YT Navigator indexes YouTube channels by extracting video metadata and transcripts, converting transcript segments into vector embeddings stored in PGVector. Users can then search using natural language queries or chat with an AI that retrieves relevant video segments with timestamps. Built with Django, LangChain, and LangGraph, it implements an agentic RAG pipeline to synthesize answers from hours of video content.