Google researchers introduced Agile Deliberation, a framework that treats the creation of subjective AI classifiers as an iterative conversation, turning the AI system into a collaborative partner that helps users deliberate and sharpen their own definitions rather than requiring humans to act as static data labelers. The work addresses the assumption that people labeling examples have a clear and stable definition of a category before they begin—an assumption that holds for objective categories like a stop sign or a golden retriever but not for subjective "gray areas" such as "healthy food," "inspiring artwork," or content moderation, where boundaries can shift with cultural context, personal philosophies, or evolving operational policies. Because standard, static AI benchmarks can't capture subjective concepts, the researchers bypassed simulated testing and evaluated Agile Deliberation across 18 live, 1.5-hour sessions with human participants, finding it drastically reduced user fatigue and anxiety by replacing manual prompt engineering with direct feedback on borderline images, resulting in more consistent, higher-performing AI models that outperform standard baselines. Agile Deliberation splits the task into two structured stages, concept scoping and concept iteration, transforming a fuzzy idea into a highly precise instructional blueprint for an AI model. The underlying code and algorithms have been open-sourced on GitHub.