Ingest Supabase Files into OpenAI Embeddings and Vector Store

Overview

This n8n workflow is designed to automate the process of ingesting documents stored in Supabase into an OpenAI-powered embedding system and a Supabase vector store. It addresses the common challenge of making unstructured data (like PDFs and text files) searchable and usable by AI models.

The workflow begins by fetching all files from a specified Supabase storage bucket using the Get All files node. It then iterates through each file, downloading its content. Based on the file type (PDF or other), it uses the Extract Document PDF node or directly processes text. The prepared content is then passed to the Embeddings OpenAI node to generate vector embeddings. Finally, these embeddings, along with the file's metadata, are inserted into a Supabase table designated as a vector store using the Insert into Supabase Vectorstore node. This process enriches your Supabase data with AI-searchable vector representations.