Document intelligence platform for managing company knowledge using PageIndex (vectorless RAG).

• Documents: PDF, Word (.docx), Excel (.xlsx), PowerPoint (.pptx)
• Interface: Chat + Search
• Access: Single company, multi-user
• Auth: Username/password, all users can view/edit/add
• Hosting: Replit (primary) or cloud alternative

• Backend: Python (FastAPI)
• Frontend: React or simple HTML/JS
• RAG Engine: PageIndex (vectorless, reasoning-based)
• Database: SQLite (dev) → PostgreSQL (prod)
• Auth: Simple JWT or session-based
• File Processing:
  • PDF: PyMuPDF / pdfplumber
  • Word: python-docx
  • Excel: openpyxl
  • PowerPoint: python-pptx

┌─────────────┐     ┌─────────────┐     ┌─────────────┐
│   Frontend  │────▶│   FastAPI   │────▶│  PageIndex  │
│  (React/JS) │     │   Backend   │     │   Engine    │
└─────────────┘     └─────────────┘     └─────────────┘
                           │
                    ┌──────┴──────┐
                    ▼             ▼
              ┌──────────┐  ┌──────────┐
              │ Database │  │  Files   │
              │ (Users,  │  │ Storage  │
              │  Docs)   │  │          │
              └──────────┘  └──────────┘

1. MVP — Upload docs, build index, chat query
2. Search — Add search bar alongside chat
3. Auth — User login/registration
4. Polish — UI improvements, error handling

• Project created
• Full PageIndex integration (native library, not approximation)
• Document upload/processing (PDF, Word, Excel, PowerPoint)
• Chat interface
• Two-stage search (optimized for 4000+ documents)
• User authentication (JWT)
• Deployment to Replit
• Production hardening

Designed for 4000+ documents:

• Two-stage search: filters documents first, then deep-searches relevant ones
• Async processing pipeline
• Efficient tree structures (no vector DB overhead)

cd app

# Create virtual environment
python -m venv venv
source venv/bin/activate  # or venv\Scripts\activate on Windows

# Install dependencies
pip install -r requirements.txt

# Configure
cp .env.example .env
# Edit .env and add your OPENAI_API_KEY

# Run
python main.py

Open http://localhost:8000

app/
├── main.py              # FastAPI application
├── models.py            # Database models
├── auth.py              # Authentication
├── config.py            # Settings
├── document_processor.py # Document extraction & indexing
├── search.py            # RAG search logic
├── requirements.txt     # Dependencies
├── .env.example         # Config template
└── static/
    └── index.html       # Frontend UI

Created: 2026-02-05