What's the relationship among LLM, Prompt, RAG, Prompt Engineering, Metadata?

So a simple way to visualize the pipeline: Prompt Engineering\\] ──designs──-> \\[Prompt Template │ [User Question] ──────────────────────┼──-> [Final Prompt] ──-> [LLM] ──-> Answer │ [Knowledge Base] ──[Metadata]──-> [Retrieved Docs] (via RAG) The key insight that took me a while to grasp: RAG doesn’t change the LLM at all. It changes what goes into the prompt. The LLM sees exactly what a non-RAG system sees — a prompt — just with more relevant context stuffed into it. Hope this helps someone still wrapping their head around it!