1
Curious builder0 XP earned · 300 to level 2
0 daysFinish a lesson to begin
Badge collection0 of 6 unlocked
51 small wins to finish your pathNext question →
What is agentic RAG, and how does it differ from classic RAG?
30-second answerSay your answer out loud first, then reveal.
What agentic RAG adds
- Query rewriting / decomposition: "Compare our Q2 and Q3 churn" becomes two retrievals.
- Source routing: docs vs SQL vs web vs API.
- Self-evaluation: grade retrieved chunks for relevance and retrieve again if they're weak (the corrective RAG / self-RAG ideas).
- Multi-hop: the answer to retrieval 1 determines what to search next.
Trade-offs
| Classic RAG | Agentic RAG | |
|---|---|---|
| Latency | 1 retrieval + 1 LLM call | Multiple loops, 3–10x slower |
| Cost | Low | Higher |
| Simple factual Q | Great | Overkill |
| Multi-hop / comparison | Weak | Strong |
Good answer. "I'd use a router. Simple questions go to classic RAG; complex or multi-hop ones go to the agentic path."
Follow-ups to expect
- How do you evaluate retrieval separately from generation? (RAGAS-style context precision/recall vs faithfulness.)
Related
Slow is fine. Stopping is the only problem.