A shop becomes one node in a larger network.

The formal business base alone is enormous. The Ministry of MSME said in July 2026 that more than 8.9 crore (89 million) enterprises were registered on the Udyam Registration Portal and the Udyam Assist Platform^[1]. The evidence review is careful about what that number means. It shows formalisation at scale, but it is not a count of retail businesses and does not measure the informal economy. Moving from “a large MSME base” to “a large, digitally invisible local supply” is plausible but unmeasured.

Visibility is also not binary. Many local businesses are partly digitised: they accept UPI, use WhatsApp and appear on a map, yet have no structured inventory, availability, service attributes, credentials or policies that an agent could query. ONDC’s seller side already accommodates distributed supply. Seller apps connect sellers to the network, digitise their catalogues, disburse payments and train them. Marketplace Seller Nodes offer goods and services from many sellers, while Inventory Seller Nodes are sellers offering their own inventory^[2]. So “put local stores online” is not the thesis; that already exists. The deeper question is whether the network can represent what those businesses are able to do, not only what they list.

Centralised commerce aggregates inventory, and dark stores do it very efficiently. A network of nodes would try to aggregate capability instead: we may be able to aggregate commercial capability without first aggregating physical inventory.

It is tempting to put a number on the infrastructure this avoids: the retail space, inventory, warehouse capacity, staff, equipment and fulfilment assets needed to offer the same breadth centrally. That number is a modelled counterfactual, not an observation. It should be reported as a range with its assumptions stated, never as a headline on its own.

Some missions will need several nodes at once. Converting a spare room into a home office might draw on an interior designer, a furniture workshop, an electrician, an acoustic specialist and a network installer. Composition like this should come later, because it raises the hardest accountability question in the model: who owns the outcome? Shift 8 takes that up.

Distributed brand networks already exist as dealerships, franchises, authorised service centres, distributor networks and partner ecosystems. The closest comparator is Urban Company, which describes a full-stack model. It trains and certifies professionals, with more than 300 full-time trainers and Skill India-certified credentials through NSDC. It gives them access to tools and consumables through its procurement network, and supports them with insurance and access to credit^{[3, 4]}. It does all this inside a single, centralised platform. The difference proposed here is structural, and it cuts both ways.

Reach and portability favour the open version. Single-party accountability favours the managed one.

“I need an AC installed tomorrow” might bring up an independent technician, an authorised brand technician, a local appliance retailer with a technician, a multi-brand service specialist and a brand-authorised local node. The consumer chooses on expertise, authorisation, availability, distance, price, warranty and past work. The network doesn’t decide that one supply model is superior; it makes the alternatives visible. “Brand-authorised” only means something if the brand attests to it in the trust layer, so brand nodes depend on brand cooperation from day one.

Manufactured supply is a different business. Recruiting, training, equipping and financing operators is franchise operations, with its own economics and quality risk. Someone has to fund equipment and working capital, whether the brand, the operator or a lender. This belongs after the discovery layer has proven itself, because trying to validate both at once would blur both results.

The brand’s incentive is unproven. Brands usually want control of price, warranty and the customer relationship, and they may not want their nodes listed beside independents where consumers compare. They might agree in order to reach districts without stores, lower acquisition costs or cover service for an installed base. This should be tested with a single brand partner.

Large brands could dominate. Thousands of brand nodes could turn the network into another distribution channel for big companies. The safeguards need to be mechanisms, not principles. Authorisation should work as a filter the user can choose, never as a ranking boost, and shortlist placement should not be for sale. Ties should rotate, node reputation should be portable, and brand nodes should interoperate without pulling the transaction into a proprietary system.