Tell Me How You Prompt. I’ll Tell You What Kind of Manager You Are.

“Show me your prompts. I’ll tell you who you really are as a leader.”

Not your title. Not your LinkedIn headline. Not your org chart.

Your prompts!

Because in the age of AI agents, management has become visible. Public. Observable. Measurable. And sometimes… uncomfortable.

A few weeks ago, 🦄 Peter Steinberger—a senior engineer—built, in a single weekend, what used to demand entire teams, endless roadmap rituals, and thousands of pull requests... and was rewarded with hundreds of thousands of GitHub stars.

One person. A constellation of agents. An amazing product.

No army. No bureaucracy. No theater... Just orchestration. This is not a story about coding. It is a story about management.

In the last couple of years, we have treated AI as an augmenter.

“Make me faster.” “Help me write code.” “Summarize this.”

That is Superman thinking. One stronger individual. One sharper sword. But Superman never saves the world alone. He needs the Justice League. Agents change the game because they let you build a league.

Not tools. Not copilots. Employees. AI employees. With roles. With responsibilities. With interfaces. With expectations. With management.

Here is the part most people miss. Great agent systems are not technical achievements. They are organizational achievements. They reflect three things:

• How well you understand users (desirability)
• How well you design systems (feasibility)
• How well you create value (viability)

Classic product thinking. Just… exposed.

When your agents fail, it is rarely “because AI is bad.” It is because:

• Your instructions are vague
• Your objectives conflict
• Your architecture is confused
• Your priorities are unclear

In other words: Your management is fuzzy. And AI does not forgive fuzziness. I often hear:

“I tried agents. It was bad.”

“I tried again. Still bad.”

“Third time. Meh.”

“Fourth time. It worked.”

That is not persistence. That is leadership. Because what happens in those iterations is not technical. It is reflective. You argue with yourself. You refine your intent. You sharpen your thinking. You clarify trade-offs. You confront blind spots. You become… a better manager. Before managing anyone else.

The best teams I see today use agents the way great leaders use people.

One agent for exploration. One for verification. One for quality. One for structure. One for risk. Like Anthropic’s “coworker” model. Not magic. Organization. You are building a team. Just faster. And more honest. Because the mirror never lies.

A former colleague told me something fascinating recently. Since working daily with agents, he writes better PRDs. Clearer specs. Sharper goals. Cleaner interfaces. Why?

Because he practiced thinking clearly with machines. And now humans benefit. His team moves faster. With less friction. With less waste. AI made him a better boss.

This is why the agentic era is not the era of engineers. It is the era of managers. Real ones. Not people with titles. People with craft. People who know how to:

Hire (design roles), train (refine prompts), coach (iterate), evaluate (observe outputs), correct (re-align), and reward (reinforce patterns).

With humans. And with machines. Same discipline. Same ethics. Same responsibility. Agents don’t replace leadership. They reveal it. They expose your thinking. They amplify your habits. They mirror your rigor. Or your chaos.

Tell me how you orchestrate your agents. I’ll tell you who you are as a builder.

If you’re curious about this shift, we discuss it in depth in the latest Silicon Carne episode (watch it here 👉 https://www.youtube.com/watch?v=_Hxg1tpkaBY&t=210s) on Stealth and the emerging agentic operating model.

The future of tech is not about better prompts.

It is about better managers.

Of humans. Of machines. Of meaning.

And of themselves.