It's notoriously difficulty to evaluate interfaces intended to support creative work – but as software creative tools proliferate, the importance of understanding whether and how these tools support user creativity continues to grow. In this talk, I discuss several related approaches to making sense of user interactions with creativity support tools. I focus in particular on how AI-supported evaluation methods can help us illuminate a design tool's expressive range; trace user trajectories through design space; and potentially even intervene to shape these trajectories while the interaction is still unfolding.