂 Folder Structure

orbitos-ai/
├── frontend/          Vite + React app (port 5173)
│   └── src/
│       ├── app/            pages (landing, auth, dashboard/*)
│       ├── components/     layout + ui primitives
│       ├── hooks/          data-fetching hooks (tasks, meetings, ...)
│       ├── stores/         Zustand stores (auth, ui)
│       ├── lib/            supabase client, api client, utils
│       └── types/          shared domain types
├── backend/            Express API (port 4000)
│   └── src/
│       ├── config/         env validation, supabase admin client
│       ├── middleware/      auth, rate limiting, error handling
│       ├── routes/         tasks, meetings, knowledge, workflows, insights, ai, users
│       └── lib/             schemas (zod), AI provider client, errors
├── supabase/
│   └── migrations/0001_init.sql   full schema + RLS policies + triggers
└── docs/
    └── ARCHITECTURE.md

🔄 Project Workflow

Meetings are initiated and recorded/captured inside OrbitOS AI. Once completed, the AI pipeline completely parses the transcription. The user reviews an Intelligent Summary detailing Action Items, Next Steps, Decisions, and Risks. Action items can then be manually selected (with Employee Assignments) and converted smoothly to tracked project tasks on the Kanban/Dashboard interface.

🚧 Future Roadmap (NOT IMPLEMENTED)

The following features are planned for future iterations and are actively NOT IMPLEMENTED yet:

⚠️ Known Limitations

🤝 Contributors

Built for the OpenAI Build Week 2026.

📄 License

MIT License