AskDocs AI

A local-first, lightweight document interaction application similar to NotebookLM. Upload documents and interact with them through Q&A, summaries, flashcards, and quizzes—all running locally on your machine.

✨ Features

• Multi-format Document Support: PDF, DOCX, PPTX, TXT, Markdown, CSV, HTML
• Intelligent Q&A: Ask questions about your documents with conversational context
• Automatic Summarization: Section-by-section summaries with overall document summary
• Flashcard Generation: Auto-generate flashcards (Q&A pairs) from document content
• Quiz Generation: Create quizzes with multiple choice, true/false, and short answer questions
• Local-first Architecture: All data stored locally, no cloud dependencies required
• Vector Search: Fast semantic search using ChromaDB and sentence transformers

🚀 Quick Start

Prerequisites

• Python 3.8 or later
• pip (Python package manager)

Installation

1. Clone or download this repository

2. Install dependencies:

   pip install -r requirements.txt
   

3. Run the application:

   python app.py
   

Optional: Webhook server (for hosting services)

Bind to the platform-provided port and expose health/webhook endpoints:

python server.py  # runs FastAPI on PORT or 5000

On Render/Heroku, ensure the process binds to $PORT.

CLI Utilities

# Download models for offline use
python manage.py download-models

# Rebuild vector index from documents
python manage.py rebuild_index

# Backup vector DB and sessions
python manage.py backup --out ./data/backups

4. Open your browser to http://localhost:7860

That's it! The application will run locally with default settings.

📋 Supported File Formats

• PDF (.pdf) - Text extraction from PDF files
• Word Documents (.docx) - Full text extraction
• PowerPoint (.pptx, .ppt) - Slide content extraction
• Text Files (.txt) - Plain text files
• Markdown (.md, .markdown) - Markdown documents
• CSV (.csv) - Tabular data
• HTML (.html, .htm) - Web pages

🏗️ Architecture

The application is organized into modular components:

AskDocs-AI/
├── app.py                 # Main entry point
├── config.py              # Configuration management
├── ui.py                  # Gradio UI interface
├── document_manager.py    # Document processing and storage
├── vector_store.py        # ChromaDB vector database
├── qa_engine.py           # Question answering engine
├── summarizer.py          # Document summarization
├── flashcard_generator.py  # Flashcard generation
├── quiz_generator.py      # Quiz generation
├── text_processor.py      # Text chunking and processing
└── parsers/               # Document parsers
    ├── document_parser.py
    ├── pdf_parser.py
    ├── docx_parser.py
    ├── pptx_parser.py
    ├── text_parser.py
    ├── markdown_parser.py
    ├── csv_parser.py
    └── html_parser.py

🔧 Configuration

Create a .env file (optional) to customize settings:

# Embedding model (local, no API key needed)
EMBEDDING_MODEL=all-MiniLM-L6-v2

# Language model (local)
LLM_MODEL=google/flan-t5-base

# Storage paths
DATA_DIR=./data
DOCUMENTS_DIR=./data/documents
VECTOR_DB_PATH=./data/vector_db

# UI settings
UI_TITLE=AskDocs AI
UI_PORT=7860
UI_DEBUG=false

Note: No API keys are required! The application runs entirely locally using open-source models.

📖 Usage Guide

1. Upload Documents

• Go to the Documents tab
• Click "Upload" and select a supported file
• Wait for processing to complete
• Select the document from the dropdown

2. View Summaries

• After selecting a document, go to the Summary tab
• View section-by-section summaries and overall summary
• Summaries are generated automatically

3. Ask Questions

• Navigate to the Q&A tab
• Type your question in the input box
• Ask follow-up questions for conversational context
• Clear history to start a new conversation

4. Generate Flashcards

• Select a document first
• Go to the Flashcards tab
• View auto-generated flashcards
• Export to JSON or text format

5. Take Quizzes

• Open the Quiz tab after selecting a document
• View generated quiz questions (multiple choice, true/false, short answer)
• Questions are auto-generated from document content

🔒 Privacy & Data

• All data stays local: Documents and embeddings are stored on your machine
• No external API calls: Everything runs locally by default
• Optional cloud integration: Can be configured to use OpenAI or Ollama if desired
• Data storage: Documents are stored in ./data/ directory

🛠️ Development

Project Structure

• Backend: Python modules for document processing, embeddings, and Q&A
• Frontend: Gradio-based web interface
• Storage: ChromaDB for vector storage, local filesystem for documents

Extending the Application

The modular architecture makes it easy to:

• Add new file formats: Create a new parser in parsers/
• Customize summarization: Modify summarizer.py
• Enhance Q&A: Update qa_engine.py with different models
• Add features: Extend ui.py with new tabs or components

🐛 Troubleshooting

Common Issues

1. Import errors: Make sure all dependencies are installed (pip install -r requirements.txt)
2. Model download: Models download automatically on first run (may take time)
3. Port already in use: Change UI_PORT in .env or kill the process using port 7860
4. Memory issues: Reduce chunk size in config.py or use a smaller embedding model

Logs

Check app.log for detailed error messages and debugging information.

📝 License

This project is licensed under the MIT License - see the LICENSE file for details.

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

🙏 Acknowledgments

• Built with Gradio for the UI
• Uses ChromaDB for vector storage
• Powered by sentence-transformers for embeddings
• Uses Hugging Face Transformers for language models