The workflow is the whole product. AI answer-machines optimize for the click that ends the session; Learn mode optimizes for the session that ends with learning. Those are not the same objective, and the difference shows up in the design.
The attempt-gate is the most user-facing rule. It’s also the one that gets the loudest complaints for the first three days and the loudest praise for the next three weeks. A cognitive load that felt like friction becomes a scaffold you count on.
What Learn mode is not
Learn mode is not a chat window with a “don’t give me the answer” system prompt. A system prompt is a stance; a stance drifts under pressure. Every rule in Learn mode is enforced by code that runs regardless of what the model was told.
It’s also not a homework helper in the copy-paste sense. It’s the tool you use when you want to actually understand what you’re doing. If you want a paragraph drafted for a topic you aren’t being graded on, Efficiency mode is the door to your right.