Best Practices for Updating a Vector Store in a RAG-Powered Chatbot

Hi everyone,

I’m building a chatbot with n8n that should be able to answer questions about a hundred different products. I’ve been exploring various RAG (Retrieval-Augmented Generation) tutorials with n8n, but I haven’t found clear guidance on how to update the vector store when products are modified, added, or removed.

What would be the best approach to keep my vector database up to date dynamically? Should I periodically re-index the entire dataset, or is there a more efficient way to update only the relevant parts?

I appreciate any insights or best practices from those who have implemented similar solutions!

Thanks! :blush:

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:

Here’s a tutorial that might help: