Generate AI explainer videos from documents or queries using Retrieval-Augmented Generation (RAG) and HeyGen API.

🔍 Overview

This project demonstrates how to:

• Use RAG (Retrieval-Augmented Generation) to extract meaningful information from documents
• Generate video scripts using OpenAI GPT / Mistral
• Convert scripts into professional videos using HeyGen’s API
• Run everything through an interactive Streamlit or Gradio interface

🚀 Features

• 📄 Upload or point to documents
• 🧠 RAG retrieves context using vector search
• ✍️ LLM generates a human-like narration script
• 🗣️ Script is turned into a talking-head avatar video using HeyGen
• 🎬 Instant video link or download

🏗️ Architecture

1. Document Indexing (FAISS / Chroma / Qdrant)
2. RAG Chain (LangChain or custom logic)
3. Script Generation via OpenAI or Mistral
4. HeyGen API for video rendering
5. Frontend: Streamlit or Gradio

🧑‍💻 Setup

git clone https://github.com/supermldev/heygenai.git
cd heygenai
pip install -r requirements.txt

Create a .env file:

OPENAI_API_KEY=your-openai-key

HEYGEN_API_KEY=your-heygen-key

▶️ Run the App

streamlit run app.py

📦 Folder Structure

heygenai/
├── rag_pipeline/
│   ├── retriever.py         # Vector DB + similarity search
│   ├── generator.py         # Script generation using LLM
│   └── heygen_video.py      # HeyGen API call
├── app.py                   # Streamlit or Gradio UI
├── examples/                # Sample docs and outputs
├── docs/                    # Architecture diagram, screenshots
├── requirements.txt
└── .env

Example Use Cases

•	🎓 Educational Explainers
•	🏢 Internal Corporate Training
•	📢 Marketing Demos
•	🧾 Document Summaries into Video

✨ Powered by SuperML.dev

Let me know if you’d like:

• The starter Python code (app.py, rag_pipeline/)
• Video upload/share functionality

Ready to scaffold the codebase too?

Github Repo: HeyGenAI