Working with Technical Teams

The interface between leadership and technical teams is where many AI projects break down. Leaders who cannot communicate requirements clearly waste their team's time. Technical teams who cannot explain constraints clearly lose leadership support. Bridging this gap is a leadership skill.

Asking the Right Questions

Instead of asking "Can AI do this?" — which almost always gets a "yes" — ask better questions:

• "What accuracy can we expect on our data?" This forces a concrete, testable answer rather than theoretical possibility.
• "What data do we need, and do we have it?" Many AI proposals assume data exists that does not, or assume it is clean when it is messy.
• "What happens when the AI is wrong?" Every AI system produces errors. Understanding the failure mode is as important as understanding the success mode.
• "What is the simplest version we could test first?" This prevents over-engineering and gets to validation faster.

Communicating Requirements

Technical teams need specificity, not vision statements. Instead of "make our customer service smarter with AI," provide concrete scenarios: "When a customer emails about a billing issue, we need the system to categorize it, extract the account number, and draft a response that the agent reviews before sending."

Include examples of good and bad outcomes. Show the team five customer emails and describe exactly what the AI should produce for each. This is more valuable than any requirements document.

Reviewing Progress

Ask to see the AI working on real examples during every review. Do not accept accuracy percentages alone — numbers can hide important failures. If the AI is 95% accurate but fails on your most important customer segment, that matters more than the aggregate number.

Establish a shared vocabulary for quality. Terms like "accurate," "relevant," and "helpful" mean different things to different people. Define them explicitly for your project.

For a deeper understanding of how to interact with AI systems yourself, see the Prompt Engineering course on FreeAcademy.