Prototype development of an AI-powered conversational coaching system for graduate research supervision

Abstract

This study developed a prototype of an AI-powered conversational coaching system to address recurring challenges in graduate research supervision. Using a design-based research approach integrated with design thinking, the study engaged 49 stakeholders—comprising students, faculty, and alumni—to pinpoint critical pain points and pedagogical requirements. These insights were distilled into a robust design framework centered on three core dimensions: problems, contexts, and learner needs (P-Q-R), integrated with the goal, reality, options, will (GROW) coaching model to facilitate goal setting and reflective practice. Expert evaluations underscored the system’s high utility, pedagogical relevance, and adaptability for resource-constrained academic environments. Beyond technical implementation, this study contributes empirically grounded design principles for AI-supported graduate supervision and offers a scalable evaluation framework for early-stage educational AI interventions.