AI-Powered Documentation Q&A with Retrieval-Augmented Generation

This n8n workflow automates the creation of a knowledge-based AI assistant for n8n documentation. It begins by systematically crawling, extracting, and indexed the entire n8n documentation into a Supabase vector database, effectively building a comprehensive yet efficient knowledge base. The workflow involves multiple nodes dedicated to web scraping, cleaning up text, splitting content into manageable chunks, and embedding these chunks into vectors with Google Gemini embeddings. These vectors are stored in a dedicated database table, allowing rapid retrieval of relevant information. The second part of the workflow activates a chatbot interface where users can pose questions. The bot uses retrieval-augmented generation (RAG) to fetch the most relevant documentation chunks based on user queries, ensuring responses are factual and grounded in official content. This setup is invaluable for creating accurate, real-time, documentation-specific AI assistants that can help users learn n8n more effectively or serve as knowledge bases for technical support.