Personalized learning systems increasingly leverage large language models (LLM) to offer adaptive learning plans and real-time support. While such systems excel at content adaptation, the other critical process, self-reflection, is still under development. Traditional reflection methods and even advanced reflective dialogue systems remain limited by reliance on self-immersed, first-person perspectives. Recent work shows Virtual Reality (VR)’s capacity for embodied presence can externalize internal dialogue and promote self-distancing, which could improve emotional regulation and foster deeper insights [3, 16]. Building on this foundation, we propose a novel framework that integrates an AI agent in a mobile-based interview and a virtual avatar in VR to provide second-person perspectives for embodied and personalized self-reflection. Based on a database of personal interviews, learners can converse with a virtual self in VR. By reframing reflection as an embodied dialogue with a virtual self, this approach aims to establish a new paradigm of agentic AI that transforms reflection from an isolated task into a dynamic conversation, thereby enabling deeper and more personalized self-reflection.