🤖 Ragie Chat - AI-Powered Document Intelligence

Transform any document, video, or audio file into an intelligent, conversational knowledge base using RAG (Retrieval Augmented Generation) and OpenAI.

🌟 What Does This Do?

Imagine having a super-smart assistant that has read all your documents, watched your videos, and listened to your audio files - and can answer questions about them instantly. That's exactly what this app does!

In Simple Terms:

1. 📤 Upload any file (PDFs, videos, audio, images, etc.)
2. 🤖 Ask questions in plain English
3. 💬 Get accurate answers based on your uploaded content

It's like ChatGPT, but it actually knows about YOUR specific documents and content.

✨ Key Features

📁 Universal File Support

Upload virtually anything:

• 📄 Documents: PDFs, Word, Excel, PowerPoint, Text files
• 🖼️ Images: PNG, JPG, TIFF, and more
• 🎬 Videos: MP4, AVI, MOV
• 🎵 Audio: MP3, WAV
• 📧 Emails: EML, MSG

🧠 Smart AI Modes

🤖 LLM Mode - Get natural, conversational answers

• AI reads your documents and gives you clear, helpful responses
• Choose from GPT-5, GPT-4.1, or lightweight models
• Customize how the AI responds with system prompts

📊 Raw Retrieval Mode - See exactly what was found

• View the exact chunks of text the AI found
• See relevance scores for each piece
• Perfect for fact-checking and research

🎯 Advanced Features

• Custom System Prompts: Tell the AI how to respond (e.g., "respond in bullet points", "be concise", "act like an expert")
• Metadata Tagging: Organize documents with custom tags and filters
• Partition Support: Group documents into categories
• View Document Chunks: See exactly how your files were processed
• Smart Processing: Different modes for text extraction vs. image/table extraction

🚀 Quick Start (Non-Technical Guide)

What You'll Need

1. Node.js (version 18 or higher) - Download here
2. Ragie API Key - Get one free at ragie.ai
3. OpenAI API Key - Get one at openai.com

Installation Steps

Step 1: Get the Code

git clone https://github.com/YOUR_USERNAME/ragie-chat-app.git
cd ragie-chat-app

Step 2: Add Your API Keys

1. Find the file called .env in the main folder
2. Open it and replace the placeholder text:

   RAGIE_API_KEY=your_actual_ragie_key_here
   OPENAI_API_KEY=your_actual_openai_key_here
   PORT=3001
   

Step 3: Install & Run

# Install everything (only needed once)
npm install

# Install backend
cd backend && npm install && cd ..

# Install frontend
cd frontend && npm install && cd ..

# Start the app
npm start

Step 4: Open Your Browser

Go to: http://localhost:3002

That's it! 🎉

📖 How to Use

1. Upload Documents

1. Click the 📤 Upload tab
2. Click the pretty upload box and select a file
3. Choose processing mode:
   • Fast: Quick text extraction (good for most documents)
   • Hi-res: Extracts images and tables (slower, great for PDFs with visuals)
   • All: Processes audio and video too
4. (Optional) Add metadata tags like category: training, year: 2024
5. Click Upload Document

2. Chat with Your Documents

1. Click the 💬 Chat tab
2. Toggle 🤖 LLM Mode ON for AI answers
3. Select your preferred AI model (GPT-4.1 Mini is recommended)
4. (Optional) Click ⚙️ System Prompt to customize AI behavior
5. Type your question and press Send!

3. Browse Your Documents

1. Click the 📚 Documents tab
2. See all uploaded documents with their status
3. Click View Details to see metadata
4. Click 🔍 View All Chunks to see how the document was processed
5. Delete documents you no longer need

🎨 Understanding the Interface

Processing Modes Explained

Mode	Best For	What It Does	Speed
Fast	Text documents, emails	Extracts text only	⚡ Very Fast
Hi-res	PDFs with images/tables	Extracts text, images, and tables	🐌 Slow (20x slower)
All	Videos, audio, all media	Processes everything including audio/video	🕐 Varies

Document Status Lifecycle

Your document goes through these stages:

pending → partitioning → partitioned → refined → chunked → indexed → ready ✅

• indexed: Ready for basic chat
• ready: Fully processed with summaries

What Are "Chunks"?

When you upload a document, it gets split into smaller pieces called "chunks". Think of it like cutting a book into paragraphs - this helps the AI find relevant information quickly.

When you ask a question, the AI:

1. Searches through all chunks
2. Finds the most relevant ones
3. Uses them to answer your question

🔧 Technical Architecture

For Developers (Click to expand)

Tech Stack

Frontend:

• React 18 with Hooks
• Vite for blazing-fast development
• Axios for API communication
• LocalStorage for persistent settings

Backend:

• Node.js + Express
• OpenAI SDK (Responses API)
• Ragie SDK for document processing
• Multer for file uploads

How RAG Works Here

1. Document Ingestion (Ragie)

   • Upload → Ragie API
   • Automatic chunking, embedding, and indexing
   • Supports multiple processing strategies

2. Retrieval (Ragie)

   • User query → Ragie retrieval API
   • Returns top-K most relevant chunks with scores

3. Augmented Generation (OpenAI)

   • System prompt + retrieved chunks + user query → OpenAI
   • Uses OpenAI Responses API
   • Returns natural language answer + source chunks

API Endpoints

Backend (Port 3001):

• POST /api/documents/upload - Upload files
• GET /api/documents - List documents
• GET /api/documents/:id - Get document details
• GET /api/documents/:id/chunks - View document chunks
• DELETE /api/documents/:id - Delete document
• POST /api/retrievals - Raw retrieval (chunks only)
• POST /api/chat - RAG chat (LLM answers)

Frontend (Port 3002):

• React SPA with three main views
• Real-time updates via React state management

Project Structure

ragie-chat-app/
├── backend/
│   ├── server.js          # Express server + API routes
│   └── package.json
├── frontend/
│   ├── src/
│   │   ├── components/
│   │   │   ├── FileUpload.jsx    # Upload interface
│   │   │   ├── DocumentList.jsx  # Document management
│   │   │   └── Chat.jsx          # Chat interface
│   │   ├── App.jsx
│   │   └── main.jsx
│   └── package.json
├── .env                   # API keys (DO NOT COMMIT)
├── .gitignore
└── README.md

🎯 Use Cases

For Businesses:

• 📊 Analyze reports and presentations instantly
• 📧 Search through emails and documents
• 🎓 Create a searchable knowledge base
• 📹 Extract insights from meeting recordings

For Students:

• 📚 Study from lecture recordings
• 📝 Get summaries of research papers
• 🎥 Extract notes from video lectures
• 📖 Search across all your course materials

For Content Creators:

• 🎬 Transcribe and search video content
• 🎙️ Extract insights from podcasts
• 📄 Organize research materials
• ✍️ Find specific quotes and references

🔐 Privacy & Security

• ✅ All data is processed through Ragie's secure API
• ✅ API keys stored in .env (never committed to git)
• ✅ No data is stored by this application (only in Ragie)
• ✅ You can delete documents anytime

Note: Review Ragie and OpenAI's privacy policies before uploading sensitive information.

🛠️ Troubleshooting

Common Issues

"OpenAI API key not configured"

• Check that your .env file has OPENAI_API_KEY=your_key
• Restart the server after adding the key

"Failed to load chunks"

• Wait for document status to be indexed or ready
• Refresh the documents tab

Upload fails for video/audio

• Make sure you're using "All" processing mode
• Videos/audio can't use "Fast" or "Hi-res" modes

Port already in use

• Change PORT=3001 in .env to another number
• Update vite.config.js proxy target to match

🤝 Contributing

Contributions are welcome! This project was built to be educational and extensible.

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

📝 License

MIT License - feel free to use this for personal or commercial projects!

🙏 Acknowledgments

• Ragie - For the amazing RAG infrastructure
• OpenAI - For the GPT models
• Claude AI - For helping build this project

---

Built with ❤️ for the AI community

Star ⭐ this repo if you find it useful!