The store is the constraint

XY Retail already runs 1,300 stores in more than 50 countries. The merchandising, the demand, the supply chain and the personalisation already have owners. What they did not have was AI that could sit on that stack without becoming a demo nobody used on the shop floor.

I am in that work now. It is not a closed case. The test is not whether the demo landed. It is whether a store can use it on a Tuesday.

The miss I am watching for

I made this miss on the lending line: I treated the platform half as the work, when the constraint was the people who had to live with a write they did not design. Fashion will punish that faster. A merchandiser who does not trust a recommendation will not argue with it. They will ignore it and go back to the number they already trust.

So operations and merchandising stay in the room from week one. The gate is the same one I use everywhere else: a write waits for a person. There is no maintenance window for a clever rewrite. The stores are open.

What “sit on the stack” means

Not a new system next to the old one. Not a workspace that only headquarters can log into. A path that already exists — demand, allocation, a store action — with an approval sitting on the write that used to be a person guessing from a report.

If the store still has to translate the output into the tool they actually use, we built a briefing, not a capability. I will take the duller path that lands in their screen over the smarter one that lands in ours.