Integrate Supabase Vector Database with n8n for RAG Workflows

Overview

This n8n workflow provides a comprehensive solution for integrating a Supabase Vector Database with your n8n automation, specifically for building Retrieval Augmented Generation (RAG) systems. It guides you through the necessary preparation steps in Supabase, including enabling the pgvector extension, creating or altering tables with embedding, metadata, and content columns, setting up appropriate policies, and defining a custom match_documents function for efficient retrieval. The workflow then handles the ingestion of data, using the Google Drive node to fetch files, the Default Data Loader and Recursive Character Text Splitter for processing, and Embeddings OpenAI to generate vector embeddings before inserting them into your Supabase table via the Insert Documents node. Finally, it sets up a retrieval and question-answering chain using When chat message received trigger, Vector Store Retriever, OpenAI Chat Model, and Question and Answer Chain nodes, allowing users to query the ingested data.