
CA

You are working on an LLM feature for a product team that needs better answers than a plain prompt can give. The team is deciding whether to add retrieval over source documents, fine-tune a model, or do both.
Compare fine-tuning and RAG. When would you choose each?