When Friction Disappears, Sprawl Begins

Internal Build in the Age of AI – Part 1

For most of the past two decades, internal build decisions carried a natural cost that shaped how organizations made them. Not because governance was especially rigorous, but because the work of building something internally (assembling the engineering capacity, coordinating the effort, absorbing the distraction) was expensive enough to suppress a significant portion of marginal ideas before they ever became real commitments. The cost itself was doing governance work, even when no governance process existed to do it deliberately.

That cost has changed materially. Teams that could not have realistically assembled a custom workflow tool, reporting pipeline, or internal approval system a few years ago can now produce something functional in days. That is a genuine shift in strategic capability, not just operational efficiency. Tools that were previously too expensive to justify, workflows too specific to the business to warrant vendor complexity, capabilities too narrow in scope to clear the old cost threshold: many of these are now rational to build in ways they simply were not before. The decision surface for internal development has expanded, and some of what is now accessible through that expanded surface represents real strategic value.

But the filter that used to shape which decisions got made has fallen without a corresponding expansion in the governance discipline required to evaluate them well. The same conditions that make more good ideas newly viable also make more bad ideas newly executable. AI has not just made it easier to build things that should not exist. It has made it newly rational to build things that previously would have been too expensive, too niche, or too operationally specific to justify; and it has done both simultaneously, without distinguishing between them.

This is not primarily a technology story. It is a governance story about what happens when the decision surface expands faster than the organization’s capacity to evaluate what is on it.

Organizations that developed internal build governance standards (formal or informal) understood something that is easy to undervalue in retrospect: not every constraint on internal development was inefficiency. Some of it was discipline. Architecture review before a tool got adopted. Approval routing before infrastructure commitments were made. Ownership documentation before a system touched a critical process. These mechanisms were slow by design, because the slowdown produced something the organization needed: a moment of confirmed understanding about what was being taken on, and by whom.

The difficulty is that productive friction and waste friction can look nearly identical from the outside. Both slow things down. Both generate friction between the team that wants to move and the process that requires them to stop. The difference is not visible in the delay; it is visible in what the delay was producing.

Waste friction slows work. It creates delay without creating accountability, documentation, or confirmed ownership. Removing it makes the operation faster and nothing else changes.

Productive friction prevents drift. It creates checkpoints where obligations get made explicit, where the question of who owns this in three years gets asked before the answer becomes irrelevant, where a handoff moment produces a record rather than an assumption. Removing it also makes the operation faster, but it removes something in the process that was carrying weight the organization may not be aware it was depending on.

The governance problem is not that organizations are now removing friction. Removing friction is often the right call. The problem is that the capacity to distinguish between the two has not kept pace with the speed at which both kinds are now being removed.

This is where the leadership question sharpens, and where it tends not to get asked clearly enough.

The question is not whether the organization removed the right friction in any given instance. That is a process design question, and it sits at the wrong altitude for how consequential the underlying issue is. The more important question is structural: who in this organization has the authority, visibility, and operating context to recognize which friction was carrying governance weight before it is removed?

In most organizations, that question does not have a clean answer. The people with the technical context to understand what a particular process step was doing are often not the people with the governance authority to decide whether it should stay. The people with the governance authority often lack the operating context to evaluate what they are looking at. And the gap between those two positions is where productive friction tends to disappear quietly, without anyone making an explicit decision that it should.

This is not an execution failure. It is a structural one. The organization has not assigned the oversight function that the condition now requires.

The consequence accumulates in a predictable pattern, even if it is rarely visible until later.

When the decision surface expands, internal build activity rises. That activity is not uniformly problematic; some of it reflects exactly the kind of strategic specificity that was previously too expensive to pursue, and the systems that result are genuinely worth owning. But every internal system the organization builds becomes something the organization is responsible for, regardless of how strategically sound or operationally convenient the original decision appeared. Maintenance, support, continuity planning, and the institutional knowledge required to keep a system legible over time: none of these compress at the same rate as the work of building.

What can be assembled in days still takes years to own well. And when the decision surface expands without a corresponding expansion in ownership discipline, the organization begins accumulating operational obligations faster than leadership can see them forming. The valuable builds and the marginal ones arrive through the same process, create the same categories of obligation, and are equally invisible to governance structures that were never designed to evaluate decisions at this volume or velocity.

That is the precise diagnostic. AI has expanded the set of viable internal build decisions faster than leadership has expanded the governance discipline required to evaluate them. The systems are being built. The obligations are being created. The authority and visibility required to distinguish which commitments the organization is genuinely prepared to carry has not been designed into the decision process.

That gap is not an IT operations problem. It is a governance design problem, and the friction that used to suppress it is no longer doing that work.