{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreigzy45cqype5be5wypjud2z4yajcreqfedirgj3cxtobprdgsp2py",
    "uri": "at://did:plc:pgryn3ephfd2xgft23qokfzt/app.bsky.feed.post/3mlid2hvaa3h2"
  },
  "path": "/t/whats-the-relationship-among-llm-prompt-rag-prompt-engineering-metadata/101061#post_11",
  "publishedAt": "2026-05-10T07:45:39.000Z",
  "site": "https://discuss.huggingface.co",
  "textContent": "So a simple way to visualize the pipeline:\n\nPrompt Engineering\\\\] ──designs──-> \\\\[Prompt Template\n\n│\n[User Question] ──────────────────────┼──-> [Final Prompt] ──-> [LLM] ──-> Answer\n│\n[Knowledge Base] ──[Metadata]──-> [Retrieved Docs] (via RAG)\n\nThe key insight that took me a while to grasp: RAG doesn’t change the LLM at all. It changes what\ngoes into the prompt. The LLM sees exactly what a non-RAG system sees — a prompt — just with more\nrelevant context stuffed into it.\n\nHope this helps someone still wrapping their head around it!",
  "title": "What's the relationship among LLM, Prompt, RAG, Prompt Engineering, Metadata?"
}