The previous unit established that AI adoption is fundamentally a people challenge—and that diagnosing cultural barriers, leading with empathy, and demanding a clear end game are the first moves any people leader should make. But once the organization is leaning in, you face a different problem: where do you actually start building? The conversation turned to this exact question, and the answer was refreshingly disciplined—not a sprawling AI roadmap, but a focused method for establishing a common language, surfacing the right problems, and letting early wins generate their own momentum.

Before diving into tools, the focus was on education and alignment. The team began by building a common vocabulary—reading the same books and learning together—to ensure everyone was speaking the same language. Only then did the approach shift to bottom-up problem discovery. The team asked two deceptively simple questions: "What bothers you the most about your job today?" and "What could technology do to fix that, help that, change that?"

The result was a surge of a thousand use cases. To avoid being "whipsawed into try this, do that," the team used a value-versus-effort rubric to sort submissions:

• High value but hard to build: Worth pursuing deliberately.
• Easy to do and super important: "Just get on with it."
• Not fully baked or governance risks: Deferred until a stronger case could be made.

This rubric narrowed a thousand ideas down to 28 actionable pilots, providing the focus needed to move from experimentation to execution.