AI-powered data analysis, running privately on your machine.
View a demo of how Waraqa works at waraqa.ai.
This snapshot shows an OpenAI-powered response. The demo video shows use of local models via Ollama, but I had to go easier on it.
- Pyodide analysis environment: Use Python in WebAssembly with Pyodide for lightweight data analysis.
- Interactive visualizations: Create and interact with dynamic data visualizations, powered by Plotly.
- Flexible model selection: Use OpenAI or Ollama as your model provider.
- Edit and re-execute code: Easily re-execute and edit code to refine your analysis.
To manually build Waraqa locally, clone this repository and run:
npm install
npm run tauri buildHaven't tested this - might work?
"Waraqa" is inspired by the Arabic word "ΩΩΨ±ΩΩΩΨ©" (waraqa), which translates to "paper" or "leaf." I thought it would be a cool name that points to the simplicity of converting data to insights with AI, as well as the lightweight nature of doing data analysis with Pyodide.
There is so much missing right now, but I wanted to force myself to put something out there. A few things missing include:
- Figure out signing/notarizing Tauri apps π
- Support for more file types (only supports .csvs right now)
- Robust response handling - currently relying on model to print string or Plotly outputs from Pyodide
- Support function calling for Ollama
- Support multi-cell responses and markdown outputs
- Visualization edits (actually already there, but commented out because it's ugly right now)
- Persist chats and chat histories on local filesystem
- and so much more.
We welcome contributions! If you would like to see a feature or come across any issues (you will), please feel free to raise a PR.