The AI teaches. It never does the student's work. The student never gets a free prompt box. Every model call is a fixed action from a closed set, and a server-side registry decides what the model may do for that action.
The agent manages the learning. The student does the learning. Given a goal and a deadline, the study agent keeps the plan feasible as the student's week drifts, and stays quiet unless a human decision is genuinely required. It can plan, assess, schedule, and adapt. It cannot solve the student's actual assignment, invent XP, or write a mastery score.
Log in or sign up for Devpost to join the conversation.