Medical simulation training received AI-powered enhancement through automated analysis of team communications during high-fidelity scenarios involving 41 healthcare professionals at University Hospital Zurich. Two large language models processed verbal interactions and generated structured teamwork reports using the Team-FIRST framework, capturing nuanced communication patterns that human observers might miss due to cognitive overload. The AI systems demonstrated particular strength in providing detailed transcripts with illustrative quotes that supported more comprehensive post-simulation feedback sessions. This represents a significant advancement in simulation-based medical education, where effective debriefing has long been constrained by facilitators' limited capacity to simultaneously observe complex team dynamics while managing session logistics. The approach addresses a critical gap in healthcare training, where communication failures contribute substantially to medical errors. However, the technology showed notable limitations in accuracy and contextual understanding that prevent immediate clinical deployment. The exploratory nature and small sample size suggest this is early-stage research requiring validation across diverse medical specialties and training environments. If refined, AI-assisted debriefing could democratize access to high-quality simulation education by reducing dependency on expert facilitators while maintaining educational effectiveness.