Question

A financial services company runs a customer-facing assistant built on a foundation model using Retrieval Augmented Generation (RAG). Product policies and interest rates change several times per week, and the assistant must always reflect the latest figures. The team wants an approach that keeps answers current with minimal cost and effort. What should they do?

Accepted Answer

Update the documents in the RAG knowledge base and re-index them so retrieval returns the latest policies and rates

Answer

Fine-tune the foundation model each time a policy or rate changes so the updated values are learned into the model weights

Answer

Increase the model's temperature setting so it generates more up-to-date responses

Answer

Switch to a larger foundation model with more parameters to improve recall of recent information