GroceryShop 2026 was the AI show. Not AI as a future conversation but AI as an active investment retailers are already trying to operationalize. The boardroom commitments are made. The urgency now is whether the data infrastructure can hold the weight.
One line on our booth screen kept starting conversations: “Your AI is only as good as your product data.”
Nearly every retail leader who stopped by had the same story. An AI roadmap that looked ready in the boardroom. A catalog that wasn’t. Search, personalization, and retail media all waiting on the same missing attributes.
Hy-Vee CEO Jeremy Gosch laid out a specific and public goal: cut the average online grocery order — currently around 40 minutes — down to five minutes, powered by AI cart preloading that anticipates what shoppers need before they open the app. Gosch confirmed Hy-Vee is in a ground-up rewrite of its digital properties, with beta testing of the new capabilities starting within the next month.
What makes a five-minute shop possible is not the app. It is the completeness and accuracy of what sits underneath it.
01 — The AI roadmap is approved. The catalog is not ready.
Almost every team we spoke with had greenlit an AI initiative in the last 12 months. Very few had audited the product data those initiatives would run against. Search accuracy, personalization, and retail media all depend on attribute completeness at the SKU level. The gap between the AI roadmap and the catalog reality is where most projects stall.
02 — Platform-bundled enrichment is not the same as enrichment.
A number of ecommerce platforms are now bundling some form of catalog enrichment into their base contracts. Retailers are discovering that enrichment scoped to enable search is a different product than enrichment scoped to enable substitutions, personalization, and AI-assisted shop. Attribute depth matters. The distinction is showing up in production.
03 — Real-time sync is no longer a roadmap item.
Digital shelf investment is accelerating. Chain-wide rollouts mean real-time product attribute synchronization is now an operational requirement, not a future-state aspiration. The retailers further along in this conversation are the ones whose product data can move as fast as the shelf.
The question retailers kept asking on the floor is the right question: what is the one data gap standing between your team and your next AI launch? In most cases, they already knew the answer. They were looking for a credible path to close it.
Those conversations are why this team spends time on the show floor. We learn more in two days of listening than in a month of guessing.