Conventional task management and study applications frequently do not cater to the distinctive cognitive profiles of students with Attention Deficit Hyperactivity Disorder (ADHD) and associated neurodivergent conditions. These stu-dents often have trouble with executive dysfunction, time blindness, and not being able to control their own dopamine levels, which makes standard "wall of tasks" interfaces mostly useless. This paper presents an innovative AI-driven web application aimed at delivering active cognitive scaffolding and real-time physical accountability. The system uses a Transformer-based Large Language Model (LLM) to break down difficult academic goals into smaller, time-limited tasks that can be done one at a time. The platform uses TensorFlow.js to add a lightweight Convolutional Neural Network (CNN) for real-time, privacy-protecting facial presence detection at the edge. This keeps people interested all the time. Also, a gamified progression system replaces delayed gratification with immediate reward loops, which keeps users interested. Preliminary evaluations indicate that the integration of active physical monitor-ing with AI task generation markedly diminishes initiation friction, reduces cognitive load, and enhances overall focus retention among neurodivergent users.

Keywords: ADHD, Artificial Intelligence, BlazeFace, Computer Vision, Executive Dysfunction, Gamification, Large Language Models, Edge Computing