The learning loop is the interesting part.
Models will improve. Operational judgment is much harder to replicate: what to trust, what to review, what to escalate, and what not to act on.
That's where institutional knowledge becomes a real asset.
@satyanadella I think a lot of the value ends up in the learning loop around the model, not the model itself. Most teams already know what good looks like. The challenge is making that judgment repeatable.