The Addressable Layer

TAM-CV.10 · The Capital View · The Approximate Mind

The first nine essays in this arc examined how capital organizes the transition from human-delivered to AI-orchestrated services in fragmented industries. The enclosure of coordination, the three tiers, the blue mug, the asymmetric deployment, the dual-asset exit. One arc. One side of the ledger.

But capital reads everything. And the Coordination cluster, which examined what happens when workers own the AI coordination layer rather than renting it from someone else, is not invisible to the people whose business is spotting structural change early enough to profit from it.

Marcus read the Coordination essays the way he reads everything: for the structural implication, then for the deal.

The structural implication was clear. If AI can perform the management function, the management layer is removable. The Coordination cluster drew the cooperative conclusion: workers could own the coordination and keep the surplus. Dale’s company replaced his managers with an AI system. Charlene’s cooperative replaced the entire corporate hierarchy. Ravi’s network replaced seven intermediaries. In each case, the insight was the same. Management was optional. The coordination it performed was real. The layer that performed it was not.

Marcus drew a different conclusion from the same structural fact.

“They’re right,” he said. “Management is optional. The question is who captures the savings when you remove it.”

He was not talking about cooperatives.

The Operating Partner

#

Nora has been doing this for fourteen years. She is a PE operating partner, the person the fund sends into a portfolio company after the deal closes, the person whose job is to find where the value is trapped and design the structure that frees it. She has done it in food distribution, in commercial janitorial services, in regional HVAC companies. She is good at it. She walks into a company, spends three weeks understanding the operation, builds the hundred-day plan, executes.

Her method has not changed in a decade. Find the duplicated functions. Consolidate them. Install better reporting. Hire a CFO who speaks the fund’s language. Replace the founders who cannot operate at scale with operators who can. Reduce the cost structure. Grow the revenue into the reduced cost structure. Exit at a higher multiple than entry.

The method works because it is grounded in something real. Mid-market companies, the kind PE buys in the lower middle market, typically have management layers that grew organically with the business. The founder hired a cousin to run the warehouse. The cousin hired someone to do scheduling. The scheduling person became the operations manager. The operations manager hired an assistant. By the time the company reaches $15 million in revenue, there are seven or eight people whose jobs are primarily coordination: making sure the right truck is at the right warehouse at the right time, that the customer’s order matches the inventory, that the invoicing reflects what was delivered, that the new employee has a schedule that does not conflict with the three people who cannot work Fridays.

These are real jobs. The people doing them are doing real work. The company could not function without the coordination they provide.

Nora has never questioned this. The coordination was necessary, and the people were the coordination, and therefore the people were necessary. The logic was circular and invisible, like most load-bearing assumptions.

Eight months ago, the circle broke.

What Changed

#

Nora’s fund acquired a regional building supply distributor. Forty-seven employees, $18 million in revenue, three locations across the mid-Atlantic. The founder, who had built it over twenty years, wanted to retire. The company was well-run by the standards of its size, which means it had reliable customers, decent margins, and a management structure that had accumulated like geological strata, each layer deposited when the company hit a size that required someone new to handle a function no one else had time for.

Nora walked in and did what she always does. She mapped the organization. She identified the roles. She built the cost structure analysis.

Of forty-seven employees, fourteen were in what she would classify as coordination roles: operations management, scheduling, order processing, inventory reconciliation, customer service routing, quality tracking. They were the nervous system of the company. They made it work.

The AI orchestration layer the fund had been developing across its portfolio was not a chatbot. Not a dashboard. It was trained on two years of operational data from similar companies in the fund’s portfolio, capable of handling the scheduling, the inventory matching, the order routing, the exception flagging, and most of the customer communication that constituted the daily work of those fourteen people.

The system could handle, from day one, roughly 80 percent of the coordination work that the fourteen people performed. Not perfectly. Not without human oversight for the exceptions. But competently enough that the oversight required three people, not fourteen.

Nora sat with the math for a week before she brought it to the fund.

The Math

#

The fourteen coordination roles had a fully loaded cost of approximately $1.1 million annually. Salaries, benefits, payroll taxes, the workspace they occupied, the management attention they required. This was not excessive compensation. Average salary in this group was around $52,000. These were people who had built their lives around what the company paid them. Some had been there for a decade.

The AI orchestration layer, amortized across the fund’s portfolio of similar companies, had a per-company annual cost of approximately $180,000. This included the platform license, the integration work, the ongoing monitoring, and the three human positions that would remain to handle exceptions, manage supplier relationships that required a phone call, and oversee the system’s output.

The net savings: roughly $920,000 annually. On a company with $18 million in revenue and historical EBITDA of about $2.2 million, this was a 42 percent improvement in operating profit. Not from growing the business. Not from finding new customers. Not from better pricing. From removing the management layer.

Nora had been improving companies for fourteen years and had never seen a single intervention produce a 42 percent improvement in operating profit. The typical PE operational playbook, fully executed over three years, produces cumulative EBITDA improvement of 20 to 30 percent across all initiatives combined.

The management strip is not an incremental improvement. It is a different category of value creation.

The entry multiple on the acquisition was six times EBITDA. The exit multiple, assuming the market prices the transformed cost structure correctly, would be ten to twelve times the new EBITDA. The return math is clean and enormous, large enough that Nora checked it three times and then had the fund’s analysts check it independently.

The analysts confirmed it. They also noted, in a footnote that Nora read twice, that the eleven displaced employees would receive severance consistent with the fund’s standard terms: two weeks per year of service, capped at twelve weeks.

What the Coordination Cluster Saw

#

Dale’s company did the same thing. Replaced human managers with an AI coordination system. Dale noticed the improvement immediately: better routes, staged parts, less time navigating the organization. The eleven managers before his twelfth were, in his description, mostly weather. Present. Occasionally significant. Something to work around.

The Coordination cluster told this story as a liberation. The management layer was removed, and the worker’s relationship to the work improved. Dale was closer to the purpose of his job than he had been when eleven humans stood between him and the lines.

The PE management strip produces the same structural outcome. The coordination is automated. The frontline workers remain. The organizational layer between purpose and execution dissolves. The work itself, for the people still doing it, improves in exactly the ways Dale described.

The difference is who captures the savings.

In Dale’s company, the savings went to the company, which is to say to a corporate entity whose distribution of the benefit was diffuse and unspecified. In Charlene’s cooperative, the savings would flow to the workers. In Ravi’s network, to the producers.

In the PE management strip, the savings go to the fund’s limited partners. The margin that Kevin’s salary consumed becomes return on invested capital. The coordination that was a cost becomes EBITDA. The EBITDA becomes the exit multiple. The exit multiple becomes the carry.

Same technology. Same structural insight. Same operational outcome. Different beneficiary.

This is the crux. The Coordination cluster examined the insight from the position of the people inside the structure. The Capital View examines it from the position of the people financing the structure. Both see the same truth. Both draw different conclusions. Both are acting on their conclusions right now.

Kevin

#

Kevin has been the operations manager for seven years. He is forty-one. He has a mortgage on a house twelve minutes from the warehouse, a payment on a truck he bought when his second child was born, and a set of professional skills that are almost entirely composed of the coordination work the AI layer now performs. He knows the routes. He knows the drivers. He knows which customers will accept a partial shipment and which will reject the entire order if a single item is missing. He knows that the warehouse in Fredericksburg runs hot in July and the adhesives on aisle six need to be moved to climate-controlled storage by Memorial Day or the returns spike in August.

Some of this knowledge is in the system now, because Kevin entered it during the data migration that he was told was for operational improvement. Some of it is not in any system, because it lives in the relationships Kevin maintains with the drivers and the customers and the warehouse staff, relationships built on years of solving problems together in ways that did not generate data points.

Kevin will receive eight weeks of severance. He will have difficulty finding a comparable position because the skills that made him valuable, the coordination skills, the people skills, the ability to hold the logistics of a mid-size distribution company in his head and route around problems before they became crises, are precisely the skills the AI layer is eliminating across every company in the sector. He is not being replaced by a better operations manager. He is being replaced by a category shift. The category he occupied no longer exists.

Dale got to stay. Dale climbs poles. His work is physical, specific, irreplaceable by current technology. The AI coordination system improved his working conditions. Kevin’s work was coordination itself, and the coordination was what got automated.

The Coordination cluster had a name for this. The quiet reversal. The two-century movement from physical labor toward cognitive labor, reversed. The lineman climbing the pole in an ice storm is doing something AI cannot replicate. The operations manager reviewing a spreadsheet is doing something the most basic AI system can replicate today. The hierarchy inverts.

Kevin is on the wrong side of the inversion.

Nora does not use the word cruelty. She would say that the company was paying fourteen people to perform a function that technology can now perform more reliably and at lower cost, and that maintaining those positions to avoid the discomfort of eliminating them would be a form of subsidy that neither the company nor its customers have agreed to provide.

She is not wrong about this, within the frame she is using.

The Replicable Play

#

The building supply distributor is not special. It is representative.

Across the lower middle market, thousands of companies with $10 to $50 million in revenue have management structures that look like this one. Built organically. Layered over time. Performing real coordination that real people depend on. And the coordination layer in each of them is the same addressable cost.

The fund models the play across its portfolio. Janitorial services: acquire a regional provider, install the AI orchestration layer, reduce management headcount by 60 percent, improve margins by 30 to 40 percent. HVAC: same play, different industry, similar math. Light manufacturing: slightly different because the coordination includes production scheduling, but the AI handles production scheduling well. Food distribution: the coordination is logistics-intensive, which is where the AI excels most.

In each case, the same pattern. Buy at a multiple that reflects the current cost structure. Install the AI layer. Remove the management layer. Sell at a multiple that reflects the new cost structure. The spread between the entry and exit multiples is the return.

The management strip is the PE playbook adapted to the structural fact that the Coordination cluster described. Management is optional. PE is very good at removing optional costs.

The fund estimates it can execute this play across fifteen to twenty portfolio companies in the current vintage. Total management positions eliminated: roughly two hundred to three hundred. Total EBITDA improvement across the portfolio: $12 to $18 million annually. Total return to limited partners: substantial, depending on exit multiples that the fund believes will be generous because the buyer is acquiring a structurally transformed cost base that no traditional competitor can replicate without performing the same strip.

The buyer of the transformed company inherits an organization that runs lean because the coordination is automated. The buyer cannot be undercut by a competitor who still carries the management overhead. The moat is structural, not operational.

What Marcus Noticed

#

He was quiet for a while after Nora presented the portfolio model. He looked at the numbers. He looked at the staffing analysis. He looked at the severance projections.

Then he said something I did not expect.

“This is the enclosure of coordination applied to the firm itself.”

He was right. The original Capital View arc described how AI makes invisible coordination legible and capital encloses what becomes legible. The daughter’s coordination of seven aging-at-home services became the horizontal composition rollup. The patient’s navigation of the behavioral health system became the care orchestration platform. In each case, coordination that was informal, unpriced, performed by someone whose labor was invisible to the market, was formalized, priced, and sold.

The management strip is the same pattern, one layer deeper. The coordination that was formal, priced, and performed by employees whose labor was visible to the market, is now performed by AI. The employees are the invisible coordinators of the next round. Their labor was visible but their removability was not, until the AI made the removability legible.

Capital does not stop at one round of enclosure. Each round reveals the next addressable layer.

Marcus saw this. He also saw something else, which he mentioned in a way that was careful enough that I understood he had been thinking about it.

“The cooperative people. In Ohio. They are doing the same thing, just with the savings flowing differently.”

He paused.

“We move faster.”

Speed vs. Durability

#

This is where the architecture note’s deeper argument surfaces, and it is worth stating plainly.

The cooperative model and the PE management strip are competing deployments of the same structural insight. Both use AI to remove the management layer. Both capture the savings from the removal. They differ on one variable: who benefits.

In the cooperative, the workers benefit. The savings flow to the people who do the work. The governance is collective, the ownership is distributed, and the margin that management consumed is returned to labor.

In the PE model, the limited partners benefit. The savings flow to the capital that funded the acquisition. The governance is the fund’s standard terms. The margin that management consumed is converted to return.

The race between these two models is asymmetric.

PE moves faster because it has money, talent, and institutional infrastructure for rapid deployment. A fund can identify, acquire, and transform a company in nine to twelve months. The AI layer is developed centrally and deployed across the portfolio. The operating partners have done this before. The lawyers know the deal structure. The debt is available.

The cooperative moves slower because collective governance is difficult and capital formation is constrained. Assembling a workforce cooperative requires trust, legal architecture, governance design, and the kind of patient capital that institutional investors are not set up to provide. Charlene’s cooperative took eighteen months to organize. The governance meetings are still difficult. The decision-making is slower than any PE operating partner would tolerate.

But the cooperative has a structural advantage that PE cannot replicate: alignment. In the cooperative, the people who own the enterprise are the people who do the work. There is no gap between owner interest and worker interest because they are the same people. There is no extraction because there is no external investor requiring a return. The surplus belongs to the people who generated it.

PE portfolios inevitably produce misalignment. The fund optimizes for exit value. The workers optimize for job security and income. These objectives coincide when the company is growing and diverge when it is not. The management strip exacerbates the divergence because the value creation comes explicitly from eliminating positions, which is to say from a decision that benefits capital and costs labor.

The cooperative does not face this tension. The AI removes the management layer, and the savings flow to the workers. Nobody is eliminated because the workers are the owners and the owners do not fire themselves.

PE is faster. The cooperative is more durable. The race is between speed and durability, and the outcome is not determined.

The Window

#

There is a window, and both sides know it.

If PE can deploy the management strip across enough of the lower middle market before cooperatives establish themselves, the market structure hardens. The transformed companies become the acquisition targets for the next round of capital. The cost structures they establish become the competitive baseline. Any company that still carries management overhead is at a structural disadvantage. The cooperative that forms after the market has been stripped competes against transformed companies whose cost base it cannot match, because the PE-backed company has the same AI coordination at the same cost but with three years of accumulated operational data and an exit-ready governance structure.

If the cooperatives can establish themselves first, the dynamic reverses. The worker-owned company does not need to generate returns for external investors. It can price lower, invest in quality, retain workers, and build the kind of institutional knowledge that the PE strip destroys when it displaces Kevin. The cooperative that is running when the PE fund comes looking for acquisitions is not for sale, because it is not owned by someone who wants to sell.

The window is the time between the structural insight becoming available and the market structure hardening around whichever deployment model gets there first. The Coordination cluster described the cooperative’s version. This essay describes capital’s version. Both are racing.

Marcus knows this. When he said “we move faster,” he was not making a philosophical claim. He was making a competitive assessment.

The question that neither Marcus nor the cooperative organizers can answer is whether governments will tilt the playing field. Sunita’s line item on page forty-seven of the budget document, the one that would establish a pilot program for AI-enabled cooperative infrastructure, is the policy version of this question. If public infrastructure supports cooperative formation, the speed disadvantage narrows. If it does not, capital’s advantage compounds.

I asked Marcus whether the management strip would work if the companies he acquired were already cooperatives. He looked at me as though the question were genuinely novel, which told me something about the limits of what capital’s field of vision includes.

“Nobody has brought me a cooperative to buy,” he said.

There was a pause.

“They wouldn’t sell.”

He said this the way you say something you have only just understood by saying it out loud. A cooperative does not exit. A cooperative persists, because persistence is the point. The PE model depends on the buy-improve-sell cycle. The cooperative does not cycle. It holds.

The management strip works on companies that can be bought. Cooperatives cannot be bought because ownership is distributed among the people who work there, and they do not want to sell because selling would convert their ownership into someone else’s return.

The cooperative is the structure that the management strip cannot reach. Not because it is protected by policy or regulation. Because it is protected by its own design.

What Nora Does Not Say

#

Nora presents the portfolio model to the fund’s investment committee. The math is clean. The returns are compelling. The execution risk is low because the playbook is proven and the AI layer is mature. The committee approves the strategy.

She does not mention Kevin by name. She mentions the headcount reduction, the severance terms, the transition timeline. She uses the language the committee expects: cost rationalization, operational optimization, structural improvement. The language is accurate. It describes the same events that the Coordination cluster describes using different words.

Dale’s company called the AI system a “decision support tool.” Nora’s fund calls the management strip “operational transformation.” Both phrases perform the same function: they make the structural change less visible in the language used to describe it.

Kevin will be told his position has been eliminated due to restructuring. He will not be told that his role has been automated. The distinction matters because “restructuring” implies that the job might exist elsewhere, that the problem is organizational rather than structural, that Kevin’s skills are transferable to a company that has not yet restructured. The truth is that every company in the sector will restructure, because the math is available to every fund, and no fund that sees the math will decline to execute it.

Nora knows this. She does not say it, because saying it would require her to engage with what happens to Kevin after the severance runs out, and engaging with that would require her to engage with whether the structure she is building is good, and “good” is not a category the investment committee uses.

She files the staffing analysis. She starts the hundred-day plan. The AI layer is already installed. The transition will be complete by the end of the quarter.

The trawler is still on Marcus’s windowsill. He has not moved it back to his desk. I notice this but do not mention it.

This is the tenth essay in The Capital View, extending the arc to examine how private equity adapts the structural insight from the Coordination cluster. The original nine essays examined how capital organizes the transition from human-delivered to AI-orchestrated services. This essay asks the next question: what happens when capital applies the same enclosure logic to the management layer of the firm itself. The addressable layer is the PE version of what Dale experienced and Charlene proposed, the same coordination automated, the savings flowing to different people. The essay that follows (TAM-CV.11) examines the competition to own the coordination infrastructure that enables both the management strip and the cooperative, the platform race between capital and the commons. TAM-CV.12 holds the genuine uncertainty about which model prevails. This essay connects to the enclosure of coordination in TAM-CV.07; to the inverted firm in TAM-RIM.6-03, where Dale’s management layer is replaced from within; to the owned factory in TAM-RIM.6-04, where Charlene’s cooperative replaces the same layer from below; to the lock and the unlock in TAM-RIM.6-SYN, where every unlock carries an enclosure within it; to the dissolved middle in TAM-059; and to the distillation thesis in TAM-072, applied here not to professions but to the firm.

References

#

Private Equity and Value Creation

Appelbaum, Eileen, and Rosemary Batt. Private Equity at Work: When Wall Street Manages Main Street. Russell Sage Foundation, 2014.

Kaplan, Steven N., and Per Strömberg. “Leveraged Buyouts and Private Equity.” Journal of Economic Perspectives, vol. 23, no. 1, 2009, pp. 121-146.

Lerner, Josh, et al. “Private Equity and Long-Run Investment: The Case of Innovation.” Journal of Finance, vol. 66, no. 2, 2011, pp. 445-477.

Management, Coordination, and the Firm

Coase, Ronald H. “The Nature of the Firm.” Economica, vol. 4, no. 16, 1937, pp. 386-405.

Graeber, David. Bullshit Jobs: A Theory. Simon and Schuster, 2018.

Williamson, Oliver E. The Economic Institutions of Capitalism: Firms, Markets, Relational Contracting. Free Press, 1985.

Worker Cooperatives and Alternative Ownership

Cheney, George. Values at Work: Employee Participation Meets Market Pressure at Mondragon. Cornell University Press, 1999.

Dow, Gregory K. Governing the Firm: Workers’ Control in Theory and Practice. Cambridge University Press, 2003.

Labor Market Restructuring

Acemoglu, Daron, and Pascual Restrepo. “Robots and Jobs: Evidence from US Labor Markets.” Journal of Political Economy, vol. 128, no. 6, 2020, pp. 2188-2244.

Autor, David H. “Work of the Past, Work of the Future.” AEA Papers and Proceedings, vol. 109, 2019, pp. 1-32.

Enclosure and the Commons

Boyle, James. The Public Domain: Enclosing the Commons of the Mind. Yale University Press, 2008.

Ostrom, Elinor. Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press, 1990.