The starting point
One person rebuilt each changed order by hand, copying lines from email, reissuing documents and checking a separate stock file. Each change took days to reconcile, invoices waited until well after shipment, and errors surfaced through buyer queries. Nobody had time to look at which stockists were selling through.
What we built
Order change intake
An AI agent reads buyer emails, matches each change to the confirmed order and drafts the update. Requests it cannot match go to a named reviewer.
Documents from one record
Packing lists, commercial invoices and customs details are generated from the updated order, so every document shows the same lines, quantities and prices.
Invoice readiness check
When goods ship, the system compares shipped quantities with the order and prepares the invoice. Mismatches go to the wholesale lead with the reason attached.
Reorder signals
Stockist sell-through reports are read every week. Retailers selling fast get a prepared reorder suggestion for the wholesale lead to send.
The result
Changes now arrive matched and checked, with the packing list and invoice already updated, and nobody retypes an order. The wholesale lead approves each one before it goes to the buyer.
- Checked on arrivalEach buyer change is matched to its order and checked before anyone opens it.
- Invoices at dispatchThe invoice is prepared from the shipped quantities as the goods leave.
- Reorders flaggedStockists selling through quickly get a prepared reorder for the wholesale lead to send.
Where people stay involved
Nothing goes to a buyer or a carrier until a person approves it. Every change keeps the original email, the proposed update and who approved it. The wholesale lead handles commercial exceptions, and engineers review recurring failures and maintain the rules.
