E-Portfolios provide a flexible and competence-oriented assessment format that supports students’ reflection processes. However, in larger courses individual feedback is not scalable. The use of artificial intelligence (AI), for example in the form of specific chatbots, can support teachers in providing formative feedback. This study examines the perceived usefulness and pedagogical quality of Generative AI-supported feedback systems as reported by instructors and students. The results demonstrate how AI-generated feedback can serve as a bridge between theoretical vision of personalized, scalable education and its practical implementation. The results indicate that AI-generated feedback is perceived as well-structured and helpful by participants. However, its usefulness is dependent on careful prompt engineering and adaptation to specific disciplinary contexts. This study utilises an analysis of quantitative metrics and qualitative assessments from educators and students to indicate that AI-supported feedback reduces instructor workload (e. g. in feedback provision) and supports competence development when implemented within a well-designed pedagogical framework. The integration of AI within competence-oriented education has the potential to enhance its scalability. However, to ensure contextual sensitivity, the incorporation of a human-in-the-loop approach remains a critical component in such endeavours.