Question and Answer chain with Qdrant Vector DB

[Rico_Yu]

Describe the problem/error/question

Hi guys I and trying to use the AI feature in the n8n, I am trying to build a RAG Question and answer chain by using the n8n feature and Qdrant vector database.

Here is a full picture of my current workflow, basicly I create a vector dabase qdrant server and wanna the it use the information in the vector database

In addtion, I add the data in to the qdrant database by following the instruction on Langchain
langchian Qdrant

So the format is like this:

Document( metadata ={} , page_content)

What is the error message (if any)?

Basicly I meet two problem.
I have try both format with meta data or without metadata in Qdrant by example like this:

The vector tore retrier seem like only return the meta data part:

I assume it do not return any useful information to the RAG chian? So how to build a right vector db by using Qdrant.

Also in the RAG chain, it seems like not reading any information from the chain. Not sure how to set the prompting or the prompt to refine the answer in this case

Therefore, I want to ask for how to properly build a RAG in n8n with own vector database.

Please share your workflow

Share the output returned by the last node

Information on your n8n setup

• n8n version: 1.31.1
• Database (default: SQLite):
• n8n EXECUTIONS_PROCESS setting (default: own, main):
• Running n8n via (Docker, npm, n8n cloud, desktop app): npm
• Operating system: Linux

[Anush]

Hi @Rico_Yu. The Qdrant node depends on LangchainJS.

The appropriate payload format is

{ 
"content": "...",
"metadata": {}
}

You’re using Langchain-py to add the documents, where the format is:

{ 
"page_content": "...",
"metadata": {}
}

This is why the node isn’t returning any content.

You’ll have add your documents to Qdrant using the add documents N8N-Qdrant node or LangchainJS as mentioned at Qdrant | 🦜️🔗 Langchain.

[Rico_Yu]

Thanks for information!
I modify it python code, and it also work, thankyou!