NotebookLlama offers a fully open-source alternative to Google's NotebookLM, enabling users to interact with documents using large language models. It is designed for researchers, students, and developers who want a customizable and transparent platform for document analysis and summarization.

How It Works

The project leverages LlamaCloud for document indexing and retrieval, integrating with LLMs like OpenAI's models and ElevenLabs for text-to-speech. It employs a multi-component architecture including a backend server (MCP) and a Streamlit frontend, with Postgres for data storage and Jaeger for tracing. This approach allows for flexible integration with various LLM providers and custom agent creation.

Quick Start & Requirements

• Install dependencies: uv sync
• Requires API keys for OpenAI, ElevenLabs, and LlamaCloud.
• Launch Postgres and Jaeger: docker compose up -d
• Run MCP server: uv run src/notebookllama/server.py
• Launch Streamlit app: streamlit run src/notebookllama/Home.py
• Additional requirement: ffmpeg
• Explore at http://localhost:8751/

Highlighted Details

• Open-source alternative to NotebookLM.
• LlamaCloud-backed for document indexing.
• Integrates with OpenAI and ElevenLabs APIs.
• Uses Postgres and Jaeger for data and tracing.

Maintenance & Community

The project is hosted on GitHub under the run-llama organization. Contribution guidelines are available.

Licensing & Compatibility

This project is provided under an MIT License, permitting commercial use and integration with closed-source applications.

Limitations & Caveats

The setup requires multiple API keys and external services (Postgres, Jaeger), and relies on specific LLM providers. ffmpeg is a necessary system dependency.