📄 DocChat — RAG Document Chat with Convex + Vapi

Upload a PDF and chat with it using text or voice.

Stack: React · FastAPI · Convex (vector DB) · OpenAI · Vapi (voice)

---

Project Structure

rag-doc-chat/
├── convex/          # Convex schema, queries, mutations, actions
├── backend/         # FastAPI — PDF ingestion, embeddings, chat streaming
└── frontend/        # React — Upload UI, Chat UI, Voice UI

---

Prerequisites

• Node.js 18+
• Python 3.11+
• Accounts: OpenAI, Convex, Vapi

---

Setup

1. Clone & open in VS Code

cd rag-doc-chat
code .

2. Init Convex

cd frontend
npm install
npx convex dev

Follow the prompts — it creates a free project and gives you a CONVEX_URL.

3. Set environment variables

backend/.env (copy from .env.example)

OPENAI_API_KEY=sk-...
VAPI_API_KEY=...
CONVEX_URL=https://....convex.cloud

frontend/.env (copy from .env.example)

VITE_CONVEX_URL=https://....convex.cloud
VITE_BACKEND_URL=http://localhost:8000
VITE_VAPI_API_KEY=...
VITE_VAPI_ASSISTANT_ID=...

4. Create a Vapi Assistant

• Go to vapi.ai → Assistants → Create
• Set webhook URL: http://localhost:8000/vapi-webhook (or your deployed backend URL)
• Copy the Assistant ID into frontend/.env

5. Run the backend

cd backend
python -m venv venv
source venv/bin/activate        # Windows: venv\Scripts\activate
pip install -r requirements.txt
uvicorn main:app --reload

6. Run the frontend

cd frontend
npm run dev

Open http://localhost:5173 — upload a PDF and start chatting!

---

Deploy

Service	Platform
Backend	Railway (set env vars in dashboard)
Frontend	Vercel (set env vars in project settings)
Database	Convex Cloud (run npx convex deploy)