How you respond to AI reveals the relationship you already have with your work.

You may have heard the old parable. Three bricklayers are laying brick. You ask the first what he’s doing. Laying brick. You ask the second. Building a wall. You ask the third. Building a cathedral.

Same task. Same bricks. Three completely different worlds.

The story is about purpose — about how far you can see from where you’re standing. It’s been around a long time for a reason. It’s true.

Now imagine something new arrives on the job site.

A machine. It lays bricks automatically. Fast. High quality. It occasionally makes a mistake — small ones, the kind a trained eye would catch.

Watch what happens to each bricklayer.

The first was already defined by the task. The machine just said it out loud.

The second was already optimizing for the deliverable. The machine made her faster at it.

The third was already building toward something larger than himself. The machine gave him more time for the work only he can do.

If your answer stops at the task, the machine is a threat. If your answer stops at the project, it’s a useful tool — but you’ll miss its greatest gift.

If you can name what you’re actually trying to build in the world — if you can see the cathedral — the machine is the most powerful collaborator your generation has ever been handed.

The machine is here.
It lays bricks.
It makes mistakes.

What are you building?

I tried to quit sugar many times. Willpower. Cutting back. Making rules. Nothing worked. I’d last a few days, maybe a week, then find myself back where I started.

Then a mentor said something to me — I don’t remember her exact words. Something simple. Maybe just: “Why don’t you give up sugar?” But something in that moment shifted. Not my willpower. Not my strategy. My identity.

That was many years ago. I haven’t had refined sugar since.

Once I became someone who doesn’t eat sugar, checking labels and passing on the cookie just happened. I wasn’t fighting myself anymore. The doing flowed from the being.

What if you aren’t getting what you want — not because of your skill or your effort — but because you’re approaching it from the wrong direction?

Most people work from a simple assumption: if I have the right things — talent, connections, experience, money — I can do what I want, and eventually be who I want to become. Have → Do → Be.

A salesperson thinks: if I had a nicer car, I’d do more sales calls, and I’d be more successful. An executive thinks: once I have the right team in place, I’ll do the hard work, and the company will be what I want it to be.

This feels logical. It’s also backwards. And when you believe it, you give up without realizing it — because you’ll never have enough to start.

The sequence runs the other way. You start with BE — your identity, how you see yourself, who you’re committed to becoming. That identity drives DO — the actions that flow naturally from who you are. And consistent action produces HAVE — the outcomes, the results, the life.

Be → Do → Have. Identity first. Always.

If you take on the identity of a dictator, you start acting like one — and you’ll get the results a dictator gets. If you take on the identity of a collaborative leader, you seek input, build buy-in, and get different results. Neither is inherently right. Sometimes you need to be a dictator. If the building is on fire, you don’t ask for consensus on which exit to take. But you’re choosing the identity deliberately — not defaulting to it.

In my work at ITX, when I’m pushing ideas that are more doing than being, I feel the difference. The effort is higher. The results are thinner. When I connect to a clear way of being first — something like I am a leader who brings out the best in people to deliver ever-increasing value to our clients — the doing gets easier and the results get better. Less force. More flow.

The declaration matters. Naming your identity out loud — to yourself, to your team — changes how you act. Not because you’re performing it, but because identity shapes perception. You start noticing things you couldn’t see before. Options appear that weren’t visible when you were grinding at the level of tasks and goals.

So here’s the question worth sitting with: where in your life are you working at the level of doing or having — and wondering why nothing is changing?

What identity would make the actions you want to take feel natural instead of forced? Who do you need to be — so that the doing just follows?

You don’t need more willpower. You need a different starting point.

AI can write 1,000 lines of code in minutes. Congratulations. You’ve just moved the bottleneck.

For thirty years, the central constraint in software development was writing code. We spent decades trying to address it — frameworks, higher-level languages, agile processes, continuous deployment. We got better at it. We never removed it.

AI just removed it.

And if you keep trying to build software the same way, you’re going to end up with a very fast, very expensive process that produces the wrong things. The Theory of Constraints is clear: remove a bottleneck, and it moves somewhere else. The question is where it moved — and whether your organization is set up to deal with it.

The Old Process Was Designed Around the Wrong Problem

At its simplest, software development follows this arc: strategy → roadmap → backlog → iterative delivery. The bottleneck was always in that last step — writing the code. So we optimized everything around it.

AI doesn’t just write code quickly. It writes epics, user stories, test cases, and strategy documents quickly. It changes the entire cost structure of building software. If you can produce a working prototype in a few hours, the cost of throwing it out and trying something different is almost nothing.

That changes everything upstream.

There’s an old story about a crew working in a jungle. They’re chopping trees, clearing land, optimizing their tools to go faster. Then someone climbs a tree, looks out, and yells — wrong jungle.

Two New Constraints

With AI, two new bottlenecks emerge. Both are harder to solve than writing code.

1. Decision Making

The question is no longer “can we build it.” It’s “should we build it — and does it connect to what the business actually needs?”

Years ago, approximately 90% of feature requests Microsoft received for Word were already in the product. Users didn’t know. That caused Microsoft to rethink the entire UI and create the ribbon. When you can produce features quickly, the natural outcome is feature bloat, user confusion, and a maintenance nightmare. The constraint becomes strategic decision making.

And here’s the problem: our current processes were designed to make those decisions at the speed of humans writing code. Slowly. As the constraint shifts, those decision loops need to speed up dramatically. Otherwise you end up with a lot of resources building things without clear direction on what to build.

There’s good news too. With the cost of working prototypes dropping dramatically, we can now vet ideas with actual software in front of real users — not focus groups, not surveys. Real feedback on real things. But only if your decision-making process is fast enough to use it.

2. Governance

The second constraint is governance. And it’s more complex than it first appears.

When AI can write 1,000 lines of code in minutes, how do you verify it’s right? How do you verify it’s safe? How do you verify it’s actually aligned to what the business needs?

The obvious answer is “have humans review it.” But humans can’t review at AI speed. And having AI write its own test cases doesn’t solve it either — that’s just asking the same system to grade its own homework.

Governance at AI speed has to be a system, not a person.

And that system has more layers than most people think about. Automated scanning. AI-vs-AI validation using multiple models to check each other’s output. Architectural design that limits how much damage a bad piece of code can do. Identity and authorization — who approved this, at what level, with what credentials. Compliance — AI doesn’t inherently know your HIPAA obligations or your SOC2 requirements, that knowledge has to be built in explicitly. Data governance — what data is the AI touching, or potentially leaking.

And yes, there’s still human review. But applied strategically, not uniformly.

Adding copy to a marketing page? Automated scans, quick approval, move on. Changing the algorithm that calculates a financial metric driving millions of dollars in decisions? Human eyes, compliance review, full authorization trail.

Not everything needs the same checks. But everything needs some checks. And your highest-risk changes need more rigor than most teams currently apply even to their most careful work.

A New Development Process

The software process needs to fundamentally change to address these new constraints. Not incrementally — fundamentally.

That means rapid prototyping paired with sophisticated deployment processes and advanced governance. Build working prototypes fast to learn what to build. Once you’ve found something with strong user signals, shift gears — rigorous governance, security reviews, performance testing. Build once you know what matters, not the other way around.

How to Adapt

No one has fully figured this out yet. But a few things are clear.

Decision making and feedback loops need to dramatically improve. You need more sophisticated ways to understand what users actually need — not what they say they need in a survey, but what real behavior tells you. You need clear business goals that connect directly to product strategy. You need prototyping fast enough to test ideas before committing to them.

Investment needs to shift toward the new bottlenecks. Quality Assurance engineers who think like security researchers. Architects who design for verification, not just functionality. Product Managers who are less focused on writing specs and more focused on identifying the right problems to solve.

How projects are staffed will change. Do we still need coders in the same way? Quality Assurance engineers? Business analysts? A dedicated governance role? These aren’t rhetorical questions — the answers will shape how you hire, train, and staff projects for the next decade. The teams experimenting with these questions now will have a significant advantage over the ones waiting for the answers to become obvious.

The bottleneck moved. The teams that figure out where it went — and adapt accordingly — will build the right things. That’s always been the harder problem. Now it’s the only problem.

J.D. Salinger wrote one book.

One. His entire life. By most business definitions, he was one of the least productive authors who ever lived.

And yet Catcher in the Rye is considered one of the greatest novels ever written. Millions of copies sold. Decades of cultural impact. Still assigned in classrooms today.

Meanwhile, somewhere out there is an author who published 200 books. You’ve never heard of them.

So who was more productive?

The Problem With How We Measure Output

In business — especially in tech — we’ve often defaulted to quantity as a proxy for productivity. Lines of code. Story points. Tickets closed. Features shipped. The more, the better.

The problem is we’ve confused output with value.

You don’t know if something mattered until after the fact — sometimes long after.

IBM famously compensated engineers by lines of code. You know what they got? A lot of lines of code. Most of it noise.

The Spray-and-Pray Trap

There’s a competing school of thought: just do a lot of things, and eventually something will stick.

Throw enough at the wall. Ship fast. Iterate constantly. The startup mantra.

And look — there’s something to that. Iteration matters. Speed matters. But iteration without depth is just activity. And activity is not the same as progress.

The real question is: when do you iterate fast, and when do you go deep?

That distinction matters enormously — and most leaders never ask it.

Start With Real Pain

Here’s where Steve Jobs and J.D. Salinger actually had more in common than you’d think.

Jobs didn’t run focus groups. But he wasn’t operating on pure whimsy either. He identified a real, specific pain point: people’s relationship with technology was frustrating, clunky, and broken. Then he trusted his intuition to solve it his way.

Salinger did the same thing, just in a different domain. Post-war America had a cultural pain point — alienation, phoniness, the difficulty of growing up. He felt it deeply and wrote from that truth.

The formula isn’t magic. It’s this:

1. Identify a genuine pain worth solving
2. Trust your intuition on the solution — don’t let focus groups water it down
3. Ask if it’s economically viable after you’ve confirmed the pain is real

Most companies do this backwards. They start with what they can monetize, then go looking for problems that fit.

The Focus Group Trap

Here’s the thing about asking people what they want: they’ll tell you. And they’ll be wrong.

There’s a famous example where a focus group was asked: red toy or blue toy? Everyone said red. Then they left a bin of both toys out — and everyone took blue.

Henry Ford said it best: “If I had asked people what they wanted, they would have said faster horses.”

People can’t see solutions that don’t yet exist. They can only describe the world as it is.

A Toyota engineer spent over a year just living with American families before redesigning the Sienna minivan. Not asking. Watching. He observed where they put their cups, how they loaded groceries, how they moved through their days. He saw pain points families had accepted as just the way things were — because they didn’t know a better design was possible.

That’s not scalable. It’s not efficient. And it’s exactly why most companies don’t do it.

But it’s also why that minivan was a hit.

What AI Changes — And What It Doesn’t

Here’s what’s different now: building is cheap.

AI has dramatically compressed the cost of making things. In software especially, you can ship faster than ever. The constraint is no longer “can we build it.”

The constraint is now: should we build it, and what should we build?

That’s a decision problem. And decisions require something AI can’t fake: genuine empathy, human judgment, and real understanding of what people actually need.

AI can analyze patterns in data. It could theoretically watch ten thousand minivans and surface behavioral trends. But it can’t interpret what those patterns mean. It can’t know that an awkward reach for a cup holder is a pain point worth solving — or feel the significance of that moment the way a human engineer can.

The Depth Question

So back to Salinger.

Maybe the real lesson isn’t that he was unproductive. Maybe the lesson is that he was working on a different problem than most people.

He wasn’t optimizing for output. He was trying to capture something true.

And when you create something that captures a true human experience — a real pain, a real feeling, something people recognize in their bones — that’s when the impact outlasts the effort.

That’s not anti-productivity. That’s the highest form of it.

The question worth asking isn’t “how much did we produce today?”

It’s: did we solve something that actually mattered?

AI can process anything. It can’t judge anything. That gap is where your value lives.

In 1893, a young lawyer arrived in India after years working in South Africa. The British Empire still controlled the subcontinent. The pressure on Mahatma Gandhi to act — to lead, to speak, to organize — was immediate and enormous.

He meditated for a year.

Not publicly. Not visibly. He traveled, observed, listened. He did what looked, to most observers, like nothing. And then, when he felt the moment was right, he acted. What followed dismantled one of the most powerful empires in history.

If Gandhi had been an employee, he’d have been fired.

That tension — between what looks like productivity and what actually creates value — is at the heart of what we believe is the most important leadership skill of the AI era.

Not knowledge. Not speed. Not output.

Judgment.

But judgment isn’t one skill. It’s a cluster of capabilities — some inner, some relational, some about execution. And the research on what separates great decision-makers from average ones points to a surprisingly consistent set of qualities.

The Foundation: Being Anchored and Centered

Every other judgment skill rests on this one.

Gandhi didn’t meditate for a year because he was avoiding the problem. He did it because he understood something most leaders never learn: decisions made from urgency, ego, or external pressure are almost always wrong. Not immediately. But eventually.

The leaders we most respect — across history and across industries — share a quality that’s hard to name but easy to recognize. They know who they are. They’re not threatened by opposition. They don’t need to look good in the moment. They can hold enormous pressure without being moved by it.

Rudyard Kipling tried to capture it in If: “If you can keep your head when all about you are losing theirs… if you can wait and not be tired by waiting…”

This isn’t passivity. It’s the opposite. It’s a stable enough sense of self that you can take in hard information, sit with uncertainty, and act from clarity rather than reaction.

Steve Jobs had it — and lost it — and found it again. In his first run at Apple, his vision was extraordinary. His judgment about what people needed from technology was decades ahead. But his ego, his impatience, his inability to see the good in the people around him — those got him fired from the company he founded.

He came back different. More anchored. Less in his own way. What followed was the greatest product run in the history of consumer technology.

The pattern repeats. When leaders operate from this centered place — patient, clear, not needing to be right — they tend to make better calls. When they don’t, even brilliant people cause damage.

The judgment skill: Know yourself. Know your values. Make decisions from there, not from fear, ego, or the pressure of the moment.

Inner Skills: How You Think

Being anchored creates the conditions for clear thinking. But clear thinking still requires practice.

See the whole system, not just the problem in front of you.

Most decisions feel like isolated choices. They’re not. Every call you make sits inside a larger system — of people, incentives, history, and unintended consequences. Judgment requires zooming out far enough to see what you’re actually deciding.

Gandhi didn’t just see a labor dispute or a legal injustice. He saw an entire colonial structure and understood the leverage point that could move it. Jobs didn’t just see a phone — he saw the relationship between humans and technology and decided it needed to be rebuilt from scratch.

Holistic thinking isn’t about having all the answers. It’s about asking: what’s the larger thing I’m actually solving for?

Know when to act and when to wait.

This one cuts against almost everything Western business culture rewards. We celebrate speed. We promote decisiveness. We’re suspicious of people who say “not yet.” But timing is a judgment call. Acting too early collapses options. Acting too late misses them. The skill is reading when the moment is right — and being patient enough to wait for it.

Pressure-test your assumptions — especially the ones you like.

AI will validate your thinking. So will most of the people around you, if you hold power. Judgment requires actively looking for what’s wrong with your best ideas. Not to kill them — to make them stronger. The question to ask before any major decision: what would have to be true for this to fail?

Understand the difference between extrinsic and intrinsic value.

We measure what we can count. But the most important things are often the hardest to quantify.

J.D. Salinger wrote one book. By most business definitions, he was one of the least productive authors who ever lived. And yet Catcher in the Rye is still assigned in classrooms today, decades after his death. He wasn’t optimizing for output. He was trying to capture something true.

The parent raising a child. The teacher who changes a student’s trajectory. The manager who believes in someone before they believe in themselves. All high value. All poorly priced by markets.

Judgment requires asking not just “what’s the return?” but “what’s actually worth creating?” Those are different questions. They lead to different decisions.

Relational Skills: How You Think With Others

Great judgment is rarely solitary.

Know whose — and what — perspective is missing.

Every decision has a blind spot. Usually it’s the voice that isn’t in the room — the person without power, the discipline that wasn’t invited, the stakeholder whose concern nobody thought to surface.

Lincoln built his entire cabinet from rivals. Not because he liked them. Because he understood that the perspectives most likely to challenge his thinking were the ones he most needed. Before any significant decision, ask: who isn’t here? What viewpoint haven’t we considered?

Read what’s unsaid.

Data tells you what happened. It doesn’t tell you why someone went quiet in the meeting, or what the VP is sitting on that she hasn’t said yet, or whether the team has actually bought in or is just complying. That takes presence. Attention. The willingness to slow down and notice.

Don’t be moved by others’ urgency.

Pressure is contagious. When everyone around you is anxious, it takes real anchoring to stay clear. But borrowed urgency produces bad decisions. The question isn’t “how fast does everyone want this?” It’s “when does this actually need to happen?”

See the good in others.

This one sounds soft. It isn’t. Leaders who assume incompetence or bad intent get less information, less honesty, and worse outcomes. Leaders who genuinely believe in the people around them — who extend trust before it’s fully earned — tend to unlock capability that otherwise stays hidden. Judgment about people requires seeing clearly — and at times seeing things they can’t yet see in themselves — not cynically.

Execution Skills: How You Move

All of this inner and relational work has to land somewhere. Judgment that isn’t acted on is just observation.

Deeply understand the people you serve.

Before you decide what to build, make, or change — understand who it’s for and what they actually need. Not what they say they need. What they need.

A Toyota engineer spent over a year living with American families before redesigning the Sienna minivan. Not surveying them. Living with them. Watching where they put their cups. How they loaded groceries. How they moved through their days. He saw pain points families had accepted as just the way things were — because they didn’t know a different design was possible.

Define the outcome, then work backwards.

Activity is easy to generate. Outcomes are harder to specify. Before any initiative, get precise: what does success actually look like for the people we’re serving? Then ask whether what you’re doing moves toward that — or just keeps people busy.

Run experiments. Test hypotheses.

You won’t always know. Good judgment includes knowing when you don’t know — and designing small, fast tests to learn before you commit fully. The goal isn’t to be right. It’s to get right, faster.

Make the call.

All of this — the anchoring, the holistic thinking, the relational work, the understanding of who you’re serving — ultimately has to result in a decision. Judgment isn’t endless analysis. At some point, with imperfect information and real stakes, you commit. And then you own the outcome. Regardless of what the data said. Regardless of what AI recommended.

Nobody Gets This Perfect

That’s the point.

Jobs succeeded when he operated from these principles. He failed — spectacularly, publicly — when he didn’t. Gandhi waited a year and changed history. Most of us can’t wait a week without feeling like we’re falling behind.

The goal isn’t perfection. It’s direction.

In an AI world, the leaders who create the most value won’t be the ones who know the most, generate the most, or act the fastest. They’ll be the ones who think most clearly, see most completely, understand the people they serve most deeply — and can make the call when it matters.

That’s judgment.
And unlike knowledge, it can’t be commoditized.

It has to be developed.

A man stood in a Washington, D.C. subway station and played a 300-year-old Stradivarius for about 45 minutes. He was one of the finest violinists alive. More than a thousand people walked past him. Seven stopped. He made about $32.

Move that exact performance into a concert hall and it sells out at a hundred dollars a seat.

Nearly twenty years later, someone posted a Monet on X — a real one, a cropped detail from his late Water Lilies — and labeled it “Made with AI.” Then asked the internet a simple question: describe, in detail, what makes this inferior to a real Monet.

The internet obliged. Confidently. Specifically.

People diagnosed the dead color. The lifeless brushwork. The missing soul. The “AI mush.” Self-described experts wrote paragraphs. One wrote 850 words.

It was Monet.

When the reveal landed, some people laughed. Some quietly deleted their replies. And some — this is the part worth sitting with — dug in. The Monet was “overrated anyway.” The work hadn’t changed. Their verdict refused to.

Here is the question everyone is asking: can you still tell what’s real?

The subway and the Monet are the same experiment run twice. Great work, wrong frame, dismissed. We like to believe we evaluate the thing in front of us. Mostly we evaluate its context and call the result taste.

That shortcut was always there. It mostly worked. If something hung in a museum, it had already survived a thousand filters before it reached your eye. The frame was a decent proxy for the quality. So we stopped checking the quality and started trusting the frame.

Efficient. A little lazy. Fine — until the frame stopped tracking the quality.

That is what AI actually did. Not “make fake art.” It broke the proxy.

And it broke it in both directions at once.

On one side: AI makes the frame look expensive. The average output now arrives polished, fluent, confident, formatted — wearing every signal we were trained to read as good. Mediocre work in a beautiful frame. We overcredit it.

On the other side: the word “AI,” attached to anything, has become its own subway station. It tells the eye what to find before the eye has looked. A Monet behind that label gets shredded by people who would have wept in front of it in Munich.

Same painting. Different sticker. Opposite verdict.

We have spent two years arguing about whether machines can make real art. Wrong fight. The thing that actually broke is older and more uncomfortable: the shortcut we have all been using instead of looking no longer points anywhere reliable.

This isn’t really about art. It’s about every judgment you make by reading the frame instead of the work.

The résumé from the right company. The idea from the most senior person in the room. The proposal that arrives beautifully formatted. The plan that sounds polished in the meeting. You are surrounded by frames that used to be honest proxies and are quietly turning into costumes.

The people who got the Monet right weren’t smarter. They did one thing the crowd didn’t. They looked at the painting before they looked at the label.

That used to be optional.