Item J: the entry point
Every demand signal needs a place to land. In the seed-treatment network that entry point is Item J — the packed, coated, primed, calibrated seed that is shipped from the Central Warehouse to the customer via the final BOD lane.
A forecast of 100 units of Item J in a given planning period is the starting assumption for everything downstream. Until that number exists, the planning engine has no reason to schedule Packing, Coating, Priming, or any upstream process.
Forecast vs. demand plan
The raw forecast from the ML model is a probability distribution — for example, “90 % confidence the customer will need between 85 and 115 units.” The demand plan is the single number committed to by the commercial team: 100 units. This is the number that the planning engine uses.
In o9 the demand plan lives on the Item J line item at the customer location. A line item is a (item, location, time) combination, and the demand plan is one of its key attributes alongside on-hand inventory, safety stock, and lead time.
The parent line item relationship
Item J at the customer is a child demand node. The planning engine traces back through the BOD edge (Ship [J] process) to find Item J at the Central Warehouse — the parent line item. The 100-unit forecast propagates to the parent: the Central Warehouse must have 100 units ready to ship.
This parent → child relationship repeats all the way up the network. Each step upstream creates a new parent that must be planned to satisfy the child.
Time offsets begin here
The customer expects delivery by a specific date. The Ship [J] process has a lead time — say, 3 days. So the Central Warehouse must have Item J ready 3 days before the customer needs it. Every hop upstream shifts the required start date back by the process lead time, creating a cascade of time-offset requirements across the whole network.