# Build an advanced multi-query RAG system with Supabase: n8n workflow

> Started by someone opening a chat with it, and works across Supabase vectors. It puts OpenAI behind a retrieval step, so answers are drawn from documents...

Source: https://millionrounds.com/workflow/build-an-advanced-multi-query-rag-system-with-supabase (Millionrounds; every figure on the page links to its filing or article)

[Automation library](https://millionrounds.com/workflows) · [AI assistants](https://millionrounds.com/workflows?category=AI+assistants#library)

# Build an advanced multi-query RAG system with Supabase

Started by someone opening a chat with it, and works across Supabase vectors. It puts OpenAI behind a retrieval step, so answers are drawn from documents you supply rather than from the model alone. 18 steps, 1 branch, 1 loop or waits: a working afternoon to understand properly.

Medium Calls a model: OpenAI Retrieval

[Download free](https://millionrounds.com/login?next=%2Fapp%2Fworkflow%2Fbuild-an-advanced-multi-query-rag-system-with-supabase)

**18 steps** Medium: a working afternoon

**1** Branch: places the flow splits

**1** Loop or waits

**1** Service it connects to

## The flow

Starts it 2 Model 7 Logic 9

The steps in the order they run, from what starts it.

- 1 **When chat message received** Chat trigger *Starts it*
- 2 **OpenAI Chat Model** Lm Chat Open Ai *Model*
- 3 **Simple Memory** Memory Buffer Window *Model*
- 4 **RAG sub-workflow** Execute Workflow trigger *Starts it*
- 5 **Query knowledge base** Tool Workflow *Model*
- 6 **Embeddings OpenAI1** Embeddings Open Ai *Model*
- 7 **Think** Tool Think *Model*
- 8 **AI Agent** Agent *Model*
- 9 **Split Out** Split Out *Logic*
- 10 **Supabase Vector Store1** Vector Store Supabase *Model*
- 11 **Loop Over Items1** Split In Batches *Logic*
- 12 **Clean RAG output** Set *Logic*
- 13 **Aggregate items** Aggregate *Logic*
- 14 **Keep score over 0.4** Filter *Logic*
- 15 **Any chunk?** If *Logic*
- 16 **Aggregate chunks** Aggregate *Logic*
- 17 **Say no chunk match** Set *Logic*
- 18 **Prepare loop output** Set *Logic*

## How it's built

- **Starts on:** Someone opening a chat with it
- **Steps:** 18, with 1 branch, 1 loop or waits
- **Size:** Medium: a working afternoon
- **Connects to:** Supabase vectors
- **Model:** OpenAI
- **Retrieval:** Answers are drawn from documents you supply, not from the model alone

## Questions

What does it need to run?

An n8n account, plus your own Supabase vectors account, and a key for OpenAI. You connect your own accounts after importing it; none are included.

What starts it?

It starts on someone opening a chat with it.

Does it use AI?

Yes. It calls OpenAI. Its answers draw on documents you give it, not on the model alone. Model calls are billed by the provider, not by n8n.

How long does it take to set up?

Plan an afternoon. 18 steps is enough that you should read each one before running it.

How do I import it?

Copy the workflow, open a new workflow in n8n and paste. The steps appear on the canvas, ready for your own accounts.

**Before you run it**

Nothing here has been tested by us. It is somebody's working automation, read and described from its own steps. Treat the connections it wants as a list of accounts it will touch, and read the 18 steps before you point it at anything that matters.

## Run it in n8n

- **Create your n8n account** Free to start. n8n is where the workflow runs. [Create an n8n account ↗](https://millionrounds.com/n8n)
- **Download the workflow** Free with a Millionrounds account. [Download free](https://millionrounds.com/login?next=%2Fapp%2Fworkflow%2Fbuild-an-advanced-multi-query-rag-system-with-supabase)
- **Paste it into n8n** Open a new workflow in n8n and paste. Then connect your own accounts.

Stored with the author's traces removed: canvas notes, instance identifiers, saved credential names and any live document they had left behind.

## Connects to

Each opens the other workflows that use it.

[Supabase vectors](https://millionrounds.com/workflows?service=Supabase+vectors#library)

## Workflows like it

Also ai assistants, with the same services where possible, closest in size first.

- [**Build a Placetel RAG voice assistant with Google Drive** 16 steps · calls a model · Google Drive, Supabase vectors](https://millionrounds.com/workflow/build-a-placetel-rag-voice-assistant-with-google-drive)
- [**Build a RAG chatbot with Gemini and Supabase** 16 steps · calls a model · Supabase vectors](https://millionrounds.com/workflow/build-a-rag-chatbot-with-gemini-and-supabase)
- [**Index and update web knowledge base with Jina and Gemini** 20 steps · calls a model · Supabase vectors, n8n data table](https://millionrounds.com/workflow/index-and-update-web-knowledge-base-with-jina-and-gemini)
- [**Power Placetel voice answers from OneDrive docs with OpenAI** 16 steps · calls a model · Supabase vectors](https://millionrounds.com/workflow/power-placetel-voice-answers-from-onedrive-docs-with-openai)
- [**Answer voice queries from a webhook over Google Drive docs** 15 steps · calls a model · Google Drive, Supabase vectors](https://millionrounds.com/workflow/answer-voice-queries-from-a-webhook-over-google-drive-docs)
- [**Ultimate RAG system (Query data and tables) with Claude** 21 steps · calls a model · Pinecone, Supabase vectors](https://millionrounds.com/workflow/ultimate-rag-system-query-data-and-tables-with-claude)