AI Is a Stress Test. Most Companies Are Failing It.

When ChatGPT shipped, the prediction was uniform: consulting was over.

Why pay McKinsey when a model could draft your strategy deck? Why hire Deloitte when an agent could build your playbook? The thesis was clean. AI commoditizes knowledge work. Consultants sell knowledge work. Math is math.

The math went the other way.

Consulting demand for AI-related engagements is up. Boutique firms specializing in AI adoption are at capacity. The Big Four have reorganized entire practice areas around it. Internal transformation offices are hiring external help, not replacing it.

Something interesting happened on the way to the funeral. AI didn't replace consultants. It revealed why companies needed them in the first place.

The Bottleneck Was Never the Work

The naive view of consulting is that consultants do work clients can't do. That was never quite true. Consultants do work clients can't get done. The distinction is small in language and enormous in practice — and AI makes it impossible to ignore.

Most large organizations don't have a capability problem. They have an integration problem. They have analysts who can analyze, engineers who can build, marketers who can market. What they don't have is a system that turns those outputs into aligned decisions.

AI doesn't fix that. AI makes it worse.

When you drop a high-output tool into a low-coordination system, you don't get more progress. You get more artifacts. Each function uses AI to produce more of what it was already producing — more decks, more code, more research, more content. None of it integrates. None of it moves a decision faster.

The bottleneck has moved. It used to sit at the work itself. Now it sits at the seams between the work.

AI Is a Faithful Amplifier of Whatever You Already Had

This is the uncomfortable executive observation. AI does not change your culture. It scales it.

If your organization was already misaligned, AI scales the misalignment. If your data was fragmented, AI generates confident answers from fragments. If your incentives rewarded visible output over real outcomes, AI gives every team a way to look busier than ever while moving in the same direction they were already moving.

AI doesn't introduce dysfunction. It exposes dysfunction that was hidden by the cost of producing output. When output was expensive, activity could be mistaken for strategy. When output is free, the gap shows.

Companies that were quietly broken are now loudly broken. The model didn't change them. It just made them faster.

The Illusion of Progress

There is a pattern that shows up everywhere AI has been deployed seriously, and executives need to name it.

Teams optimize for what is visible. AI makes more things visible. So teams produce more — more pull requests merged, more research summaries circulated, more drafts generated, more dashboards refreshed. The visible metrics improve. The actual outcomes do not.

I have watched product organizations triple their experimentation throughput and not move a single business KPI. I have watched marketing teams generate ten times the content and lose share. I have watched engineering organizations halve their cycle time and ship the same number of meaningful features.

This is not a tooling failure. It is a measurement failure. Companies were already measuring the wrong things. AI just helps them hit those wrong targets faster.

The Consulting Paradox

So why are consultants busier than ever?

Because the AI vendors cannot do the part that actually matters. Vendors sell models, platforms, and integrations. None of that is transformation. Transformation is a political, organizational, and narrative act. It requires someone in the room who can name uncomfortable truths to a CEO without losing their job the next quarter.

That has always been part of what executives quietly bought from consultants. It is more valuable now, not less.

There is a second reason. Executives running AI initiatives need three things their internal teams cannot reliably provide: external validation that the direction is correct, a coherent narrative that aligns the board and the organization, and an accountability buffer when things go sideways.

These are not deliverables. They are functions. AI does not produce them. It increases the demand for them.

The Job Is Changing

The honest version: traditional consulting was already overdue for a reckoning. Decks built from secondhand research, frameworks lifted from prior engagements, recommendations the client could have written themselves. AI is killing that work, and good.

What is left is sharper and more valuable.

The new role is closer to a systems integrator of reality. Less "here is the answer," more "here is why your decisions are not converging." Less "here is the framework," more "here is the redesign of how you actually decide what to build, what to fund, and what to kill."

The deliverable is no longer the recommendation. The deliverable is a functioning decision system.

The consultants who matter in this era are the ones who can walk into an organization, see the wiring, and rewire it. The ones who treat AI as a diagnostic instrument first and a productivity tool second. The ones who can tell an executive that the model is working perfectly and the company is still failing — and explain exactly why.

What Executives Should Take From This

If your AI rollout is producing more output and the same outcomes, you do not have an AI problem. You have an organizational problem that AI is now making visible.

The instinct is to respond with more tooling, more pilots, more agents. That is the wrong move. The right move is to take the diagnostic seriously.

Where is the work piling up between functions? Which decisions take six weeks and should take six days? Which metrics are improving while the business is not? Which teams are producing more and influencing less?

Those questions are not answered by a model. They are answered by leadership willing to look at the system honestly — usually with help from someone whose job is to tell the truth.

AI doesn't remove the need for thinking. It exposes where thinking never existed in the first place