/

LLMs & Reasoning

/

RAG (Retrieval-Augmented Generation)

RAG (Retrieval-Augmented Generation)

/ rag-retrieval-augmented-generation /

Retrieving relevant documents at query time and grounding the model’s response in them — the standard defense against hallucination.

Retrieving relevant documents at query time and grounding the model’s response in them — the standard defense against hallucination.

Why it matters

RAG is the standard architecture for keeping agents factual: answer from retrieved documents, not memory. Its failure modes — bad retrieval, ignored context — need their own tests.

Related — LLMs & Reasoning