Retail AI has a reality problem
It happens all the time - the inventory system says there are six units in-store.
Replenishment has nothing to flag. E-commerce tells shoppers the item is available. But the shelf is empty, and no one knows the product is simply in the wrong place.
The technology hasn't necessarily failed. Those six units are in stock, but the system is missing something fundamental: the product isn't where shoppers can actually find it.
A store can be digitally correct and operationally wrong.
After spending much of my career focused on improving retail operations, I’ve learned how easily this disconnect can be overlooked—and how much more consequential it becomes as retailers rely on AI to guide more of their operations.
Inventory, pricing, e-commerce, and workforce management systems tell us where merchandise should be, which promotions are running, and what needs attention.
But anyone who's spent time on a sales floor knows how quickly that picture can fall apart. Displays are set incorrectly. Promotions go live without the right signage. Products are in the building but nowhere to be found on the shelf.
Retailers are investing in increasingly sophisticated technology to make decisions, and yet many still lack an accurate picture of what's happening inside their own stores.
The reality is that shelf data has a shelf life, too. Products move. Deliveries arrive. Online orders pull inventory from the floor. Spills happen.
Historically, retailers have managed these changes through aisle walkthroughs, cycle counts, and associate reporting. These processes still matter, but they only offer periodic snapshots of an environment that's constantly changing.
I've seen how much time and effort store teams dedicate to finding problems before they can even begin solving them.
Physical AI can help change that.
Fixed sensors, computer vision, robotics, and related technologies provide a more continuous view of inventory, product placement, and other changing store conditions.
The opportunity isn't simply collecting more data. Retailers have plenty of that already. It's about connecting what our systems think is happening to what's actually happening, giving store teams the intelligence to act.
Sometimes the real problem isn't where you think it is.
Take McCoy's Building Supply, for example. A specific feed bucket kept showing up as out of stock. At first glance, it looked like a routine shelf execution issue.
The pattern revealed something different: a vendor shortage.
What appeared to be a shelf problem was really a sourcing problem. McCoy's ultimately identified another supplier offering the same product at a better price, improving both availability and cost.
That's what makes this kind of intelligence so valuable.
Is demand really weakening, or is a top-selling product consistently unavailable? Is inventory misplaced, or was it never delivered?
These are very different problems. Physical AI can help retail teams make those distinctions and identify patterns that isolated reports and manual scans often miss.
For shoppers, "in stock" or "available for pickup" is a promise. When the shelf tells a different story, they don't care about why or which system was right. They care that the retailer didn't deliver.
And in a world where shoppers have less time, higher expectations, and more choices than ever, every unfulfilled promise chips away at something retailers spend decades building: customer loyalty.
I believe the next competitive advantage won’t be about which retailers adopt the most AI. It will come down to who uses AI to make their stores work better.
Physical retail will always be messy. Thousands of products, constant movement, changing priorities, and people doing their best to keep everything running. We can't eliminate that complexity, but we can give retailers a much clearer view of it.
When the store changes but the system doesn't, every decision that follows risks being based on yesterday's reality.
And no matter how intelligent our technology becomes, it can't solve a problem it doesn't know exists.
Chris Green is chief commercial officer, Badger Technologies.



