Consider a company called Outlet

My fifth or sixth fantasy company.

Outlet is auto cash compatible, and offers unified inventory control amongst licensees.  Outlet competes directly with Amazon and Walmart.  Outlet is a protocol driven contract, agreed to by partner firms, like Mervyn's and Macy's and Target and Home Depot and so on.  The protocol requires them to reserve presentation space and inventory space that matches their market, statistically tested.  The Outlet protocol includes the skill to package inventory for delivery, each partner retaining that capability.

The partner firms are free to market and purchase in their sector, but the intelligent protocol analyzer is free to bring in complementary products to use excess flow in their outlets.

So, a company with clothes outlets for the family might get stuck setting up a family sports center, or offer appliances they are not used to. Behind the scenes our AI is running the on-line shop, featuring and co-featuring theirs and its products. In this way, each partner has an observable inventory control agent from which to compute in and outs, locally.

The decision trees used  are arbitrary semantic graphs with probabilities at the nodes, by my definition.  We get to do that because all parties agree on the semantics of retail distribution, they can fill out their contracts accurately, and correct them as needed.

So, we have an agreement on prime separability,  (horses are not groceries), and relative proportion over a set of sectors.  Watson becomes the arbiter. Ad hoc semantic matching gets improved with a probability index, gets matched to real goods flow. With agreement on semantics, Watson goes to town with all the world's abundant products categorized, by agreement ,among partners.  Sort of the world's shopkeeper.

The Outlet Initial Fintech Offering

Outlet needs Watson, so IBM is in.  After that we go for all the regional and mid sized chains.  The Offering has to be coin because we want everyone to use the Outlet discount coin, it is how the franchise manages internal inventory volatility.  Everyone has this, the store managers, uber delivery drivers, department managers and customers.

Store manages see all products, indexed by semantic and congestion, they see all possible in and outs.  The store evolves by contract with their original franchise, the outlet, and local manager decisions.  Al done via decision trees, upgraded and pruned each morn.

Underneath, Watson will keep very close measurements of excess capacity, and keep it optimum, cooperating with the store manager and department heads.