DocuMind

DocuMind is a RAG (Retrieval-Augmented Generation) system that enables you to chat with your documentation. It ingests content from a provided URL, indexes it, and allows you to ask natural language questions to receive context-aware answers.

Features

• RAG Architecture: Retrieves relevant context from documentation to prevent hallucinations.
• URL Ingestion: Scrapes and processes content directly from documentation websites.
• Semantic Search: Uses local embeddings (@xenova/transformers) for fast, private vector search.
• Smart Answers: Integrates with Groq (Llama 3 70B) for fast and accurate responses.
• Responsive UI: Clean interface with dark mode support.
• Serverless Ready: Optimized for deployment on platforms like Vercel.

Tech Stack

Frontend

• React (Vite)
• Vanilla CSS
• Lucide React

Backend

• Node.js & Express
• @xenova/transformers (Local embeddings)
• Cheerio (Web scraping)
• Groq SDK (LLM inference)
• In-Memory Vector Store

Getting Started

Prerequisites

• Node.js (v18 or higher)
• A Groq Cloud API Key

Installation

1. Clone the repository

2. Backend Setup

   Navigate to the backend directory and install dependencies:

   cd backend
   npm install

   Create a .env file in the backend directory:

   GROQ_API_KEY=your_groq_api_key_here
   PORT=3000

   Start the backend server:

   npm run dev

3. Frontend Setup

   Open a new terminal, navigate to the frontend directory:

   cd frontend
   npm install

   Create a .env file in the frontend directory:

   VITE_API_BASE_URL=http://localhost:3000

   Start the frontend:

   npm run dev

4. Usage Visit http://localhost:5173 in your browser.

Deployment

Vercel

This project is configured for Vercel deployment.

1. Push to GitHub: Ensure your code is in a GitHub repository.
2. Import to Vercel: Import the repository in Vercel.
3. Environment Variables:
   • Add GROQ_API_KEY to the Vercel project settings.
   • Configure VITE_API_BASE_URL if necessary (Vercel automatic URL handling usually works, you may need to set the production backend URL).
4. Deploy: Vercel will detect vercel.json and deploy both the frontend and backend.

Architecture

1. Ingest: The backend scrapes text from the URL, chunks it, generates embeddings, and stores them in memory.
2. Query: The backend generates an embedding for the question, finds the top similar chunks, and sends the context + question to the Groq LLM.
3. Response: The LLM generates an answer based strictly on the provided context.

License

MIT