Quivr - Your Second Brain, Empowered by Generative AI
Quivr is an innovative platform that aims to act as your "second brain," leveraging the power of Generative AI to become your personal assistant. It helps users efficiently manage and access information across various file formats and sources.
Key Features 🎯
- Opiniated RAG: Quivr includes a Resource Accumulation Graph (RAG) that is both speed-optimized and efficient, allowing you to concentrate on your primary tasks.
- LLMs Compatibility: It seamlessly integrates with a wide range of Large Language Models (LLMs) such as OpenAI, Anthropic, Mistral, and Gemma.
- File Flexibility: The platform supports multiple file types, including PDF, TXT, and Markdown, and users can incorporate their custom parsers.
- Customizable RAG: Quivr provides the flexibility to modify your RAG, such as adding internet searches and integrating additional tools.
- Megaparse Integration: It works harmoniously with Megaparse, enabling users to ingest files effortlessly and utilize the RAG effectively.
Quivr manages the technical aspects of the RAG, so you can divert your attention to other focal areas, merely by installing quivr-core into your project.
Getting Started 🚀
The journey with Quivr begins with documentation available on their website. To set up, users need Python 3.10 or later. Installation is simplified to a few key steps:
-
Install the Package:
pip install quivr-core -
Create a RAG in Python: Use the provided code snippet to set up a basic RAG.
-
Configuration: Set environment variables with API keys, create configuration files, and initialize a "brain" using provided commands.
-
Launch a Chat: Interact with your "brain," asking questions and receiving responses to test retrieval strategies.
Advanced Features
Quivr facilitates further exploration through internet search additions and tool integrations. Detailed guidance is available in the documentation.
Contributors & Participation
Quivr attributes its success to a dedicated team of contributors. Users are encouraged to contribute by reviewing open pull requests or proposing new ideas on their project board available on GitHub.
Partners ❤️
Quivr's development is supported by partnerships, including Y Combinator and Theodo, reflecting a collaborative effort in enhancing the platform.
License 📄
Quivr is released under the Apache 2.0 License, ensuring open-source distribution and collaboration.