The New Divide

Individuals adopt tools. Organizations adopt change. The distance between the two is where the next decade of competitive advantage forms.

AI adoption at the personal level is already widespread. People across every function are using AI to draft, summarize, analyze, and build. The momentum is accelerating.

But productive individuals do not make a productive firm. George Sivulka names the core tension: a team where everyone uses AI daily is not the same as a team that has redesigned how its work moves. Sivulka calls the missing capability “institutional intelligence,” the organizational capacity to absorb new technology into how decisions get made, how work gets coordinated, and how processes evolve. Most organizations are only beginning to develop it.

The economics of each path are not close. Individual productivity gains plateau. Organizational operating improvements compound: each cycle reduces exception volume, speeds resolution, and makes the next cycle more valuable. Organizations that make the harder transition build capabilities their competitors cannot replicate by handing out better tools.

And the cost of waiting is not neutral. Shadow workflows multiply. Teams begin handling the same process in different ways. What started as enthusiasm becomes fragmentation, and fragmentation is harder to reverse than to prevent.

Where the Two Speeds Diverge

The speed gap is not distributed evenly across the organization. It clusters at the frozen middle: the layer where operating leaders sit between individual teams doing new work and executive leadership counting on results.

Middle management is caught between two incompatible mandates. Keep the operation running today. Transform it for tomorrow. Individual adoption is already happening on the teams they oversee; they see the energy, the experiments, the friction with legacy processes. But they are structurally constrained by operating models, exception protocols, reporting structures, and accountability frameworks built for stability, not change. The frozen middle is where institutional intelligence has to be built, but where the leverage to build it is weakest.

Every major technology wave follows a version of this pattern. The individual capability arrives first. The institutional adaptation follows later, and the distance between the two is where value gets lost. Fiber reached the backbone before it reached the home. Cloud migration moved the easy workloads first; the last data center took years. Mobile devices were everywhere before mobile-first business processes existed. In each case, the technology was ready before the organization was. And in each case, closing that last mile was the hardest, most expensive part of the transition.

Marshall McLuhan called this pattern driving to the future via the rearview window. Organizations adopt new capabilities by pattern-matching against what they already know. The first instinct is to bolt the new technology onto existing workflows without changing how the work moves. Ivan Zhao describes the current moment as the “waterwheel phase”: when steam engines first arrived, factory owners swapped out the waterwheel but kept the same floor plan, the same power distribution, the same workflow. Productivity gains were modest. The real gains came when owners redesigned the operation around the new capability rather than fitting it into the old one.

Most organizations are in the waterwheel phase with AI. Individual adoption is running well ahead of operating change. People are faster. Processes are the same. The gap between what individuals can do and what the organization can absorb is growing, not shrinking.

What AI Reveals About Operating Foundations

OpCo Intelligence’s 2026 State of AI Transformation survey surfaced a pattern that matches what we see in practice: organizations are active in AI initiatives and confident in individual adoption, yet success metrics remain absent and accountability is fragmented. Ninety-one percent of respondents reported feeling personally accountable for AI transformation in their role. Their organizations showed no such clarity: accountability was spread across at least six different structures, with no single model adopted by more than a quarter of respondents. The energy is individual. The infrastructure is not.

Google’s DORA 2025 research offers a diagnostic lens, and the initial finding was counterintuitive: across the full sample, more AI use correlated with worse delivery stability and throughput. The resolution explained the anomaly. AI magnifies organizational operating foundations. Organizations with strong foundations (fast feedback loops, loosely coupled systems, clear ownership of decisions) saw meaningful productivity gains. Those without saw dysfunction amplified: more exceptions, clearer bottlenecks, confusion about who decided what. AI did not create these problems. It exposed them.

This reframes the readiness question entirely. It is not that organizations are not ready for AI. It is that AI reveals what was already true about the organization. An operating foundation that works well with humans works better with AI partners. One that is fragmented, slow to decide, and heavy on exception overhead becomes painfully visible under AI acceleration.

The organizations that escaped the waterwheel phase in each previous wave did so by redesigning the operation around the new capability. The same applies here. The institutional side catches up in two moves: measure the distance, then close it through operating action.

Assessment identifies where individual adoption has momentum, which operating areas are strongest, and where to focus first. Action means picking one workflow and one team, then making the operating decisions that individual adoption never forces. Where does authority sit when the AI is wrong? Who owns the exceptions? How does the team know whether the process improved? How does what the operation learns feed back into the workflow?

Each decision answered is an operating outcome. Each outcome makes the next cycle more valuable. And the frozen middle is where these decisions matter most. The operating discipline to make them, tested through live operations rather than planning exercises, is where advantage compounds.