Document Chat Application

A RAG (Retrieval-Augmented Generation) application that allows you to chat with documents from a local directory using Google's Gemini AI.

Architecture

This app consists of:

• Frontend: React + TypeScript (Vite) running on port 3000
• Backend: Express server running on port 3001 that handles Gemini API calls and reads local documents

Prerequisites

• Node.js (v18 or higher)
• A Gemini API key from Google AI Studio

Setup

1. Install dependencies

Install frontend dependencies:

npm install

Install backend dependencies:

cd server
npm install
cd ..

2. Configure the backend

Edit server/.env and set:

• GEMINI_API_KEY: Your Gemini API key
• DOCUMENTS_DIR: Path to your documents directory (default: /Users/aneal/Library/CloudStorage/OneDrive-UniversityofEdinburgh/national_security_analysis/gemini_PDF_processor/output/clean_final)

Example:

GEMINI_API_KEY=your_actual_api_key_here
DOCUMENTS_DIR=/path/to/your/documents
PORT=3001

3. Run the application

You need to run both the backend and frontend:

Terminal 1 - Backend:

cd server
npm run dev

Terminal 2 - Frontend:

npm run dev

The app will be available at http://localhost:3000

How it works

1. First startup: The app automatically loads all documents from the configured directory and creates a persistent RAG store (5-10 minutes for 607 documents)
2. Subsequent startups: The app uses the cached RAG store and loads instantly (2-3 seconds)
3. You can chat with the documents - the AI will search through them to answer your questions
4. Clear Chat button: Clears the conversation history but keeps the documents loaded
5. Reload button: Forces re-upload of all documents from the directory (use this if you add new documents to the folder)

Supported File Formats

• .txt - Plain text
• .pdf - PDF documents
• .doc, .docx - Word documents
• .md - Markdown
• .html - HTML files
• .json - JSON files