A fully private, 100% offline AI chat assistant that runs on your local machine using Ollama and Streamlit or Flask.
Chat with your PDF, TXT, and Markdown files safely and locally.

• 📴 100% Offline – No data leaves your computer
• 📄 RAG (Retrieval Augmented Generation) – Chat with your documents
• ⚡ Fast & Efficient – Uses llama3.2:1b model
• 🧠 Persistent Memory – Local storage via ChromaDB
• 🖱️ Multiple Interfaces – Streamlit (web) & Flask Cyberpunk (modern UI)
• 🐧 Cross-Platform – Works on Windows, macOS, and Linux

• Python 3.8+
• Ollama (ollama.com)
• Git (to clone the repository)

1. Clone the Repository

   git clone https://github.com/code-glitchers/IncognitoAI.git
   cd IncognitoAI

2. Run Setup (Choose your platform)

   • Windows: Double-click START_PRIVATEAI.bat
   • Linux/macOS: cd linux && chmod +x setup.sh && ./setup.sh

3. Start Ollama (in a separate terminal)

   ollama serve

4. Launch the App

   • Streamlit Version:

     streamlit run app.py

   • Flask Cyberpunk Version (Linux):

     cd linux && chmod +x bot.sh && ./bot.sh

Run the one-click installer:

START_PRIVATEAI.bat

Follow the Linux/macOS Installation & Setup Guide

• Open http://localhost:8501
• Upload PDF, TXT, or Markdown files
• Toggle RAG mode to search documents or ask general questions

• Open http://localhost:5000
• Dark-themed neon aesthetic
• Real-time streaming responses
• Toggle between RAG and general chat modes

IncognitoAI/
├── app.py                    # Streamlit application
├── requirements.txt          # Python dependencies
├── START_PRIVATEAI.bat       # Windows launcher
├── linux/
│   ├── bot.py               # Flask Cyberpunk app
│   ├── setup.sh             # Linux setup script
│   ├── start.sh             # Start Streamlit (Linux)
│   ├── bot.sh               # Start Flask app (Linux)
│   ├── templates/           # Flask HTML templates
│   ├── static/              # CSS and JavaScript
│   └── README.md            # Linux-specific guide
└── .chroma_db/              # Local vector database (auto-created)

• LLM: llama3.2:1b - Fast, efficient language model
• Embeddings: all-minilm:latest - Fast embedding model for RAG

Models are automatically downloaded on first run via Ollama.

We welcome contributions! Feel free to:

• Report bugs and issues
• Suggest new features
• Submit pull requests
• Improve documentation

MIT License - feel free to modify and distribute!

For questions or support, open an issue on GitHub.