Can AI retrieval systems find the right document, not just a relevant one?

How can we tell whether an AI retrieval system found the right financial document, not just a relevant one? In this recorded micro1 Forum session, we introduce CortexRetrievalBench, a benchmark for evaluating retrieval systems on financial documents. In our tests, authoritative documents appeared in the top 10 results for 92-99% of queries, but ranked first for only 42-69%. We discuss what that gap means when a weaker source shapes the evidence an AI model relies on. Featuring: - Mark Esposito, PhD, Chief Economist at micro1 - Shaudi Mahdavi, Member of Technical Staff at micro1 - Strib Walker, Head of Product at Rogo