We present an intelligent, emotion-aware educational planner designed to deliver a highly personalized, self-paced learning experience. The system integrates task decomposition, graph-based reasoning, utility-driven scheduling, and affective state modeling into a cohesive, adaptive framework. High-level learner goals are broken down into actionable tasks via language models and structured into graph representations, which are analyzed using graph neural networks. A utility function—modulated by affective and cognitive state—guides task prioritization, while a contextual multi-armed bandit model dynamically schedules daily activities. The system also incorporates agentic AI to support user autonomy, mid-session adaptation, and long-term engagement. By merging planning, emotional intelligence, and explainable automation, this work proposes a modular architecture for learner-centric, holistic task management.