This research introduces WHAM, a generative AI model designed to support creative ideation in game development. The authors identified key capabilities—consistency, diversity, and persistency—needed for AI to effectively assist creatives through divergent thinking and iterative practice. They trained WHAM on gameplay data, demonstrating its ability to generate consistent and diverse gameplay sequences and incorporate user modifications. The model's architecture, training process, and evaluation metrics are detailed, along with a concept prototype called the WHAM Demonstrator. The work emphasizes the importance of aligning AI development with the specific needs of human creatives, and also underscores the need for generative AI models to broaden creativity support to other domains.