Where agentic meets reality.
We built models to predict grocery baskets. The better they got, the smaller the baskets became. Google, Meta, and now OpenAI have tried to bolt shopping onto their platforms. The disruption never lands. The gap is not compute power. It is a misunderstanding of how people actually shop.
The vision of everyone vibe-coding their own tools sounds productive. Until you multiply it by a few thousand employees. Then you don't have innovation. You have chaos, security nightmares, and a fundamental misunderstanding of what makes people successful in organizations.
Dark kitchens bypassed dining rooms. AI dark stores will bypass websites. When AI agents become your primary shoppers, they never see your homepage or brand story. They parse data: price, specs, delivery time. Retail is about to learn what restaurants did: experience is a premium, not a given.
Retailers and tech giants race to define AI shopping. Amazon lists products from other stores without asking. Walmart sells ads in AI assistants. Microsoft adds shopping to Copilot. Google aims to standardize the industry. The rules are being written with real customers, right now.
Grocery retailers have a decisive advantage in AI shopping: their data. Generic AI agents struggle with product discovery, but retailers possess years of purchase history revealing individual tastes. The race is about combining targeted AI with proprietary data.
AI shopping agents face a fundamental problem: the less you see, the less you buy. By focusing on efficiency, they eliminate the product discovery that drives grocery sales. You can't ask for what you don't know exists, and impulse purchases represent over half of grocery revenue.