Merchandising Is Intent Engineering. It Always Has Been.
What a 25-year career in product lifecycles taught me about the four AI disciplines everyone is suddenly talking about.
Twenty-five years. Heritage brands. Lifestyle brands. Founder brands built on a point of view — the kind where the product carries a story and the customer knows it.
Recently I wrote about context engineering — how the brands that win are the ones who structure what they know so it’s actually usable. This is the layer above that.
So when I watched Nate B Jones’ latest video — 10x AI Users Aren’t Smarter Than You. They Just Know Something You Don’t. — I didn’t just nod along as a technologist. I recognized the problem from the other side of the business.
He lays out four disciplines that are redefining what it means to work with AI:
Prompt Craft — the original skill. Table stakes now.
Context Engineering — curating the right information environment for the AI to operate within.
Intent Engineering — encoding what the system should want. Goals, values, trade-off hierarchies, decision boundaries.
Specification Engineering — writing your entire organizational knowledge base in a way that autonomous agents can execute against it.
He uses Klarna as the proof case for why intent engineering matters:
“Their AI agent resolved 2.3 million customer conversations in the first month — but it optimized for the wrong thing. It slashed resolution times, but it didn’t optimize for customer satisfaction. Klarna got into big trouble, had to rehire a bunch of human agents, and is still dealing with the customer trust aftermath.”
— Nate B Jones
And that’s where it landed for me.
That’s a bad buy.
Any merchant who has ever watched a fast-moving category cannibalize a brand’s integrity knows exactly what that feels like. You hit your turns. You hit your sell-through. The numbers look great — right up until the customer doesn’t come back. Not because the product failed. Because the brand failed them. Because somewhere in the hierarchy of what we were optimizing for, trust got left off the list.
The Klarna AI didn’t fail because it was dumb. It failed because nobody encoded what the brand was actually for.
Merchandising has always been a four-layer stack.
Prompt Craft is the product brief. Clear specs, clear descriptions, clear intent for the factory.
Context Engineering is the assortment. The right product in the right door, the right season, the right customer moment — curating the information environment so the customer can act.
Intent Engineering is the edit. The merchant’s point of view. Not just what sells, but what the brand is optimizing for. How the customer should feel about the brand after they buy.
Specification Engineering is the brand’s entire product philosophy — written clearly enough that a new team member, a new vendor, or now an AI agent could execute in the spirit of the brand without you in the room.
The best merchants I’ve ever worked alongside were doing all four of these things, intuitively, long before anyone called it prompting.
The worst AI deployments I’ve seen — and the worst buys I’ve seen — fail for the same reason: someone skipped intent. They measured what was easy to measure. They forgot what they were actually building toward.
Why this matters right now.
Nate makes the point that as AI agents run longer and more autonomously — hours, days, weeks — the cost of poor intent engineering scales with them. A bad prompt wastes a morning. Bad intent engineering at the organizational level produces Klarna.
But here’s the flip side: the people and organizations who get intent right — who can clearly encode what they stand for, what they’re optimizing for, and where the lines are — those are the ones whose AI investments will compound.
For those of us who have spent careers building brand-right product at meaningful companies, this is not a new skill to learn. It’s a familiar discipline, applied to a new medium.
The edit has always been the job.
Inspired by Nate B Jones and his excellent breakdown: 10x AI Users Aren’t Smarter Than You. They Just Know Something You Don’t. — worth every minute.