AI Assisted Reflective Writing: Enhancing Student Engagement and Critical Thinking in Masters Project Modules

Reflective writing is a criticalcomponentof postgraduate education, to encourage metacognition and professional development. Yet, students often struggle with structuring deep, meaningful reflections. This pilot study investigates the design and evaluation of an AI-assisted reflective writing agent, positioned within the emerging paradigm of Agentic AI. The agentleveragesNatural Language Processing (NLP) and prompt engineering to guide learners through key stages of reflection—description, analysis, and evaluation—while providing real-time feedback and suggestions aligned with academic rubrics. Deployed via Microsoft Copilot as a pilot to a small sample ofMastersstudents, the AI agent guided users through structured reflection using the Gibbs Cycle. Mixed-methods analysis of post-intervention surveys reveals high perceived usefulness, significantly improved confidence, and excellent usability. Qualitative findings highlight the agent’s success as a proactive ‘co-agent’ in structuring thought and providing guidance. However, tensions are noticedregardingthe rigidity of imposed reflective models and a desire for more critical, formative feedback. This paper argues that the AI functioned not as a passive tool but as an active pedagogical co-agent, scaffolding the writing process. We discuss the implications for designing educational AI that balances proactive guidance with learner autonomy, aligns with institutional rubrics, and upholds ethical transparency. The study contributes a practical framework for integrating Agentic AI into authentic assessment tasks whileidentifyingkey areas for future development.The research is positioned to look at scalable solutions involving AI agents to support large cohorts without compromising quality and complement lectures. Ethical considerations such as transparency, academic integrity, and GDPR compliance remain central to the design.