The digital transformation challenges teachers to constantly keep up to date with new technologies, often leading to a lack of time for further training and uncertainty in using these technologies. To address these challenges, this conceptual paper introduces the Teacher Copilot, a training and assistant system based on Large Language Models (LLM) and showcases its integration into the curriculum. The aim is to explore the use of generative AI in the teaching context, thereby supporting teachers in their continuous training, teaching, and work processes through AI-based educational technologies. We use the TPACK framework as a guiding structure to foster the types of knowledge necessary for effectively integrating generative AI tools into teaching. The prototype web application was integrated in a course design, serving as a sandbox for future and practicing teachers to experiment with AI-based learning scenarios and to design their own learning scenarios. The evaluation of the course design highlights the novel role of teachers as designers in an AI co-creation process and suggests a potential impact of generative AI on TPACK knowledge dimensions that extends beyond the mere integration of a new tool or technology. The challenges of uncertainty, interdisciplinarity and information overload can be addressed by empowering teachers to integrate AI through co-creation course designs and teacher support systems. [For the full proceedings, see ED665357.]