Don’t be data poor — Anuj Iravane, Anterior
Healthcare AI companies can't retain or reuse PHI, making high-accuracy agents hard to build
“95% is not good enough.”
Anuj Iravane of Anterior (Sequoia/NEA-backed healthcare AI) describes building production agents for prior authorization and payment integrity over highly unstructured fax-based medical records. The central constraint is that PHI regulations prohibit retaining, anonymizing, or deriving information from patient data, leaving teams perpetually 'data poor.' The talk sets up this problem but the content is cut off before the solution is presented.