It Just Doesn't Get It
A common complaint is floating through the tech industry as teams adopt AI more deeply into their workflows. It's a whisper of frustration: "I have to repeat myself constantly," "It makes basic mistakes," "It just doesn't get what I'm asking for." People are talking about AI, but they sound exactly like someone complaining about a junior employee.
After integrating AI agents into my own teams and workflows, a powerful and slightly uncomfortable truth has become clear. Once you get past the technical basics, you realize that an AI is a mirror. It reflects, with brutal honesty, the quality of your own instructions and the clarity of your own thinking.
The skills that make someone effective at working with AI are the same skills that make them an effective leader and manager of people. If you're bad at working with AI, it's a strong signal that you might also be bad at delegating to and empowering your team.
The Parallels Are Uncanny
This isn't about technical proficiency. It's about the soft skills of delegation, communication, and strategic thinking. The parallels between managing a talented human and steering a powerful AI are striking.
1. Providing Context vs. Giving Vague Instructions: A weak manager gives a team member a vague, unhelpful task: "Can you look into our churn problem?" The team member is left to guess at the scope, the goal, and what success looks like.
An effective leader provides context: "Our churn rate for mid-market accounts has spiked by 15% this quarter. We need to understand the root cause. Please analyze the last 90 days of support tickets and user feedback for that segment and come back with a summary of the top three complaint categories by Friday."
This is precisely how one must work with AI. A vague prompt like "Analyze our churn" will yield a generic, useless response. A high-agency prompt provides the same level of context: "You are a senior data analyst. I am providing you with a CSV of the last 90 days of churn data. Your task is to identify the top three reasons for churn specifically for customers on the 'Pro' plan and present your findings as a markdown table."
2. Iterating with Feedback vs. Micromanaging the Process: A micromanager dictates every single step of a task, robbing their team of autonomy and the ability to apply their own expertise. A great leader sets a clear goal, trusts their team to produce a first draft, and then provides high-level feedback to guide the next iteration.
Working with AI is the same. Ineffective users try to control the AI's output at the sentence level, getting frustrated when it doesn't write the exact phrase they had in their head. Effective users treat the AI as a creative partner. They ask it to generate a first draft of a product spec, then provide iterative feedback: "This is a good start, but make the tone more formal," "Expand on the section about security requirements," or "Reframe this from the user's perspective." It is a collaborative loop, not a dictation.
3. Defining Outcomes vs. Assigning Tasks: The least effective leaders manage by assigning a checklist of tasks. The most effective leaders manage by defining a desired outcome and empowering their team to figure out the best way to achieve it. They focus on the "why," not just the "what."
This is the most advanced skill in working with AI. Instead of giving an agent a series of small, prescriptive steps, you give it a high-level, outcome-oriented goal. "You are the CPO of a B2B SaaS company. Draft a comprehensive go-to-market strategy for a new AI-powered analytics feature targeting enterprise customers. Include sections for pricing, positioning, and a Q3 launch plan."
This approach leverages the AI's ability to reason and plan, turning it from a simple tool into a strategic partner. It requires a leader who is clear on the ultimate objective, not just the immediate next step.
A New Litmus Test for Leadership
This insight has changed how I think about hiring and developing leaders. The ability to work effectively with AI is becoming a powerful proxy for core leadership competencies. It's a test of a person's ability to communicate with clarity, to think strategically, and to delegate effectively.
This isn't about blaming people who find AI frustrating. It's an observation that the very act of trying to get better at using AI forces you to get better at the foundational skills of modern leadership. It forces you to be more explicit about your goals, more thoughtful about the context you provide, and more disciplined in your feedback.
Conclusion
AI is more than a tool; it's a catalyst for professional development. The frustration of a poorly understood prompt is a signal—a mirror reflecting a gap in communication or a lack of strategic clarity. The leaders who will thrive in this new era are the ones who see this signal not as a failure of the AI, but as an opportunity to sharpen their own thinking.