Coaching in the era of AI tools
When AI coding tools generate experimental code, the time needed to test ideas shrinks dramatically. What once took two days now takes two hours. That’s a game-changer. But every time we remove a bottleneck, another one emerges. Faster iteration shifts the bottleneck from execution to ideation. Each experiment is a learning opportunity: Do users prefer a drop-down menu? Does regularization reduce overfitting? Does the new messaging increase engagement?
The limiting factor is not how fast we can test ideas – it’s how quickly you can get more good ideas. Each result leads to a decision and a next step: What did we learn? What should we try next?
The way we reward our teams didn’t change (nor did the career progress):
If an engineer proposes a thoughtful idea that fails, reward the learning.
If the idea works, reward the impact.
If the idea was weak from the start, the lack of reward teaches the value of quality thinking.
AI can help with brainstorming, too. The bottleneck is knowing how to express a problem clearly, and recognizing a good idea when you see it.
Our coaching now must emphasize clear communication, especially diagnosing and articulating problems crisply – skills that matter most when the stakes are high.
Live skills also matter more than ever. AI can polish your deck or draft your pitch. But it won’t be in the room to convince the CEO. Can you defend your idea? Can you win support in the moment?
That’s what we should train for.