RAG Workbench

A modular desktop RAG application with a PyWebView shell, Python backend services, and a professional multi-page English UI.

This project was developed as part of LEVEL UP AI | ROKETSAN Artificial Intelligence Hackathon - AI Workshop Day 1, under the Build Your Own AI Agent: An Intelligent Assistant That Understands Documents challenge. Here is the partipicating certification: https://coderspace.io/sertifikalar/7f8f47c9-951d-4196-af20-55261decd605

Run

python -m venv .venv
.\.venv\Scripts\Activate.ps1
pip install -r requirements.txt
python main.py

Build EXE

python build_exe.py

Providers

The app is structured for OpenAI, Anthropic, Gemini, DeepSeek, NVIDIA NIM, Ollama/local, and custom OpenAI-compatible endpoints. API keys are entered inside the app from the Providers page.

Current Scope

This first implementation includes the desktop shell, multi-page UI, provider/model configuration, local key storage hooks, document collection scaffolding, RAG settings, and backend bridge methods. Provider calls and full indexing can be extended through the adapter interfaces in app/providers.