The case for async AI office hours

Universities spend enormous resources on synchronous advising that scales poorly. What would a product look like that used RAG + persona modeling to make expert judgment available at scale — without pretending it's the real thing?

// in-progress