We're About to Run Out of Phil Jacksons

Phil Jackson never made an All-Star team. Spent most of his playing career on the bench, career average of 6.7 points per game, highlight reel you could watch during a bathroom break. Then he became the greatest coach in NBA history — eleven championships, the architect of two separate dynasties.

I grew up in Chicago watching all three versions of how leaders get made, and I'm worried we're about to lose two of them.

The Three Paths to the Coaching Chair

Steve Kerr was never Michael Jordan. He was the guy who stood in the corner and hit threes when the defense collapsed. No poster dunks, no signature shoes. But he was in the room. He felt what it was like when the play broke down, when the ball stuck, when the chemistry went cold. Now he's one of the best coaches in basketball — not because he copied Phil Jackson's system, but because he built his own from ten thousand hours of playing the game.

Greg Maddux came back to the Cubs at the end of his career, not because his 42-year-old arm was special, but because of what happened in the third inning. A young pitcher would start to unravel — velocity dropping, location drifting — and Maddux was already walking to the mound before the pitching coach could get out of the dugout. He'd been that struggling pitcher ten thousand times. He knew the feeling in the body before the kid could find words for it.

Phil Jackson couldn't demonstrate the fadeaway. Couldn't show you how to post up or play help-side defense at an elite level. But he could build the system that made the fadeaway happen at exactly the right moment. Triangle offense. Zen psychology. The architecture of peak performance. He understood the game at a level the players themselves couldn't articulate because he'd spent decades watching from the inside — first as a marginal player, then as an assistant coach absorbing every pattern.

Three paths: learning by doing, learning by struggling, learning by seeing the whole system.

The Factory Is Closing

Here's what I'm stuck on: if AI agents become the players, where do the future Steve Kerrs and Greg Madduxes come from?

I was on a call last week with a finance team implementing AI agents to handle their month-end close process. Sophisticated stuff — the agents catch reconciliation errors, flag anomalies, route exceptions. The CFO was thrilled. Cycle time cut in half. His senior accountants could "focus on strategic work."

I asked him: "Where are your future controllers going to learn to spot a bad reconciliation?"

Long pause.

The judgment that makes a great controller — the pattern recognition, the smell test, the ability to look at a balance sheet and know something's off before you can prove it — comes from reps. From staring at reconciliations until the anomalies jump off the page. From making the mistake, feeling the consequences, building the scar tissue.

If AI removes the playing field, we lose the factory that built expertise.

Steve Kerr learned to lead by feeling the game from the inside. Greg Maddux learned to lead by having been the struggling player. They didn't learn from a textbook or a dashboard. They learned from reps.

The System-Thinker Problem

Maybe the future is all Phil Jacksons — people who never played but can see the whole system, orchestrate the pieces, build the architecture.

But here's the uncomfortable part: Jackson had decades of watching from the inside. He was a marginal NBA player for eleven years. He was an assistant coach for another decade before he got the top job. He may not have had the skills of his players, but he had the context. He knew what the game felt like from the court.

What does the shortcut version of that produce?

I've watched three major technology disruptions up close. Client-server to internet. On-premise to cloud. Manual to AI. Every single time, we assumed the next generation of leaders would figure it out. And every single time, we lost a cohort of people who would have been great — because we automated away their training ground before they got their reps.

The consulting firms that moved juniors straight to "strategic work" instead of grinding through Excel models? They're struggling to promote anyone to partner now because nobody learned to smell a bad assumption. The banks that automated trade execution? They're scrambling to find people who understand market structure because nobody spent years on the trading floor anymore.

We keep automating the base of the pyramid and then wondering why the top is unstable.

The Question Nobody's Asking

Here's what I can't resolve, and I've been doing this work for twenty-five years:

If AI agents do the month-end close, who learns to be a controller?

If AI agents write the code, who learns to be a software architect?

If AI agents handle the discovery process, who learns to be a litigator?

Maybe we don't need controllers and architects and litigators anymore. Maybe the future belongs entirely to the system-thinkers, the orchestrators, the people who manage the AI.

But I've never met a great system-thinker who didn't earn it through reps. Even Phil Jackson played in the NBA. Even Bill Belichick spent years as a position coach breaking down film. They saw the system from the inside before they could architect it from above.

What happens when we remove the inside?

The Uncomfortable Middle

Look, I'm not arguing we should keep people doing work AI can do better. I'm not nostalgic for ten-hour reconciliation marathons or manual code reviews. The efficiency gains are real. The technology works.

But I am arguing that we're making a trade we don't fully understand yet.

We're gaining productivity today by consuming the seed corn that grows expertise tomorrow. And unlike previous disruptions — where the new technology created new training grounds — I'm not sure what the equivalent is here. When bank tellers became obsolete, they moved to relationship management and learned different skills. When factory workers were automated, they moved to machine operation and maintenance.

But when AI agents do the cognitive work — the analysis, the pattern recognition, the judgment — what's the training ground? Watching the AI? Reviewing its output? That's not the same as doing the work, feeling the failure, building the intuition.

What to Do Monday Morning

I don't have a clean answer. But I know the wrong answer: pretending this resolves itself.

If you're implementing AI agents in your organization, here are three questions worth sitting with:

1. What skills are we automating that used to be the training ground for the next level? Make the list explicit. If your AI handles month-end close, write down what your future controllers won't learn.

2. Where are the new reps? If the old playing field is gone, what's the new one? How do people get the pattern recognition, the judgment, the scar tissue? Simulation? Rotation programs? Something we haven't invented yet?

3. Who's tracking the expertise pipeline? Not just the efficiency gains. Someone needs to own the uncomfortable question: are we still producing people capable of leading this function in ten years?

I watched Phil Jackson win eleven championships. But I also watched the NBA nearly run out of great coaches in the early 2000s because the development pipeline broke. They fixed it — G League coaching roles, expanded assistant positions, more pathways. They had to build the factory deliberately after they realized it was closing.

We're at that moment now with AI. The efficiency is here. The productivity is real. But if we don't build the new factory deliberately, we're going to wake up in ten years wondering where all the Phil Jacksons went.

And unlike basketball, we can't just import them from Europe.

What's your organization's plan for growing the people who can eventually see the whole game? Because the AI agents can't answer that question. Only you can.