Why Every Technology That Makes Creation Easy Makes Governance Hard
For three decades I have sat at the intersection of business problems and technical solutions — translating between executives who need outcomes and practitioners who build them. That vantage point has come with a recurring experience. A new technology arrives promising to put powerful capability in more hands. It delivers. And then, reliably, it leaves behind something that didn’t show up in the productivity numbers.
This is not an argument against technology. The tools this piece examines, from Excel to AI, delivered real value to real people solving real problems. That value was genuine. So was the cost that followed it.
What follows is an attempt to name the pattern, explain the mechanism, and offer a more useful question than the one most organizations are asking when the next wave of powerful tools arrives. Not what can we build with this — but what are we deciding, and what will it cost when someone eventually has to make those decisions anyway.
The Governance Wasteland
Every senior manager can remember the moment the Excel problem became clear to them. Maybe it was the meeting where three departments brought three versions of the same number and the next hour was spent figuring out which one to believe. Maybe it was the day the person who built the model left and took the institutional knowledge with them. Maybe it was the audit.
The spreadsheet itself wasn’t the problem. It put analytical capability into the hands of anyone with a PC. A financial analyst could do in an afternoon what previously took a team a week.
But that productivity miracle contained a hidden cost. The person who built the spreadsheet experienced all of the upside. Governance consequences like version sprawl, fragmentation, and institutional knowledge locked in a file on someone’s desktop arrived later, diffusely, and usually landed on others.
Decades later, organizations are simultaneously still excavating Excel debt created in the 1990s and creating new Excel debt today. Because the decision to build the spreadsheet made complete sense to the person making it. The value was visible and immediate. The cost was invisible and deferred.
This is the pattern: every technology that makes creation easy makes governance hard.
The productivity miracle and the governance wasteland are not opposites. They are the same thing, seen from different distances and different points in time.
Excel wasn’t the last technology to make this promise. And it won’t be the last to leave the same bill behind.
The Promise Has Always Been The Same
That promise has a history worth tracing.
Every wave of technology that puts creation in more hands arrives with the same promise: you no longer need specialists to build things. After Excel came Microsoft Access — a database in anyone’s hands. Then SharePoint — content management and collaboration for the masses. Then Canvas Apps — everyone can be an app developer, and here’s a SharePoint list as your default backend. It’s free. What could go wrong.
Each wave delivered value, adopted by people solving real problems. And each left behind something that didn’t show up in the productivity numbers.
The Access databases living on shared drives that IT departments are still replacing. The SharePoint environments that grew into ungovernable labyrinths of sites, subsites, and permissions that nobody fully understands. Canvas Apps built on SharePoint lists that worked for twelve people and quietly collapsed at two hundred.
Notice the pattern. Each wave arrived before the previous cleanup was finished. Organizations are now carrying layered debt — Excel debt, SharePoint debt, Canvas App debt — simultaneously, each layer complicating the one beneath it.
This is governance debt. The accumulated cost of creation decisions whose consequences were deferred, diffused, and inherited by someone who didn’t make them.
Now the promise is being made again. Natural language. AI. Anyone can build anything. The barrier has never been lower.
Which means the rate at which governance debt can be created has never been higher. A Canvas App, despite its limitations, is at least inspectable — you can open it, click around, and discover what it does. An application built through natural language, with logic embedded in agent prompts, may not be.
The promise has always been the same. So has the bill.
Some will argue this time is different — that AI won’t just accelerate creation, it will solve the governance problem too.
A familiar promise.
Governance Debt Has No Balance Sheet
To understand why, it helps to understand how governance debt actually accumulates.
Every month, making the minimum payment on a credit card keeps your account in good standing. The bank agrees. You are meeting their definition of success. The debt grows anyway.
You don’t decide to accumulate debt. You decide to add a field. Fix a report. Handle a new requirement. A few hours a month, for a decade.
In three decades of working with organizations on business applications, I’ve seen the MS Access problem many times. A system built with enough skill to work, maintained and extended over a decade, mission critical long before anyone fully recognized it as such. When it finally breaks — and it always breaks — the replacement cost is measured in years and hundreds of thousands of dollars.
Nobody decided to create that debt. They decided to make the system work. The tools rewarded the behavior that was creating the problem.
Governance debt has no balance sheet. While financial debt is visible, measured, and reported, governance debt doesn’t appear until the system breaks and the invoice arrives. By then it is no longer a governance conversation. It is a crisis.
This is the structural condition built into every tool designed to put creation in more hands. The people who built and maintained those Access databases developed the skills the tool required but not the skills that would have revealed what was accumulating underneath.
The capability gap isn’t a client failing. It’s a design condition — accessible enough to build with, not transparent enough to reveal the consequences.
Every wave of tools that put creation in more hands works the same way. The tool lowers the barrier to creation. Organizations gain the ability to build without gaining the visibility to see what they are accumulating.
The decision to build feels like progress. The decision about how it fits into a governed system — who owns it, where the data lives, what happens when it needs to change — gets quietly deferred.
AI doesn’t change this structure. It accelerates it. Previous waves deferred decisions at human creation speed — the time it took to build something created a natural pause where a governance conversation could have happened. That pause is closing. An application that would have taken weeks to build now takes an afternoon. The governance conversation that should accompany it doesn’t automatically compress at the same rate.
The question for every executive approving the next wave of tools that put creation in more hands isn’t what can our people build with this. It’s what decisions are we about to defer that we should be making explicitly — and what will it cost when someone eventually has to make them anyway.
And increasingly, the approval conversation may never happen at all. Every employee already has access — official or otherwise — to tools capable of creating complexity that won’t surface until something breaks.
The governance debt is accumulating whether or not anyone decided to let it.
They didn’t choose the debt. They just kept building.
The Burden Shifts Up
That accumulation doesn’t just create a problem for organizations. It creates an industry.
Sprawl creates opportunity. Consulting firms are trained to find it. The governance debt that tools that abstract and compress development generate is not just a problem for client organizations — it is also a revenue stream. Rescue projects arrive with urgency, visible impact, and the opportunity to look like the hero who solved the crisis.
Most hands-on practitioners don’t experience it that way. They see someone else’s decisions, a cleanup mandate, and work that only has value in that it clears the way for something better. Demolition isn’t architecture. The skills required to unwind a decade of Access debt are real skills. They are not skills that compound.
This is where the burden lands.
The promise of every wave of tooling that abstracts and compresses development is that organizations need fewer specialists. The operational reality is that specialists are busier than ever — just not on the problems that build durable value. Senior architects and consultants are being consumed by licensing archaeology. By modeling token consumption for AI features whose usage patterns aren’t yet understood. By evaluating which preview features are stable enough to build on. By AI security considerations.
None of this compounds. Every hour spent making sense of a governance wasteland is an hour not spent building the organizational patterns and architectural foundations that prevent the next one.
The cruelest part of the structure is who absorbs this cost. The tools were designed to reduce dependence on senior technical expertise. The governance consequences of those tools are legible only to senior technical expertise. The people whose judgment is scarcest and most expensive are being redirected from problems that build toward problems that clean up.
The most valuable intervention is not the rescue project. It is the conversation that happens before the thing gets built. Before the data model gets decided by default. Before the ownership question gets deferred. Before the tool gets chosen because it’s included in the license. That conversation requires being in the room at the decision moment — not after the invoice arrives.
That is a different engagement model. It is a harder conversation to initiate with a client who has come to you with something they want built. It requires saying: before we build this, let’s talk about what we are actually deciding. What foundation are we laying. What lifecycle are we committing to manage.
Not every building needs to be an architectural masterpiece. But every building deserves a foundation that was chosen deliberately rather than inherited by accident.
Deliberate or Default
Some sprawl is inevitable. So is some governance debt. The question was never whether your organization would accumulate it. It was always whether the decisions that created it were made deliberately or by default.
Every business application deployment involves decisions whether or not anyone makes them consciously. Data model decisions get made by default when nobody makes them explicitly. Ownership decisions get deferred when nobody claims them. Lifecycle decisions get avoided when the tools that abstract and compress development make it easy to start without asking what happens next. The tools always have a default answer. The question is whether your organization has a better one.
This is what governance theater gets wrong. Steering committees. Policy documents. Approval workflows that slow creation without informing it. These are the organizational equivalent of the minimum payment — they create the appearance of governance without addressing the decisions that actually matter.
The decisions that matter are smaller and harder than a policy document. Who owns this. Where does the data live and why. What happens when the person who built this leaves. What does change look like in two years. What is the lifecycle we are committing to manage. These questions don’t require a committee. They require a conversation that happens before the thing gets built rather than after the invoice arrives.
Not every application warrants an architectural review. But every application deserves a conscious decision about what it is, who owns it, and what happens when it needs to change. And every organization building on platforms that make creation easy deserves a clear set of principles for managing the hard parts of the lifecycle — not as bureaucracy but as the organizational memory of decisions that were actually made.
The current wave makes this more urgent than it has ever been. AI is compressing the time between decision and deployment faster than any previous wave of development tooling. The natural pause where a governance conversation could have happened is closing. Organizations that establish deliberate decision practices now — before the wave fully crests — will be managing a lifecycle. Organizations that don’t will be managing a crisis.
Every senior manager can remember the moment the Excel problem became clear to them. That moment always arrives. The cost is set long before it does.
What small hard decision is your organization avoiding right now that a tool, a default, or a deadline made it easy to defer?