By Jim Koetting
Somewhere in your company’s financials there is an expense no accountant has ever booked. It has no budget line, no owner, and no year-end review, yet you fund it every single day. It is the payroll you spend paying good people to reconstruct things your organization already knew: decisions that evaporated after the meeting ended, ownership that everyone assumed and no one confirmed, context that walked out the door when somebody changed jobs.
You know the symptoms, even if you’ve never connected them. Meetings keep getting longer while less gets decided. Projects drift. You ask for the same status update three times in one week, and here is the part worth sitting with: it is not because your people are incapable. It is because reality inside your company has become genuinely difficult to see, for them and for you.
Most leaders diagnose this as a communication problem, and honestly, that’s a reasonable guess. It also happens to be wrong. Your organization communicates constantly. Emails, meetings, Slack threads, presentations, calls, dashboards, project boards. If communication volume could fix this, it would have been fixed years ago. The problem is that communication does not preserve understanding. Information moves between people all day while organizational memory quietly falls apart, and no amount of additional messaging repairs that.
Software engineers wrestled with a version of this problem for decades before they found the words for it. What happened when they finally did is worth your attention, because it points to a way out.
The Debt That Doesn’t Live in Code
Picture a city that has grown continuously for a hundred years. A road was widened without moving the buildings around it. A temporary bridge went up because there was no time for a permanent one, and forty years later it is still carrying traffic. Power lines got rerouted around new construction. Old buildings were grafted onto infrastructure never designed to hold them. Here’s the thing: almost every one of those decisions was rational when it was made. Each solved a real problem, satisfied a pressing need, or postponed an expense the city could not afford that year. Nobody was foolish. Nobody was lazy.
But residents don’t experience the city one decision at a time. They inherit the cumulative effect of thousands of choices that were never designed to work together. Navigation gets confusing. Maintenance gets expensive. Simple improvements require engineers to untangle yesterday’s compromises before they can build tomorrow’s solutions. The city still functions, but a growing share of its energy goes to working around its own history.
Software ages the same way. Even a well-designed application collides quickly with commercial reality. Customers demand features. Competitors force pivots. Security holes need patching today, not next quarter. So developers make practical trade-offs: duplicate some code instead of redesigning the subsystem, skip the documentation because shipping matters more, patch the aging component because it still works well enough. Each choice saves time now by borrowing complexity from later.
In the early 1990s, a programmer named Ward Cunningham gave that accumulated burden a name his executives could understand: technical debt. The brilliance of the metaphor is that debt is not an insult. Businesses borrow to build factories and fund growth; borrowing is often the smart move. Debt becomes dangerous only when it’s invisible, when interest compounds faster than repayment, or when nobody knows the true balance. A company drowning in technical debt hasn’t lost its talent. It has simply made too many short-term accommodations for too long without ever reserving time to reconcile them.
If you lead a hospital, a manufacturer, a bank, a law firm, a university, a nonprofit, or an agency, you are carrying the same kind of debt. Yours just doesn’t live in code. It accumulates through undocumented decisions, ambiguous ownership, disconnected systems, contradictory instructions, and knowledge that exists only in one person’s inbox or one person’s head. I call it cognitive debt: the burden created when people and organizations repeatedly trade sustained thought, shared understanding, and sound judgment for speed, convenience, and immediate output.
You’ve watched this happen, probably without naming it. A small company holds its essential knowledge in a few closely connected minds, and it works. Then the company grows, and the shared picture fragments. Sales carries one version of reality, operations another. Finance understands constraints product has never seen. Marketing knows things about customers that engineering will never hear. Every department acts rationally on what it can see, while the whole becomes harder and harder to understand. Again: nobody failed. The structure did.
Or watch a single decision travel through your company this week. It gets made in a meeting. Fragments land in handwritten notes; other fragments land in a Teams thread. Someone sends a summary email that half the recipients skim. The project manager updates the board, but only after two priorities have already shifted. One executive mentions the decision in a hallway, another references it on a customer call, and a third remembers it differently. Nobody set out to create confusion. By Friday there are six slightly different versions of the same reality, and your people are choosing between them blind.
Now consider what that costs the humans involved, because this is where the debt stops being abstract. Before your people can do the work, they must reconstruct the conditions around it. They hunt for documents, compare contradictory instructions, reread old threads, and quietly ask a colleague who actually owns this. The brainpower that should have gone to analysis, creativity, and judgment gets consumed recovering context the organization failed to preserve. If you have ever wondered why smart, motivated people seem to move slowly in your company, this is very often the answer. They aren’t slow. They’re taxed.
Like all debt, this one charges interest. It shows up in meetings that begin with twenty minutes of reconstruction. In every new hire who spends months relearning what the company already paid to learn once. In talented professionals burning hours searching for information they know exists somewhere. In teams that postpone decisions because nobody trusts the available version of the facts. None of it appears on your P&L, but every dollar of it is payroll, spent recovering clarity that should never have been lost.
One more layer, and it’s the one leaders least like to look at. The debt hides in structural places: decisions never durably recorded, ownership implied rather than established, context trapped inside departments. But it also hides in a relational place: connections too thin to carry information reliably. When trust is weak, even perfectly preserved information gets questioned, filtered, withheld, or misread. Whatever clarity you manage to create starts leaking away in the space between people. You cannot fix that with software.
David Allen captured the personal version of this in Getting Things Done: the mind is for having ideas, not holding them. Every uncaptured commitment occupies bandwidth, because the brain keeps reminding itself that something remains open. You have probably felt that at eleven at night, staring at the ceiling, running the list. Organizations have been slower to learn the collective version of the lesson. They behave as though memory can scale indefinitely, as though hundreds of people can coordinate thousands of commitments across disconnected systems without friction. They can’t. So companies start mistaking effort for clarity and activity for alignment, and everyone works harder while understanding less.
When AI Makes Output Abundant
If you feel a certain unease about AI arriving in the middle of all this, trust it. That instinct is sound.
Cognitive debt was expensive before generative AI, but its growth had a natural governor: the speed at which humans can produce information. A person can only write so many reports in a week. AI removes the governor. Your organization can now produce more documents, analyses, summaries, proposals, and recommendations than your people have the time or mental capacity to evaluate. Read that sentence again, because the constraint that just disappeared was quietly protecting you.
AI is an amplifier, and amplifiers don’t care what signal they receive. Hand it to an organization with clear decisions, explicit ownership, and strong shared understanding, and it converts that clarity into extraordinary leverage. Hand it to a company already drowning in cognitive debt, and it accelerates the drowning: more reports nobody reads, more summaries of meetings where nobody truly aligned, more confident versions of reality generated faster than anyone can reconcile them.
This is the paradox at the center of every AI adoption conversation you’ll have this year. The cost of producing an answer is collapsing. The cost of determining whether it’s the right answer is not. A report generates in seconds, but someone still has to decide whether its assumptions hold, whether its information is current, whether it reflects the decision actually made, and whether acting on it will work. That someone is you, or somebody you pay. AI increases the volume of material without increasing anyone’s capacity to understand what it means.
When output becomes unlimited, judgment becomes priceless. Judgment decides which questions deserve asking, which information matters, which trade-offs are acceptable, and which consequences cannot be delegated to a machine. As production gets easier, discernment gets scarcer and therefore more valuable. The rare resource in your company will no longer be the ability to create another report. It will be the presence of a mind capable of deciding what deserves attention.
There’s a subtler danger underneath, and it may be the most important paragraph in this piece. Cognitive debt doesn’t grow only when information gets lost. It also grows when attention fragments, when people bounce reflexively between screens, when difficult thought gets replaced by rapid production, and when we accept fluent answers before forming our own questions. That last one deserves honesty: it is seductive, it is available to all of us right now, and none of us is above it. An organization can increase its visible output while quietly weakening the human understanding underneath, and from the outside the two look identical for a surprisingly long time.
This is why cognitive debt will become one of the defining leadership challenges of the next decade. The companies that pay it down will convert AI into compounding advantage, because their people supply the context and judgment the technology lacks. The companies that ignore it will run increasingly powerful engines on increasingly fragmented fuel, producing more while knowing less.
Naming What We Owe
Here’s the most hopeful part of the technical debt story, and the reason this essay exists. Engineers didn’t discover a new problem in 1992. The burden had existed before it was named and developers felt the drag every day. But “the code is getting harder to work with” sounded like an internal complaint. The language of debt made the future cost of present compromises visible to people who controlled budgets. Naming the problem didn’t repay it. It made repayment conceivable.
You are standing at the same point with cognitive debt, and so is nearly every leader you know. The drag is real and the symptoms are everywhere, but they keep getting misdiagnosed as separate problems: poor communication, ineffective meetings, weak accountability, the wrong technology, resistance to change. Each diagnosis contains a sliver of truth. None of them explains why capable people can work this hard while the organization becomes harder to understand. If you’ve cycled through those diagnoses and watched the fixes fail, the problem was never your effort. You were treating symptoms of a condition that didn’t yet have a name.
Once it has a name, you can trace it. You can start asking questions this week that most leaders never ask. Do decisions survive the meetings where they’re made? Is ownership explicit, or assumed? Can context travel across departmental boundaries, or does it die at the border? Do your people have enough cognitive capacity left to exercise the judgment you hired them for? Is your technology preserving understanding, or just increasing the volume everyone must process? And the uncomfortable one: do people trust each other enough to share what they know, challenge what looks wrong, and admit when reality has changed?
The organizations that thrive in the age of AI will not be the ones that generate the most output. They’ll be the ones that preserve enough shared understanding to know which output matters, which decisions must remain human, and which version of reality can be trusted. Cognitive debt is what piles up when the capacity for that judgment gets spent reconstructing what the organization should already know.
The balance is invisible. The interest is not: you pay it daily in time, attention, trust, execution, and opportunities that quietly expire. The first act of repayment costs nothing and starts today. Learn to recognize what your organization owes.
How much cognitive debt is your organization carrying? Explore the Cognitive Debt Organizational Assessment at cognitivedebt.io and begin identifying where the interest is accumulating.