RagReader

Inspiration

The need to provide the right context from a file or website to AI, ensuring more accurate answers.

What It Does

Retrieves information from files like PDFs, TXT, URLs, and CSVs, then generates answers based on the given prompt and context.

How We Built It

Using Vite, React, and Tailwind CSS for the frontend, and Django REST with Python for the backend. We implemented Langchain for RAG and integrated OpenAI GPT, Mistral, and Anthropic Haiku as AI models. The app is deployed on DigitalOcean.

Challenges We Ran Into

Customizing the RAG system to support multiple file types, including PDFs, TXT, CSVs, and URLs.

Accomplishments We're Proud Of

It works!

What We Learned

How to implement RAG with Langchain and handle file submissions from the frontend to the backend.

What's Next for RagReader

• Supporting more file formats, including images (vectors).
  
• Integrating additional AI models.
  
• Developing a desktop app.
  

Built With

• ai
• chatgpt
• claude
• digitalocean
• django
• mistral
• openai
• python
• react
• rest
• typescript
• vite