Choice of vector databases

Questions
1

Hello everyone!

I’m new to using vector databases and I need your expert advice.
I would like to get some feedback on the use of vector databases.

I create a lot of chatbots that need access to documents, and I have several questions:

  • Vector databases like QDRANT seem overkill for my needs (at least $60 per month), so if I create one for each chatbot, my costs will be huge.
  • I think that if I use one cluster to store the documents for all my chatbots, the responses might not be relevant, and in addition, some confidential data could potentially leak.
  • Maybe using the vector databases available in n8n could be interesting if I increase the storage on my virtual machine where n8n is hosted?

If you could share your experiences on all of this, it would be super helpful!
Thank you all!

2

It looks like your topic is missing some important information. Could you provide the following if applicable.

  • n8n version:
  • Database (default: SQLite):
  • n8n EXECUTIONS_PROCESS setting (default: own, main):
  • Running n8n via (Docker, npm, n8n cloud, desktop app):
  • Operating system:
3

Nobody have any experience about that ?

4

Hi @TomLaizy

While n8n does not provide a built-in vector store we have several other integrations, such as with Pinecone, Supabase or Zep.

I think it comes down to your personal preference and I would recommend to give each of them a try.

You could also check out our tutorial section for vector store topics - Tutorials - n8n Community or try out some of our workflow templates

1 Like
5

@TomLaizy

I think that if I use one cluster to store the documents for all my chatbots, the responses might not be relevant, and in addition, some confidential data could potentially leak.

You can use qdrant’s filters to segment customer, department, project, documents etc.

  • When inserting into Qdrant, be sure to add the correct metadata in the document loader.
  • When retrieving data, use the “search filter” option in the qdrant node to apply the filter.

image
image3584×1938 295 KB

4 Likes
6

Thanks a lot

7

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.