The Direct Chain

TAM-RIM.6-05 · The Reimagined, Cluster 6: The Coordination · The Approximate Mind

The road between Tirupur and Coimbatore is thirty-seven miles of trucks.

Cotton trucks heading to the mills. Yarn trucks heading to the knitting units. Fabric trucks heading to the dyeing houses. Finished garment trucks heading to the export houses. Each truck carries material from one stage of a process that turns raw cotton into a t-shirt, and each stage has a business that employs people who know their piece of the process with a specificity that looks, from outside, like it could be replaced, and from inside, like it could not.

Ravi’s mother runs a knitting unit. Fourteen machines, eight workers, a concrete building with a corrugated roof on a lane off the Avinashi Road. She has been doing this for twenty-two years. She knows which machines can handle 40-count yarn without adjustment and which need the tension recalibrated. She knows which of her workers is best on the collar rib and which one can recover a dropped stitch fast enough that the production loss is negligible. She knows these things because she has been paying attention for twenty-two years, and the knowledge lives in her hands and her eyes and her judgment about when to push and when to let a problem resolve itself.

She sells her output to an export house in Tirupur. The export house consolidates production from dozens of units like hers, manages the quality standards for international buyers, handles the export documentation, arranges the shipping, negotiates the contracts, absorbs the currency risk. The export house takes a margin. The consolidator who connects her to the export house takes a margin. The buying agent who represents the American brand takes a margin. The brand that puts its label on the shirt takes a margin. The distributor who moves the labeled shirt to the retailer takes a margin. The retailer takes a margin.

Ravi’s mother receives approximately three dollars for a shirt that sells for thirty in a store in Columbus, Ohio. She does not know what the shirt sells for. She has never seen the store. The chain between her knitting unit and the consumer is seven intermediaries long, and each intermediary is performing a function that is, at its core, informational: knowing something that the parties on either side of it do not know, and charging for that knowledge.

Ravi is twenty-three. He studied computer science at Anna University in Chennai and came back to Tirupur eight months ago, to his mother’s initial confusion and his father’s quiet satisfaction. He came back because he could see something that the seven intermediaries could not see, or could see but had no incentive to name: that every function they performed was an information processing function, and information processing was no longer scarce.

The Toll Booths

#

The supply chain between Ravi’s mother and the consumer in Columbus is a series of toll booths. Each intermediary controls access to a connection that the parties on either side cannot make on their own. The manufacturer cannot reach the consumer. The consumer cannot find the manufacturer. Everyone in between profits from that mutual ignorance. The intermediary does not create the shirt. The intermediary does not wear the shirt. The intermediary stands between making and wearing and charges for the passage.

Some of this charging is legitimate. The export house manages genuine complexity: customs regulations, quality certifications, shipping logistics, buyer negotiations. These require knowledge, relationships, and operational capability. The export house earned its margin by doing things the manufacturer could not do alone.

But the legitimacy of the function does not make the function permanent. The knowledge the export house holds, regulations, certifications, logistics protocols, buyer requirements, is codifiable. The relationships the export house maintains, with shipping companies, customs brokers, certification agencies, are transactable. The operational capability the export house provides, consolidating orders, managing quality, coordinating delivery, is a coordination function.

Every toll booth in the chain is an information asymmetry that charges rent. AI does not eliminate the information. It eliminates the asymmetry.

What Ravi Built

#

Ravi did not set out to disintermediate his mother’s supply chain. He set out to solve a specific problem: the export house was late paying for the last three shipments, and his mother was covering the gap by borrowing from a local moneylender at rates that ate most of her margin. The cash flow problem was a symptom of a structural dependency: his mother had one buyer, the export house, and the export house had all the leverage because it controlled access to the international market.

He built the AI layer in stages over six months, working from a desk in the room above the knitting floor, testing against real production data, failing in ways that taught him things no computer science curriculum could.

The procurement module monitors cotton and yarn prices across suppliers, identifies quality-price combinations that match the unit’s machinery capabilities, and negotiates purchase terms directly. It replaced the intermediary his mother used for raw material sourcing, a man named Prakash who had been taking a seven percent markup for the service of knowing which suppliers were reliable. The AI learned which suppliers were reliable within three months of transaction data. Prakash’s twenty years of knowledge was valuable, genuinely, and also reproducible.

The production coordination module connects his mother’s unit with forty-nine other small manufacturers in Tirupur. Each unit has different capabilities, different capacity at any given time, different strengths. The AI allocates orders across the network based on current load, skill match, and delivery timeline. This is the function the consolidator performed: knowing who could make what and when. The AI does it with more precision and without the consolidator’s margin.

The compliance module handles export documentation, customs requirements, quality certifications. These are rule-based processes with high detail and low ambiguity, exactly the kind of work AI handles better than humans because the penalty for a small error is disproportionate and AI does not have tired Fridays.

The direct-to-consumer module is the part Ravi is most uncertain about. A website, digital marketing, the shopping experience, pricing strategy. He built this last because it is the part that faces the consumer, and the consumer is where the intermediary chain captured the most value. The brand. The retailer. The story that turns a three-dollar shirt into a thirty-dollar shirt. The story is not information processing. It is something closer to identity, and identity is harder to automate than logistics.

He has not solved this. The shirts sell on the website for twelve dollars. The manufacturers receive nine. The consumer pays less than half what the store in Columbus charges. But the volume is small. The brand recognition is zero. The consumer in Bangalore who buys from the cooperative’s website is buying on price, not on story, and price competition is a race the cooperative will eventually lose to someone with lower costs or better AI.

Ravi knows he needs a story. He does not yet know what it is.

The Domestic Frame

#

Here is where the standard narrative about supply chain disintermediation goes wrong, and where Ravi’s instinct goes right.

The standard narrative is about exports. The t-shirt made in Tirupur for the consumer in Columbus. The supply chain that crosses oceans. The trade agreements and tariff structures and currency risks. The intermediaries positioned along the international chain, each one taking a cut as the product moves from the Global South to the Global North.

Ravi is not interested in Columbus. He is interested in India.

India is 1.4 billion people. The domestic garment market is enormous, fragmented, and intermediated at every level. Between the manufacturer in Tirupur and the consumer in Lucknow, there are regional distributors, wholesale markets, retail chains, and local shops, each adding a margin, each controlling access to a connection the manufacturer cannot make alone. The three-dollar shirt sells for fifteen or twenty in a shop in a tier-two city, and the intermediation, though shorter than the international chain, is proportionally just as extractive.

The domestic market does not require export compliance, customs documentation, international shipping, or currency hedging. It does not require navigating American trade policy or competing with brands that spend more on advertising than Ravi’s entire cooperative earns in a year. It requires reaching Indian consumers who want affordable, well-made clothing and are increasingly comfortable buying online.

India has the infrastructure for this. UPI processes billions of transactions per month with near-zero friction. ONDC, the Open Network for Digital Commerce, is a public protocol that allows any seller to reach any buyer without going through a private platform. Aadhaar provides universal digital identity. The logistics networks, built to serve the e-commerce boom, reach tier-two and tier-three cities with delivery times that would have been unimaginable a decade ago.

The rails exist. What has not existed is a producer collective with an AI coordination layer riding on them.

Ravi does not need Amazon. He does not need Flipkart. He does not need a brand or a distributor or a retail partner. He needs the AI layer he is building, the public digital infrastructure the government has already built, and a product that is good enough to sell on its own terms to consumers who have been paying intermediary margins their entire lives without knowing it.

The consumer in Lucknow who buys a twelve-dollar shirt from the cooperative’s website, a shirt that would cost twenty in the local market, does not need to know about supply chain theory or toll booth economics or the political philosophy of worker ownership. She needs to know the shirt fits, the quality is good, and the price is right.

The rest is invisible. As it should be.

The Ownership Distinction

#

There is a version of this story that has already happened, and it ended differently.

Tech platforms have been disintermediating supply chains for two decades. Amazon. Alibaba. Flipkart. Each one eliminated intermediaries between manufacturers and consumers. Each one reduced the number of toll booths in the chain. Each one lowered prices for the consumer and increased market access for the manufacturer.

And each one became the new intermediary.

Amazon did not eliminate retail intermediation. It became retail intermediation. The manufacturer who used to depend on seven intermediaries now depends on one, and that one controls discovery, pricing, reviews, fulfillment, and the consumer relationship. The toll booths were removed and replaced by a single, larger toll booth with better technology and higher walls.

The distinction between what the platforms did and what Ravi is building is ownership.

When a platform disintermediates a supply chain, the platform captures the coordination value. The manufacturer gets better market access but loses control of the customer relationship, the pricing, the data. The consumer gets lower prices but becomes dependent on a platform whose incentives are not aligned with theirs.

When a producer cooperative disintermediates a supply chain through an AI layer it owns collectively, the coordination value stays with the producers. The manufacturers control the customer relationship. They set the pricing. They own the data. There is no platform between them and the buyer, only the AI layer that they collectively own and that operates on public rails rather than proprietary infrastructure.

This is not a subtle distinction. It is the distinction between the cooperative and the corporation, applied to the digital economy. And it has a specific consequence: the value that the intermediary chain used to extract, the twenty-seven dollars between the three-dollar shirt and the thirty-dollar price tag, does not flow to a new intermediary. It flows partly to the producer, as higher income, and partly to the consumer, as lower prices. The margin that used to sustain seven businesses now sustains the people who make the thing and the people who use it.

What Is Lost

#

The intermediaries were not only extracting value. They were also providing functions that the cooperative must now provide for itself.

The export house managed quality. Not just checking the product but establishing the standards, communicating them to the manufacturers, rejecting work that did not meet them, and bearing the reputational cost when quality failed. The AI can monitor quality metrics. It cannot develop the taste, the judgment about what “good enough” means for a specific market, that an experienced quality manager brought.

The brand told a story. The story may have been manufactured, the marketing may have been manipulative, the premium may have been unjustified by any material difference in the product. But the story solved a real problem for the consumer: the problem of trust. The consumer in Columbus who bought the branded shirt was paying for the assurance that someone had vouched for the quality, the sizing, the consistency. The cooperative must build this trust from nothing, in a market where trust is expensive and attention is scarce.

The distributor absorbed risk. When demand dropped, the distributor held inventory. When a shipment was damaged, the distributor negotiated with the insurer. When a buyer defaulted, the distributor absorbed the loss. The cooperative, without a distributor, absorbs these risks collectively, and collective risk absorption requires collective risk tolerance, which is another governance problem added to the governance problems the previous essay described.

Ravi is learning these things in real time. The AI handles the information processing. The functions that are not information processing, taste, trust, risk tolerance, judgment about what quality means in a market you have never visited, these are the functions that require something the AI does not have: the accumulated experience of operating in the space between maker and buyer for long enough to know what each side actually needs, as opposed to what each side says it needs.

The intermediaries were inefficient. They were also experienced. Eliminating them eliminates both.

Fifty Families

#

The cooperative now includes fifty manufacturing units. Some are as small as Ravi’s mother’s operation, eight workers on fourteen machines. Some are larger, thirty or forty workers with newer equipment. Together they employ roughly four hundred people, which is close to the number the Lordstown plant employed at peak.

The difference is that these four hundred people work in fifty different buildings, on fifty different lanes, in a city where the garment industry is the economy and the economy has been squeezed by the same forces that squeezed Lordstown: global competition, margin compression, the steady migration of production to places where labor is cheaper and regulations are lighter.

Ravi’s AI layer did not save the Tirupur garment industry. The industry is too large and too complex for one twenty-three-year-old and a coordination algorithm. What it did is give fifty families a different relationship to the market. Instead of selling their output to an export house at whatever price the export house offers, they sell through a channel they collectively control, at a price that reflects the value of the product rather than the leverage of the buyer.

The income increase is real. Not dramatic. The manufacturers who were receiving three dollars now receive eight or nine. The increase is not evenly distributed, because the AI allocates orders based on capability and capacity, which means the better-equipped units get more volume. This creates internal tensions that the cooperative’s governance structure, such as it is, must navigate.

I wonder whether the uneven distribution will be the thing that breaks it. Whether the cooperative can sustain solidarity when the AI’s allocation logic produces outcomes that feel unfair to the units that receive less, even if the logic is transparent, even if the allocation is optimized for the collective’s overall performance. Fairness and optimization are not the same thing, and the gap between them is where cooperatives have historically fractured.

Ravi does not have an answer to this. He has a dashboard that shows each unit’s production, revenue, and allocation. He has meetings, which are longer than he expected and more difficult than any technical problem he has solved. He has his mother’s judgment, which he relies on more than he admits, because she has been navigating the relationships between manufacturing units in Tirupur for twenty-two years and she knows things about those relationships that no algorithm can model.

He has the desk above the knitting floor. The sound of the machines comes through the floor. His mother is downstairs, inspecting a collar rib with her hands, finding a flaw that the quality sensor missed, correcting it with a gesture so practiced it looks automatic and is not.

The chain is shorter. Whether it holds is something only time will answer.

This is the fifth essay in The Coordination, a cluster within The Reimagined examining what happens to the structure of the firm when AI can perform the coordination function. The previous essays traced the one-person firm (TAM-RIM.6-01), the zero-person firm (TAM-RIM.6-02), the inverted firm (TAM-RIM.6-03), and the worker-owned factory (TAM-RIM.6-04). This essay extends the proposition across the supply chain, asking what happens when producers own the AI coordination layer that connects them to consumers. The essay that follows (TAM-RIM.6-06) asks what happens when the workforce itself becomes fluid, assembled and disassembled by AI for specific projects. This essay connects to the toll booth economy in TAM-033 and TAM-051; to the monoculture in TAM-050, where AI recommendation systems destroy the habitat for small-scale economic life and this essay asks whether AI coordination might rebuild it under different ownership; to the enclosure of coordination in TAM-CV.07, here inverted because the producers enclose the coordination rather than capital; and to the distillation thesis in TAM-072, applied not to the profession but to the supply chain itself, stripped to the irreducible connection between the person who makes the thing and the person who uses it.

References

#

Supply Chain Economics and Intermediation

Gereffi, Gary, and Karina Fernandez-Stark. “Global Value Chain Analysis: A Primer.” Center on Globalization, Governance and Competitiveness, Duke University, 2nd edition, 2016.

Milberg, William, and Deborah Winkler. Outsourcing Economics: Global Value Chains in Capitalist Development. Cambridge University Press, 2013.

Rivoli, Pietra. The Travels of a T-Shirt in the Global Economy. John Wiley and Sons, 2005.

Indian Manufacturing and Digital Infrastructure

Nilekani, Nandan. Imagining India: The Idea of a Renewed Nation. Penguin Press, 2009.

Raghavan, Srinath. The Most Dangerous Place: A History of the United States in South Asia. Penguin Books, 2018.

Kurien, Verghese. I Too Had a Dream. Roli Books, 2005.

Platform Economics and Market Power

Khan, Lina M. “Amazon’s Antitrust Paradox.” Yale Law Journal, vol. 126, no. 3, 2017, pp. 710-805.

Srnicek, Nick. Platform Capitalism. Polity, 2017.

Zuboff, Shoshana. The Age of Surveillance Capitalism. PublicAffairs, 2019.

Cooperative Economics and Collective Ownership

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

Restakis, John. Humanizing the Economy: Co-operatives in the Age of Capital. New Society Publishers, 2010.

Zamagni, Stefano, and Vera Zamagni. Cooperative Enterprise: Facing the Challenge of Globalization. Edward Elgar, 2010.