AI Won't Fix a System That's Already Overloaded
A CIO told me recently that her board had mandated an AI strategy by Q2.
When I asked what the business expected AI to deliver, she paused.
“That’s a good question.”
That pause is not unusual. Right now, it may be one of the most expensive pauses in enterprise technology. Organizations are committing significant budget to AI without a clear picture of the problem they are solving. And that gap, between mandate and diagnosis, is where implementation money disappears.
What happens when the mandate comes before the diagnosis?
The pressure to show an AI strategy is real. Boards are watching. Analysts are publishing benchmarks. Competitors are announcing rollouts.
So organizations do what they do under pressure: they act.
Vendor evaluations. Governance frameworks. Pilot programs. Visible momentum.
What most are skipping is the work that should come first. Before any organization can deploy AI effectively, it needs an honest picture of how work flows today. Where do employees spend the most time for the least value? Where is the organization held together by people because no clear process exists? Where is the manual effort so disproportionate to the outcome that it signals a broken way of working, not a headcount problem?
Without that picture, you are not deploying AI strategically. You are spending budget hoping something lands.
Why smart leaders skip the diagnostic step
This is not a failure of intelligence. It is a predictable response to pressure.
Daniel Kahneman’s research identified two modes of thinking: fast and automatic versus slow and deliberate. (He called them System 1 and System 2, distinct from the Performance Mode and Preservation Mode framework we have discussed in this series.) Under urgency, the brain defaults to fast. It reaches for the action that most visibly reduces the threat.
An AI roadmap qualifies. A governance model qualifies. A vendor selection process qualifies.
The diagnostic work — understanding how your organization operates, where effort is wasted, where handoffs break down — requires deliberate, unhurried thinking. That is precisely the thinking that gets displaced first when pressure is high. The leaders under the most pressure to move fast on AI are the ones least positioned to slow down and ask the right questions first.
Evidence shows that skipping the diagnostic step has significant cost
Bent Flyvbjerg studied over 16,000 major projects across industries and decades. His finding is consistent: the single strongest predictor of project failure is insufficient planning before execution begins.
Organizations that invest heavily in upfront diagnostic work execute faster, overspend less, and deliver closer to their original intent. His principle: think slow, act fast.
The organizations skipping that step are not saving time. They are borrowing it, at a high interest rate, payable during implementation.
Think of it as organizational debt. Not debt in the code or the technology — debt in the operating model. The workflows nobody has mapped. The handoffs held together by institutional memory rather than process. The manual effort that has never been questioned because everyone has been too busy to stop and ask whether it should exist at all.
AI does not retire that debt. It inherits it.
The ERP wave of the early 2000s is instructive. The systems were delivered. The business outcomes weren’t. Not because the technology failed — but because the processes it was meant to improve had never been properly examined before the project began. AI is not immune to the same sequence.
AI is an accelerant
This is the part that does not get said clearly enough.
AI amplifies whatever it lands on.
A well-designed organization with clear workflows, defined decision rights, and people focused on judgment work? AI compounds that. A fragmented organization held together by people plugging gaps that should be processes? AI amplifies the noise — faster activity, more fragmentation, less clarity about what really matters.
The aspirin analogy is useful here. A persistent headache can be treated with aspirin. It works, for a while. But aspirin treats the symptom. It does not address the root cause. Water. Fresh air. Too many hours without a break. No amount of aspirin fixes the underlying condition.
Deploying AI into an overloaded organization is the same pattern. It suppresses the visible pain long enough to declare a win, while the structural problem compounds underneath.
The organizations that will see genuine AI ROI are the ones willing to do the unsexy work first.
3 Questions You Should Ask
Before your next AI investment conversation, three questions worth putting to your leadership team:
What specific business problem are we designing this to solve? Not “improve efficiency.” A problem with a name, an owner, and a cost.
Do we know where our most expensive human effort is going right now? If the answer is no, that is the first thing AI should help you find — not the last.
Are we treating the symptom or the root cause? An AI governance framework with no business requirements beneath it is aspirin for a structural headache.