> RAG is basically good old information retrieval with LLMs doing the querying.

No - rag is doing search before you call the llm to give it context from some corpus like your helpdesk articles.