qrydocs is a document-based AI assistant platform that transforms static business documents into intelligent, interactive assistentes tailored to specific organizational needs. The platform addresses the critical problem of knowledge silos and inefficient information retrieval in enterprises, where employees waste significant time searching through scattered documentation, manuals, and internal resources.
Users upload their documents (contracts, manuals, policies, etc.) and the system creates a personalized AI assistant that understands domain-specific terminology and workflows. The market potential is substantial, targeting the growing enterprise AI automation sector where businesses seek to improve operational efficiency, reduce support costs, and democratize access to institutional knowledge across teams of varying technical expertise.
- 📄 Multi-format support: PDF, DOC, DOCX, TXT, MD files
- 🎨 Drag-and-drop interface: Simple file upload interface
- 🔍 Smart search: Hybrid semantic + keyword search
- ⚡ Fast responses: Powered by local AI models
- 🔒 Privacy: Your data stays on your infrastructure
- Python 3.13+
- Node.js 20+
- Docker
# Get the code
git clone git@github.com:levyaraujo/qrydocs.git
cd qrydocs
# Start database
docker-compose up -d
# Install Python dependencies
pip install uv
uv sync
# Install frontend dependencies
cd frontend && pnpm install && cd ..# Install Ollama
curl -fsSL https://ollama.ai/install.sh | sh
# Download embedding model
ollama pull mxbai-embed-large
ollama pull qwen3:4b# Start backend
uv run python app/main.py
# Start frontend (new terminal)
cd frontend && pnpm devOpen http://localhost:5173 and start uploading documents!
- Upload: Drag your documents into the web interface
- Wait: Documents are processed and indexed automatically
- Chat: Ask questions about your documents
- Get Answers: Receive accurate responses with source references
- Backend: FastAPI + LangChain + Qdrant
- Frontend: React + TypeScript + Tailwind
- AI: Ollama (local embeddings)
- Database: Qdrant vector database