Physical Intelligence: Improving Retail AI Decision-Making With Real-World Data

Retail enterprises across the globe are investing heavily in artificial intelligence to improve inventory management and solve operational challenges. The investments that prove most successful are using real-time data about the items they make, move and sell to automate and optimize manual tasks. The early results speak for themselves: Inditex leveraged item-level data combined with AI to minimize unsold, wasted inventory — 0.57 percent at the end of 2024, compared with 10 percent to 20 percent industry averages. For other retailers, the question is now whether their in-store and supply chain operations are generating the item-level data their AI systems need to unlock similar optimizations.

Enterprises are discovering that their AI systems are only as effective as the real-world data that powers them. Without item-level visibility and traceability, what’s actually happening inside stores and warehouses remains a blind spot for AI systems: AI can only guess as to what’s happening in the physical world, unable to optimize what it can’t see.

Providing retail AI systems with trusted, real-time item data requires wirelessly connecting items using RAIN RFID tags that allow enterprises to see what items they have and track when and where these items are moving. Zara, Nike, Walmart, Macy’s, Kroger, Chipotle, UPS, H&M, Uniqlo and many more utilize RFID solutions to provide this accurate real-world data on tens of billions of items every year. This data about items and their movements connects the physical world to digital AI systems, enabling true physical intelligence across store floors, back rooms, and within the supply chain.

Moving Beyond Error-Prone, Periodic Inventory Snapshots

Retailers previously relied on periodic, manual inventory counts to update their systems. However, modern retail doesn’t operate periodically: It’s constantly in flux as items arrive, move, sell in-store and online, get stolen, and get returned. The gap between what systems show and reality is called inventory distortion, leading to stockouts and overstocks that cost retailers $1.7 trillion worldwide.

Advances in RFID are helping retailers move beyond periodic inventory snapshots toward more frequent and accurate item-level visibility. By automating inventory processes and capturing data about items as they move through stores and back rooms, retailers can identify missing or misplaced merchandise sooner, reduce inventory blind spots, and provide AI systems with more timely and reliable real-world data.

Physical intelligence enables more than faster inventory counts. Retail AI systems gain accurate, current data that informs what happens next.

From Seeing Inventory to Acting on it

Item-level traceability enables superior replenishment, fulfillment and management of retail operations. Retail AI systems can flag inventory shortages or surpluses sooner, restock shelves before limited supplies become completely out of stock, and determine where merchandise needs to move based on exact quantities demanded.

Today’s customers expect to get what they want when they want it — with frictionless transactions — or quickly shift to other stores and brands. While inaccurate availability data leads shoppers and employees to hunt for items that are actually out of stock, physical intelligence enables inventory visibility and traceability, empowering retailers to quickly locate products and fulfill orders. Customers get what they want without needing to understand the technology.

Over time, every piece of retail merchandise will be visible and traceable in real time, at scale, across thousands of stores, warehouses, and fulfillment centers. Physical intelligence will give retail AI systems the reliable data streams they need to see what's missing before it matters, act on what's moving before it's lost, and reduce waste to near-zero levels. Replenishment, fulfillment, and other everyday retail operations will benefit from automations that save labor costs and improve availability. The result will be more satisfied customers and less financial waste from overstocking or unnecessary replenishment.

Matt Branda is vice president of product and solution marketing at Impinj, a leading RAIN RFID provider and Internet of Things pioneer.