Skills Build Teams. Judgment Builds Enterprises.

Kathleen Walch is right. In “Agility Is the New IT Currency,” she argues that CIOs who treat AI skills development as a core strategic function rather than an HR afterthought will outrun those who don’t. To keep pace with AI-driven change, organizations need continuous learning systems that map current capabilities, create role-based learning paths, reinforce new skills through practice, and give teams safe ways to experiment.

Her roadmap is practical, timely, and aimed squarely at the right problem.

But there’s a level above it that the roadmap doesn’t fully reach. And that level is where enterprise agility either compounds or collapses.

Skills build teams. Judgment builds enterprises.

In AI-accelerated environments, that distinction matters because enterprise agility depends on adaptive capacity: the ability to adjust under pressure without losing coherence. The difference becomes visible in moments when knowledge is not the problem.

May 10, 1996. Rob Hall stood above 28,000 feet on Everest with his client Doug Hansen. Hall was the most experienced high-altitude guide of his generation. Five successful summits. A reputation built on discipline and sound judgment.

He had declared a turnaround threshold of 2 p.m. The rule was non-negotiable.

He pushed past it anyway.

Competitive pressure, emotional obligation to a client who had failed to summit the year before, and summit fever did the rest. The storm hit. Both men died.

Hall didn’t lack skills, knowledge, or experience. What failed was judgment — specifically, the reflective discipline that keeps experienced people from overriding what they know to be true under pressure.

The most dangerous decisions aren’t made in ignorance. They’re made despite knowledge.

That distinction matters enormously as AI accelerates decision velocity across every level of the enterprise.

Knowledge is not capability

Kathleen’s framework — skills mapping, modular learning paths, and continuous reinforcement — builds knowledge effectively. That work is necessary, but on its own, it isn’t enough.

Capability is a different construct. It emerges from knowledge, experience, and judgment working together under real conditions. Most AI upskilling programs invest heavily in knowledge. They assume experience accumulates naturally. They rarely address judgment directly.

But judgment doesn’t develop through experience alone. It develops through reflection on consequence-bearing experience — decisions made under real exposure, with real stakes, examined afterward with honest analysis.

A team member who completes an AI literacy program has more knowledge. A leader who makes a difficult call, owns the outcome, and reflects carefully on the reasoning behind it develops judgment.

Those are not the same developmental events.

Enterprise agility requires judgment distributed across the system, not concentrated at the top. An enterprise that scales skills without scaling judgment builds execution capacity. It does not build adaptive capacity.

When conditions shift — and in AI-accelerated environments they shift faster than training programs can track — execution capacity without judgment produces faster drift, not better decisions.

Enterprises rarely lose adaptability all at once. They drift into dependence gradually — one outsourced judgment call to AI at a time.

The OODA loop and the augmentation question

John Boyd designed the OODA loop — Observe, Orient, Decide, Act — for fighter pilots making split-second decisions under life-or-death pressure. Boyd’s ideas became influential across military strategy, business, and decision-making theory.

The insight that made it powerful wasn’t speed. It was orientation. Boyd’s ideas became influential across military strategy, business, and decision-making theory. The pilot who updates their mental model faster than their opponent doesn’t just react more quickly. They see a different situation entirely.

AI accelerates the Observe step dramatically: more data, faster synthesis, patterns surfaced in seconds that would take analysts days. That’s real and valuable.

The question is what happens at the Orient step — where mental models update, pattern recognition sharpens, and judgment lives.

This is where the distinction between automation and augmentation becomes critical.

It isn’t primarily a technology question. It’s a developmental one.

Automation takes over the Orient step. The AI does the sense-making. The human receives a recommendation and acts. Decisions get faster, but human judgment gets less exercise. Over time, the capacity to recognize when the system is wrong quietly erodes.

Augmentation keeps humans responsible for orientation. AI surfaces patterns, flags anomalies, and accelerates analysis. Humans interrogate the recommendation, apply contextual judgment, and own the decision.

The loop gets faster.

The human gets sharper.

Simultaneously.

In “Artificial Organizations,” Barry O’Reilly puts it directly: in the age of AI, experience only compounds if it is augmented. Otherwise, it calcifies.

That’s not a warning about technology. It’s a design requirement.

The human-AI arrangement must be deliberately built to keep judgment developmental, or organizations become faster at producing answers and weaker at knowing whether those answers are sound.

Anthropic’s 2026 survey of 81,000 AI users found that 16% already identified cognitive atrophy as a firsthand concern. Educators reported it at nearly three times that rate.

The loop that should strengthen judgment can erode it instead.

The difference is whether the arrangement augments humans or replaces them.

Two systems for decision quality

Daniel Kahneman’s dual-process framework gives us a useful starting point for thinking about decision quality.  First, we need to understand the two types of decisions.

System 1 decisions are fast, pattern-based, and often driven by recognizable signals, and data-dependent. System 2 decisions are slower, more reflective, qualitative, and uncertainty-heavy.

Most AI upskilling programs build toward stronger System 1 performance — faster analysis, better optimization, and improved data-driven execution. That’s appropriate for exploitation-dominant environments where evidence standards are clear and the future behaves enough like the past for historical patterns to matter.

But enterprise agility matters most when the future stops behaving like the past.

System 2 decisions — exploratory, qualitative, high-uncertainty decisions — require something different.

Not gut instinct, but pattern recognition earned through consequence-bearing experience and disciplined reflection.

Phil Knight’s early Nike bets looked reckless by traditional process quality standards. He ordered inventory without the cash to cover it and made decisions long before evidence stabilized.

But Knight had spent years handling shoes, talking to runners, and watching the market move. The pattern recognition that preceded articulable evidence wasn’t recklessness. It was earned perception — the right to make a call before the evidence catches up.

AI upskilling programs that build only System 1 capability leave leaders poorly equipped for the decisions that matter most.

Enterprise agility requires both systems — and the judgment to know which one a given decision demands.

Adaptive capacity builds enterprise agility

Kathleen’s roadmap builds toward high-performing IT teams. That’s the right starting point.

But enterprise agility requires something broader: adaptive capacity.

Adaptive capacity in leadership is the ability to sense change, make sound judgments under uncertainty, build capability in others, and adjust direction without losing coherence or purpose.

Enterprise agility is the organizational expression of that capability at scale — the ability to adapt direction, reconfigure how work happens, and respond to disruption without fragmenting the enterprise.

It depends on balancing tensions that never fully disappear: speed and reflection, autonomy and alignment, innovation and operational discipline.

Individual skills improve execution capacity.

Distributed judgment creates adaptive capacity.

One helps teams perform.

The other helps enterprises evolve.

O’Reilly’s 80/80 inversion captures the stakes precisely: 80% of leadership time goes to administration. 80% of leadership value comes from judgment.

AI should flip that ratio — freeing leaders from administrative burden so they can focus on what only humans can do.

But the promise only holds if the reclaimed time goes into consequence-bearing decisions, honest reflection, and the deliberate development of judgment across the enterprise.

Otherwise, organizations don’t become more adaptive.

They become operationally faster while strategically weaker.

The AI era will not primarily reward organizations that generate the most answers.

It will reward those that preserve the human judgment needed to discern which answers are worth trusting.

Tags: Agile | Artificial Intelligence | Upskilling | Teams

Related Insights

You May Also Like