# RAG-powered document chatbot with OpenAI and Gemini: n8n workflow

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

Source: https://millionrounds.com/workflow/rag-powered-document-chatbot-with-openai-and-gemini (Millionrounds; every figure on the page links to its filing or article)

[Automation library](https://millionrounds.com/workflows) · [Documents and files](https://millionrounds.com/workflows?category=Documents+and+files#library)

# RAG-powered document chatbot with OpenAI and Gemini

Started by someone opening a chat with it, and works across in-memory vectors. It puts Gemini and OpenAI behind a retrieval step, so answers are drawn from documents you supply rather than from the model alone. 32 steps, 1 branch, 1 loop or waits: large, and worth reading before you trust it with anything live.

Large Calls a model: Gemini, OpenAI Retrieval

[Download free](https://millionrounds.com/login?next=%2Fapp%2Fworkflow%2Frag-powered-document-chatbot-with-openai-and-gemini)

**32 steps** Large: read it before trusting it with anything live

**1** Branch: places the flow splits

**1** Loop or waits

**1** Service it connects to

## The flow

Starts it 2 Model 10 Logic 5 Service 15

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

- 1 **OpenAI Chat Model** Lm Chat Open Ai *Model*
- 2 **Window Buffer Memory** Memory Buffer Window *Model*
- 3 **When chat message received** Chat trigger *Starts it*
- 4 **Embeddings Google Gemini** Embeddings Google Gemini *Model*
- 5 **Recursive Character Text Splitter** Text Splitter Recursive Character Text Splitter *Model*
- 6 **On form submission** Form trigger *Starts it*
- 7 **Embeddings Google Gemini1** Embeddings Google Gemini *Model*
- 8 **Simple Memory** Memory Buffer Window *Model*
- 9 **AI Agent** Agent *Model*
- 10 **Code** Code *Logic*
- 11 **Simple Vector Store** Vector Store In Memory *Model*
- 12 **Default Data Loader** Document Default Data Loader *Model*
- 13 **Query Data Tool1** Vector Store In Memory *Model*
- 14 **Split Out** Split Out *Logic*
- 15 **Loop Over Items** Split In Batches *Logic*
- 16 **Aggregate** Aggregate *Logic*
- 17 **Switch** Switch *Logic*
- 18 **Convert to File** Convert To File *Service*
- 19 **Convert to File1** Convert To File *Service*
- 20 **Convert to File2** Convert To File *Service*
- 21 **Convert to File3** Convert To File *Service*
- 22 **Convert to File4** Convert To File *Service*
- 23 **Convert to File5** Convert To File *Service*
- 24 **Convert to File7** Convert To File *Service*
- 25 **Convert to File8** Convert To File *Service*
- 26 **Extract from File** Extract From File *Service*
- 27 **Extract from File1** Extract From File *Service*
- 28 **Extract from File2** Extract From File *Service*
- 29 **Extract from File3** Extract From File *Service*
- 30 **Extract from File4** Extract From File *Service*
- 31 **Extract from File6** Extract From File *Service*
- 32 **Extract from File7** Extract From File *Service*

## How it's built

- **Starts on:** Someone opening a chat with it
- **Steps:** 32, with 1 branch, 1 loop or waits
- **Size:** Large: read it before trusting it with anything live
- **Connects to:** in-memory vectors
- **Model:** Gemini, 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 in-memory vectors account, and a key for Gemini and 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 Gemini and 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?

Take your time. At 32 steps it is large: read it before trusting it with anything live.

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 32 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%2Frag-powered-document-chatbot-with-openai-and-gemini)
- **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.

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

## Workflows like it

Also documents and files, with the same services where possible, closest in size first.

- [**Generate business requirement documents with Google Workspace** 28 steps · calls a model · Google Drive, Google Sheets](https://millionrounds.com/workflow/generate-business-requirement-documents-with-google-workspac)
- [**Evaluate RAG response accuracy with OpenAI:** 21 steps · calls a model · in-memory vectors](https://millionrounds.com/workflow/evaluate-rag-response-accuracy-with-openai)
- [**Evaluation metric example: RAG document relevance with OpenAI** 19 steps · calls a model · Google Sheets, in-memory vectors](https://millionrounds.com/workflow/evaluation-metric-example-rag-document-relevance-with-openai)
- [**Answer product queries with WhatsApp and OpenAI** 18 steps · calls a model · WhatsApp, in-memory vectors](https://millionrounds.com/workflow/answer-product-queries-with-whatsapp-and-openai)
- [**Build a retrieval-based chatbot with Telegram and OpenAI** 13 steps · calls a model · Google Drive, Telegram](https://millionrounds.com/workflow/build-a-retrieval-based-chatbot-with-telegram-and-openai)
- [**AI-powered company documents Q&A assistant with Google Drive** 12 steps · calls a model · Google Drive, in-memory vectors](https://millionrounds.com/workflow/ai-powered-company-documents-q-and-a-assistant-with-google-d)