What is Agentic RAG?
Agentic RAG combines AI agent capabilities with retrieval-augmented generation. Instead of a single retrieve-then-generate step, an agent orchestrates multiple retrieval rounds, reformulates queries, and synthesizes information iteratively.
Why It Matters
Basic RAG fails on complex queries requiring information from multiple documents or follow-up reasoning. Agentic RAG can ask "I found pricing info, but I also need the terms—let me search for that separately."
Basic RAG vs Agentic RAG
Basic RAG
- • Single retrieval step
- • Fixed query (user input)
- • Top-k results only
- • No reasoning about gaps
Agentic RAG
- • Multiple retrieval rounds
- • Query reformulation
- • Multi-source synthesis
- • Identifies info gaps
Agentic RAG Patterns
- • Self-RAG — Agent decides when retrieval is needed
- • Corrective RAG — Verify retrieved docs, re-search if irrelevant
- • Adaptive RAG — Choose retrieval strategy based on query type
- • Multi-hop RAG — Chain retrievals for complex questions