See projects
See projects
← AI Automation

Independent project · n8n

RAG Agent with Postgres Memory + Supabase

A chat agent that answers questions from your own documents. Files from Drive are embedded into a Supabase vector store, and conversation history persists in Postgres.

Project
RAG Agent with Postgres Memory + Supabase
Sheet
03 / 03
Date
Mar 2026
Type
AI Automation
n8n canvas of the RAG agent workflow.

Vector store

pgvector

Memory

Persistent

01

How it works

An ingestion workflow loads documents from Google Drive, splits and embeds them with OpenAI, and stores the vectors in Supabase. The chat agent queries that store as a tool and keeps conversation history in Postgres, so follow-up questions keep their context.

Chat input and grounded answer from the RAG agent.
Supabase documents table with embedded chunks.
Embedded document chunks in Supabase
Postgres chat history table.
Persistent chat memory