If you had asked a warehouse manager five years ago about their company’s inventory management practices, you’d probably have heard a lot about the frequency of cycle counts—both full and partial. The answer also might have included particulars on the equipment used, whether it was pencils and clipboards or scanning guns. And at the largest companies, you might even have heard about the emerging use of predictive analytics to better align future supply with demand.
But in 2026, inventory management is not just about managing stock-keeping units (SKUs); it’s also about carefully tending the data that’s produced by every count. That’s because the data provides the raw material for artificial intelligence (AI) analytical engines, and companies are looking for returns on their big AI investments.
However, even though most operations share the goal of producing clean inventory data, the sheer speed and volume of inventory flowing through supply chains mean there are plenty of opportunities for error—as well as opportunities for the data to become outdated. While companies today use a variety of technologies to help track goods—including fixed barcode scanning stations, aerial drones, autonomous robots, mobile barcode or radio-frequency identification (RFID) scanners, and machine vision cameras—there still seem to be chances for errors to creep in, such as when a driver loses track of the barcodes they’re scanning or when a worker who’s doing a count on a mobile computer types in the wrong numbers.
“Control and aligning of that data is a primary topic. And the ability to orchestrate that data well is important, especially as the number of inputs grows,” says Wes Coleman, industry principal – warehouse at Zebra Technologies Corp.
Adding to the challenge, counts can occur at different points in the flow of goods, perhaps beginning with shipment by the vendor, then again at the receiving dock and in the warehouse racks, and finally each time an item is picked, packed, or shipped. “At each point [a discrepancy can emerge]—a worker may say the inventory wasn’t there, or they needed six and there were only four, or they picked the wrong item, which actually changes the inventory [count] for two SKUs,” Coleman says.
TALLY UP THE NUMBERS
Combining all those inputs to arrive at a single, accurate total can be devilishly tricky. As Zebra sees it, an important tool in the process is cloud computing, which offers a way to collect data at many different times and from many different sources and then align it with a company’s system of record, such as its warehouse management system (WMS), manufacturing execution system (MES), or enterprise resource planning (ERP) software. To help users accomplish that, the company offers its Workcloud Integration & Orchestration (Workcloud IO) product, saying it serves as the central nervous system for data flow, providing a standardized integration layer that connects front-line workers to core business systems.
Other logistics technology developers are addressing the problem in different ways—such as by adding inventory counting capabilities to their warehouse robots. For example, autonomous mobile robot (AMR) manufacturer Locus Robotics offers collaborative robots for picking and putaway tasks that are also able to update inventory counts every time they touch an item. There are other robotic counting technologies on the market as well; they include Gather AI’s flying drones and sensors for forklifts, Corvus Robotics’ flying indoor drones, and Dexory’s aisle-cruising AMRs.
Such tools are increasingly improving the inventory counting process by applying “physical AI,” which is an approach that combines the analytical ability of “white collar” AI with the real-world physical machine-led interactions that are part and parcel of warehouse cycle counts. And some of those tools are leveraging agentic AI as well.
For example, enterprise software vendor Oracle recently announced it had added agentic AI technology to its Oracle Fusion Cloud Supply Chain & Manufacturing (SCM) product to help users boost supply chain performance. The newly enhanced product includes four “Fusion Agentic Applications,” one of which is an “Inventory Planning Command Center” that Oracle says helps supply chain teams improve inventory availability, increase service levels, and resolve stockouts faster. “This shifts inventory management from manual tracking to an automated, business-driven workflow that helps teams reduce disruptions and improve inventory responsiveness,” the company said in a press release.
The number of tools on the market designed for counting and managing warehouse inventory has never been greater, industry experts say. But with that capability comes the added challenge of managing the data that’s collected.
“Organizations require data integrity. And that means they need governance and control of their data to get ahead of the AI challenge,” Zebra’s Coleman says. “Businesses have come to understand the need to move the ball forward. Data accuracy and its impact on AI has forced them to take action, because data is the fuel of AI.”
