YouTube RAG Search with Web Frontend (Apify, Qdrant, AI)

This n8n workflow enables you to create a Retrieval-Augmented Generation (RAG) search system over YouTube videos. It scrapes video transcripts from specified channels using Apify, breaks them into chunks, generates embeddings with OpenAI, and stores them in Qdrant vector database for efficient similarity search. A web frontend powered by a webhook allows users to query videos naturally, retrieve relevant results, and play embedded videos directly.

The workflow operates in two stages: first, populate the vector store by running initial steps to scrape, process, and index transcripts with metadata for filtering. Second, activate the webhook to serve the interactive web app, which uses Qdrant's search groups API for broader, more diverse results. Rate limiting is included for production use, but can be disabled for personal setups.

Benefits include rapid research across video content without manual transcription, scalable to many videos, and customizable LLM integration for precise answers. Ideal use cases: analyzing conference talks, educational channels, office hours (e.g., n8n videos), market research, or content discovery. Saves hours of manual searching and enables AI-powered insights from video archives.

Setup requires Apify, Qdrant, and OpenAI credentials. Run steps 1-3 once to index data, then activate for instant frontend access via the webhook URL.