DocAnalyzer AI

Intelligent document analysis desktop application powered by IBM Granite AI and Docling.

Features

• 📄 Multi-format document support (PDF, DOCX, XLSX, TXT, Images)
• 🤖 AI-powered natural language queries using local IBM Granite models (no API key required!)
• � Run models locally with GPU acceleration or use cloud API
• �🔍 OCR support for image-based documents
• 📊 Multiple output formats (tables, charts, summaries, JSON, CSV)
• 🔒 100% local processing - your documents never leave your machine
• ⚡ Fast document analysis with real-time previews
• 📦 Batch processing for multiple documents
• 🎯 Specialized Docling models optimized for document understanding

Prerequisites

• Node.js 18.x or higher
• Python 3.10 or higher
• npm 9.x or higher

Installation

Development Setup

1. Clone the repository:

git clone <repository-url> docanalyzer-ai
cd docanalyzer-ai

2. Install Node.js dependencies:

npm install

3. Install Python dependencies:

cd src/python
pip install -r requirements.txt
cd ../..

4. (Optional) Configure environment variables for cloud API:

cp .env.example .env
# Edit .env and add your IBM Granite API key (only needed if not using local models)

5. Run the application in development mode:

npm run dev

Using Local AI Models

No API key needed! The application supports running IBM Granite models locally:

1. Launch the application
2. Go to Settings → AI Configuration
3. Enable "Use Local Model"
4. Click "Manage Models" to download your first model
5. We recommend starting with Granite Docling 258M (optimized for documents, ~1GB)

Models are downloaded from Hugging Face and cached locally. GPU acceleration is automatically used if available. See MODELS.md for detailed information.

Benefits of Local Models:

• ✅ No API costs
• ✅ Works offline (after initial download)
• ✅ Complete privacy - no data sent to cloud
• ✅ Faster response times
• ✅ Document-optimized models available

Build

Build for all platforms:

npm run build

Platform-specific builds:

npm run build:win    # Windows
npm run build:mac    # macOS
npm run build:linux  # Linux

Testing

npm run test:unit           # Run unit tests
npm run test:integration    # Run integration tests
npm run test:e2e           # Run end-to-end tests
npm run test:coverage      # Generate coverage report

Project Structure

docanalyzer-ai/
├── src/
│   ├── main/          # Electron main process
│   ├── renderer/      # Vue 3 application
│   ├── preload/       # Electron preload scripts
│   └── python/        # Python worker for AI/document processing
├── resources/         # Application resources
├── tests/            # Test files
└── docs/             # Documentation

Technology Stack

• Frontend: Vue 3 with Composition API, Pinia, Chart.js, Element Plus, TailwindCSS
• Backend: Electron, Node.js, NeDB
• AI/Processing: IBM Granite AI, Docling, Tesseract OCR, LangChain
• Build Tools: Vite, Electron Builder

Documentation

See the documentation for detailed information:

• Local Models Guide - Complete guide to using local AI models
• Build Guide - Building and packaging the application
• Design Document
• Technical Documentation

License

MIT

Support

For issues and questions, please visit:

• GitHub Issues: [repository-url]/issues
• Documentation: [docs-url]