Mike will demonstrate how to train an LLM-as-a-Judge and then use it to optimize fuzzy generative tasks — where evaluations are informal, subjective, and difficult to measure. This approach unlocks new ways of making AI agents more dependable in creative and ambiguous domains.