In the field of education, numerous studies focused on the integration of artificial intelligence in aspects related to the student’s learning experience. However, less focus has been given to how AI can help educators, especially in STEM subjects like physics, a field that requires visual and dynamic representations. This study presents a teacher-focused system that uses generative AI to create physics animations for primary school. It combines a language model (GPT) to understand teacher input and create scripts, with Manim, a mathematical animation engine, to produce the visuals. A readability assessment was conducted to make sure that the GPT-generated text was accessible to young learners, and a pilot study with primary school teachers revealed levels of satisfaction with the generated videos. This study aims to shift the focus to teachers and to explore the potential of AI generated visual content in improving conceptual understanding and fostering student engagement.