Artificial intelligence is reshaping medical education and helping to address clinical assessment design challenges. Developing high-quality OSCE stations is time-consuming, prone to inconsistency and difficult to scale. Large language models make it possible to create realistic, diverse and customizable clinical scenarios. The article presents 10 practical, evidence-based tips for using AI to create pedagogically sound and clinically authentic OSCE stations. The tips are based on a literature review, experience with AI-assisted case development, and a consensus of medical education experts. Adherence to tips increases efficiency, maintains validity, fairness, safety and ethical standards. These recommendations prepare students for the complexities of modern clinical practice.