fallback

Standard RAG trusts whatever it retrieves, even when it's irrelevant. Corrective RAG scores retrieval quality and acts on it. The signal is the grade-then-correct loop: re-retrieve, fall back to web search, or discard.

The moment a company runs LLMs in more than one place, it needs a gateway. The signal is the cross-cutting concerns (cost, reliability, observability, governance) that a gateway consolidates, not merely 'proxy the API.' Here is the design.

A provider outage should cost you quality, not the feature. The answer that scores is a pre-decided fallback ladder, a timeout policy that treats slow as down, and the discipline to test the path you hope never runs.