Live app: https://thanos-theta.vercel.app (root Next.js proxy → frontend app + backend API)
THANOS is an advanced AI orchestration system that unifies multiple AI models into a single intelligent platform.
It is not just a chatbot — it is a Prompt-to-Application Generator powered by multiple AI agents.
Thanos/
├── frontend/ # Vite + React SPA (the app UI)
│ ├── src/components/
│ ├── src/features/ # API calls (auth, chat, billing, agent)
│ ├── src/pages/
│ └── src/redux/ # user / conversation / message state
├── backend/ # Express microservices (deployed as serverless functions)
│ ├── api/ # Vercel serverless entries (gateway, auth, chat, agent, billing)
│ ├── gateway/ # API gateway (auth middleware + proxy)
│ ├── services/
│ │ ├── auth/ # Firebase auth + user model
│ │ ├── chat/ # conversations & messages
│ │ ├── agent/ # multi-agent LLM orchestrator
│ │ └── billing/ # Razorpay payments & credits
│ └── shared/redis/ # shared Redis client (serverless-safe)
└── app/ # root Next.js app — thin proxy (frontend + /api → backend)
- Root app — https://thanos-theta.vercel.app (project
thanos) - Frontend SPA — served by the root app via rewrite (project
thanos-frontend) - Backend API — https://thanos-backend.vercel.app (project
thanos-backend)
Every push to main auto-deploys all three projects. All traffic is served through
https://thanos-theta.vercel.app: / renders the frontend app and /api/* is proxied
to the backend gateway.
# Backend microservices
cd backend
npm install
npm run dev
# Frontend SPA (set VITE_SERVER_URL=http://localhost:8000 in frontend/.env)
cd frontend
npm install
npm run dev- Frontend: Vite, React, Redux Toolkit
- Backend: Node.js, Express, Redis, MongoDB
- AI: GPT, Claude, DeepSeek, Google Gemini, Tavily, Qdrant
- Payments: Razorpay
- Auth: Firebase
- Deploy: Vercel (serverless)