Task → do → verify

The single biggest difference between teachers who get magic from AI and teachers who get slop: the loop. Give it a task, let it do the task, give it a way to verify its own result. For a sub-plan drafter, verification means checking the draft against the lesson plan AND against a rule list (no student named, no level or score it was not given, no decision made for me, bell to bell) before you ever see it. And one good run proves less than it seems. The model is non-deterministic: the same prompt on the same files can come back different next time. So the check has to run every time, not once when you first set it up.