At our company, we’re starting to include AI and automated workflows to streamline some internal processes. One of our current projects is building an AI agent that can answer questions about our products and their inventory.
The main challenge is that the inventory changes frequently. We’re looking to implement a RAG setup that stays as up to date as possible.
What would be the best approach for this?
Any recommended tools, patterns, or real-world examples using n8n?
To keep your knowledge base current with the latest information, I would suggest have a workflow which runs on either a scheduler to pull the latest information into your vector store or having some sort of a trigger to listen to file drops on a storage location (Google drive, S3, etc etc).
I would recommend this video to see how to update your vector stores and some different approaches of implementing your RAG
I dont have direct experience with it but Ive heard really good things about pinecone and the accuracy of its results. Apparently Pinecone’s Assistant is pretty powerful according to this video