Adding documents to a Vector Database is easily done with n8n. But what happens if that data becomes outdated and needs to be replaced?
Vector databases have become an integral part of the AI world. It is often used for historical data. But what if you want to use a knowledge database or something similar (that changes regularly) in a RAG setup?
A simple hack for n8n allows upserting documents in a vector store. It also works dynamically for embeddings split into multiple chunks.
