Learning path · Production RAG · 50
Retrieval-Augmented Generation
Retrieve relevant external documents at query time, inject them into the prompt, then generate an answer grounded in that evidence.
Why it matters
- Primary pattern for organizational knowledge without retraining.
- Separates reasoning (LLM) from facts (retrieval index).
- Requires eval on faithfulness, not fluency alone.
Key ideas
- Retrieve then generate
- Grounding
- Citations
Top resources
- 01PaperLewis et al.
Retrieval-Augmented Generation for Knowledge-Intensive NLP
Why this resource. The original retrieve-then-generate formulation.
Covers in this concept
- retrieve
- generate
- grounding
- 02DocsLlamaIndex
Understanding RAG
Why this resource. How teams assemble RAG as a product, not a demo.
Covers in this concept
- pipeline
- evaluation
RAG fetches documents at question time, puts them in the prompt, then generates. Naive RAG stops at embed-and-stuff. Production RAG adds hybrid search, reranking, citations, and a way to refuse when retrieval is weak. Score faithful answers against the retrieved passages. Users prefer "I don't have that" to a fluent answer citing the wrong page.
Updated 2026-08-09 · Full learning path