Agentic AI as Adaptive Partner in Leadership Education

Graduate leadership education faces a persistent and structural challenge: the complexity of real-world leadership practice does not translate readily into the static forms through which it ismost commonly taught. This article examines the design architecture of an agentic live case study platform developed for MLS3311 (Self-Management and Delegation) at the National Institute of Education, Nanyang Technological University, Singapore. Drawing on intelligent tutoring systems research, andragogical theory, and contemporary AI agent architectures, the article proposes a multi-layered design in which Agentic AI functions as a responsive and adaptive partner in the learning process. The platform integrates five distinct input streams — structured lecture content,facilitatedlearning, group discussion, individual learner input, and an Agentic AI Learning Agent — within a unified and coherent learning environment. The Learning Agent architecture,comprisinga Critic, Learning Element, Performance Element, and Problem Generator, drives an adaptive feedback loop with state memory, delivering real-time diagnosis,personalisedformative feedback, and Socratic prompting grounded in the learner's own professional context. The design is grounded in adult learning principles and addresses three core student needs: ongoing self-assessment of learning progress, immediate and actionable feedback, and a pathway for continual professional development beyond the formal course. The article argues that when thoughtfully integrated as adaptive heterarchical partners, AI agents can transform abstract leadership constructs into tangible, trackable, and professionally generative learning experiences — while preserving the human educator as the essential authority over standards, values, and judgement.