Retail Operations · Frontline AI
Twenty-eight years leading large-format retail operations. I build the systems, and the operator capability, that make AI actually land on the floor.
28 yrs operations700-person storeFull P&L
Most retail AI pilots do not fail because the model is weak. They fail because no one owns the translation between the AI and the frontline workflow, and because the scoreboard used to judge that work was built for a pre-AI store.
Name the role, the Workflow-First AI Leader™, and rebuild the measurement, The Operator's Scoreboard™. That is the prerequisite for the J-curve to bend. Start with the work, not the model.
The book behind the framework — twelve chapters on making AI land on the floor. Chapter 1 is free to read; the rest are available as early access. Pre-publication draft
A six-month AI pilot dies in a single shift — and nobody touched the model.
Four real explanations, none sufficient. The residual is the work no one owns.
AI’s first job isn’t to do the work. It’s to finally let you see it.
Diagnosis begins where the knowledge actually lives — on the floor.
Earn the right to see the real work, then make your method explicit.
Intent dies in translation. The fix isn’t willpower; it’s artifacts.
Re-rank what you measure, or miss AI landing and stalling entirely.
Ad-hoc, Mapped, Validated, Codified, Compounding — and knowing which you are in.
The role retail AI requires, defined — and why the org chart lacks it.
The gate is the price of shipping. A real gate produces a documented no.
A ninety-day sequence you can start Monday, without anyone’s permission.
Spend your attention on the unbuyable 80%. That is the whole method.
© 2026 Jordan K. Jones. All rights reserved. Pre-publication draft of Workflow-First: The Operator’s Field Guide (forthcoming).
An AI ordering system recommends the daily produce buy. The veteran orderer overrides it for a cold front and a stadium match nearby, and on those days she is right. The override is not an exception to suppress. It is the variable the forecast never had.
A shelf scanner flags gaps for replenishment. A third are a planogram change the system has not caught; the shelf is not empty. The scoreboard counts flags closed. The real job is the availability the partner's triage protects. Those are not the same number.
I have spent twenty-eight years in large-format grocery retail operations, including leading a 700-person store operation with full profit-and-loss responsibility. I can cite Brynjolfsson, Nonaka, and Edmondson in one paragraph and name the produce-forecasting workflow that broke on a Tuesday in the next.
The frontline is where my work starts and where it is tested. Adoption is not coerced. Visibility is not weaponized. The people closest to the work keep their dignity.
Whether you are building a retail AI capability or hiring someone to lead operations, the work is the same.