The project aims to develop an AI-supported system for the automatic generation of structured patient summaries in geriatrics. Using deep learning methods, a wide variety of multimodal patient data is integrated, outdated and irrelevant information is filtered out, and clinically relevant aspects are highlighted. The prototype is intended to assist physicians with medication and treatment planning, reduce the burden of documentation, and increase safety. An explainable, trustworthy AI architecture (“Explainable AI” and “Trustworthy AI”) ensures transparency and trust. The system is being designed, implemented, and evaluated in collaboration with the geriatrics department, laying the groundwork for future clinical development.