What the Industrial Organization Was Quietly Doing, and What Has to Replace It

A three-part series

CONTENTS

Part One:   The Changing Economics of Coordination

Part Two:   What the Boxes Were Carrying

Part Three:   What Replaces the Boxes?

---

Part One: The Changing Economics of Coordination

Part one of a three-part series on the changing architecture of coordination.

A small-business owner applies for credit after two quarters of weakening receivables. The numbers suggest deterioration. The person who has worked with the business for years knows the largest customer temporarily extended its payment terms during an acquisition while orders remained strong. That context changes the decision.

For much of the Industrial Age, information, expertise, judgment, authority, and institutional memory like this tended to sit relatively close together. The organization held the records. People developed expertise inside defined roles. Relationships accumulated history. When something did not fit the standard process, there was usually a recognizable place for the exception to go.

The same decision can now be assembled very differently. The application enters through a digital interface. Identity evidence comes from one provider, transaction data from another, fraud screening from a separate system, and risk assessment from a model. Pricing can be automated. Servicing may sit on another platform. Some of the underlying capability may belong to organizations the customer never sees.

Each component can become faster and more capable while the outcome depends on a larger and more distributed system. That shift is changing one of the basic economic assumptions behind how we organize work.

Digital technology made information progressively cheaper to move. What remained expensive was understanding what that information meant in a particular situation. AI is beginning to change that part of the equation too.

Why permanent structure made economic sense

For most of the Industrial Age, organizations managed complexity by putting capability into relatively permanent structures: jobs, departments, functions, institutions, reporting lines, and professional disciplines. It is easy to look at those structures now and see bureaucracy, but they solved a real problem. Coordination was expensive.

Finding information took time. Expertise was difficult to locate. Communication across distance was costly. Work had to be checked. Decisions moved slowly between functions. Crossing an organizational boundary introduced contracts, delay, uncertainty, and risk.

Permanent structure reduced those costs. Expertise could be concentrated. Roles made responsibility clearer. Recurring work could be standardized. Records created institutional memory. Hierarchy provided a path for resolving disagreement and moving decisions. Long-running relationships meant people did not have to rebuild understanding every time a new problem appeared.

The organizational chart was therefore more than a map of reporting lines. It also showed where capability lived and gave the organization a repeatable way to bring that capability together. The underlying economics favored organizing capability in advance, because assembling it from scratch every time a need appeared was too slow and too expensive.

What AI extends, and what it does not

Digital technology has been weakening that assumption for decades. Communication became nearly instantaneous. Information became easier to retrieve. Specialized providers became reachable across distance. Software made it possible to coordinate increasingly complex activity across company boundaries.

AI extends that progression. It can find relevant knowledge, synthesize large bodies of information, locate expertise, interpret unfamiliar material, communicate across specialized domains, monitor dependencies, generate alternatives, and increasingly take bounded action. Some of the coordination that once had to be built permanently into roles, departments, and reporting relationships can begin to happen around the problem itself.

The history of electricity offers a useful way to think about the transition. Early factories often replaced steam power with electric motors while leaving the layout of the factory largely intact. The technology initially improved the existing system. Larger gains came later, when factory design changed around what distributed electric power made possible. Production no longer had to be organized around the physical constraints of shafts and belts driven from a central power source.

AI will follow its own path, but the structural lesson matters. General-purpose technologies can spend years improving work inside an existing operating model before institutions begin reorganizing around what the new capability makes possible.

Much of today’s AI adoption is still in that first phase. We are making existing tasks faster, automating parts of existing workflows, and giving people increasingly capable tools while leaving the surrounding organization largely intact. The bigger organizational effects will emerge as firms begin asking which coordination requirements still need to be encoded permanently in structure and which can be handled more dynamically.

The pressures arriving at the same time

That question is arriving just as the system organizations are trying to coordinate becomes harder to contain. A single customer experience can already depend on several companies, platforms, data providers, contractors, and automated systems. Cloud platforms carry capabilities that once lived inside the enterprise. Expertise can sit almost anywhere. Supply chains and operating ecosystems routinely cross institutional and national boundaries.

The clock has changed too. Transactions, information flows, market responses, and machine decisions increasingly move faster than traditional escalation. Hierarchy remains useful, but a model built around information moving upward for review and a decision later moving back down becomes less effective when the underlying condition changes before the loop is complete.

Demographics adds a different kind of pressure. Aging workforces, longer careers, caregiving burdens, and shortages of specialized skills make it harder to assume that the needed expertise will always be sitting in the right department at the right moment. In that environment, AI is more than a productivity tool. It can help stretch scarce expertise, make specialized knowledge accessible to more people, and bring capability to problems that fixed staffing models struggle to cover.

These forces reinforce one another. The work is more distributed, the operating clock is faster, and some of the expertise required to manage complexity is becoming scarcer. AI accelerates parts of that shift while also providing new ways to coordinate through it.

Haven’t we heard this before?

There is a good reason to be skeptical. We have heard versions of this argument before. As communication and transaction costs fell in the 1980s and 1990s, many expected electronic markets and networks to weaken the rationale for large firms. If companies could cheaply find suppliers, exchange information, contract across distance, and verify transactions, why maintain so much capability inside permanent institutions?

Firms did not disappear. Many became much larger.

Part of the explanation is straightforward. The same technologies that reduced the cost of coordinating across company boundaries also reduced the cost of managing huge organizations internally. Scale, capital, regulation, networks, data, brands, and other advantages continued to make large institutions economically powerful.

But the earlier technologies left a critical coordination cost largely intact. They made information easier to move without making its meaning much easier to interpret. Organizations still needed people who understood the history behind the numbers, recognized when a case did not fit the category, connected information that sat in different places, and exercised judgment when evidence was incomplete. Those capabilities lived in experienced people, working relationships, professional communities, and accumulated institutional knowledge. They remained difficult to reproduce outside the structures that developed them.

That is the part of the coordination equation AI now begins to reach. What makes AI different is its ability to materially lower the cost of contextual interpretation at scale.

Return to the loan decision. The explanation for those weakening receivables was never necessarily hidden. It could sit in the customer’s payment history, in the timing of the change, in public information about the acquiring company, and in evidence that orders remained strong. What the relationship manager contributed was not privileged access to that information. It was the ability to assemble those signals and recognize what they meant together. That is the coordination work AI is increasingly able to do at scale.

AI does not eliminate judgment, and it certainly does not make every form of human understanding computational. But it can participate in work that previous generations of coordination technology left almost entirely to people.

What actually shifts

That creates a different possibility from the electronic-market predictions of the 1990s. The important question is no longer whether cheaper transactions will cause firms to disappear. It is how the boundary shifts between coordination that requires permanent organizational structure and coordination that can be assembled dynamically around a need.

Large institutions will remain. Hierarchy will remain. Permanent teams, deep expertise, and organizational identity will remain valuable. What changes is the amount of coordination those structures have to carry themselves.

That opens real possibilities. Organizations can bring expertise to a problem without routing everything through departmental boundaries. They can make scarce capability available more broadly, respond faster when circumstances change, and reduce handoffs whose primary purpose was moving information from one organizational box to another.

But those boxes were holding more than capability. They were also carrying context, accountability, and the means by which people developed judgment, much of it through ordinary relationships and routines that nobody thought of as coordination. As technology changes the need for some of the structure, those functions remain.

Before we redesign the boxes, we need to understand what they were quietly doing for us. That is the subject of Part Two.