GE Appliances turns to AI and robotics to improve warehouse efficiency - here's how it achieved 99.9% inventory accuracy

When you're moving millions of boxes through warehouses every year, losing track of even a single part can trigger expensive delays. GE Appliances—the inventor of the self-cleaning oven and whose products are found in one in four American households—has found a solution that's delivering huge efficiencies through data analytics and warehouse-based robots.

The adoption of these technologies is helping GE Appliances avoid costly inventory issues that can ripple through its just-in-time manufacturing operations, explains the company's Senior Director of Central Materials, Harry Chase:

It's critical that we keep inventory accuracy as high as we can possibly keep it; if we go to a location and a part is not there, we can't waste time on somebody driving around looking for it.

At the level we operate at, because we're moving millions of boxes through warehouses every year, we don't even want a 0.1% discrepancy if we can avoid it.

I don't know if we'll ever get to 100% inventory accuracy, but we wanted to get up to 99.9%. And for the 15 months we've had this system, we are consistently hitting that number.

A massive logistics operation

This isn't a small-scale operation. Founded in 1907 and headquartered in Louisville, Kentucky, GE Appliances operates as a leading US manufacturer of home, kitchen, and laundry appliances under brands including GE, Monogram, Café, GE Profile, and Hotpoint. Since 2016, the company has been part of Chinese multi-national Haier.

Chase leads an internal team dedicated to optimizing and automating GE Appliances' global coordination across a huge network: nine US manufacturing sites (though he jokes that "some people say eight and a half," as two operate in different buildings at one facility), four inbound warehouses, and five consolidation centers spanning Asia, China, Thailand, and Vietnam.

The scale of investment matches the ambition. Over the last eight years, GE Appliances has poured over $2 billion into logistics optimization—and this isn't about cutting jobs. The company has actually added over 2,000 employees to its operations as it pushes to transform this side of the business.

The manual counting nightmare

The problem Chase faced was both universal and uniquely challenging for a manufacturer of GE Appliances' scale. Every warehouse needs accurate stock counts, but manual stock-taking is notoriously time-consuming and error-prone - which is why many operations simply avoid it.

For GE Appliances, that avoidance isn't an option. The company operates on just-in-time manufacturing, where having the wrong parts available can shut down production lines. When your warehouses are managed by third-party logistics partners with high staff turnover, the risk of scanning errors multiplies.

The solution lies in regular 'cycle counts' - auditing specific portions of inventory to verify records match reality. But here's where the scale becomes a problem.

Chase explains:

We have five boxes of Part A and they're supposed to be in location one, two, three, four, five, say. But when you go in and count, somebody could have pulled inventory out or lost inventory or broke it and not recorded it, or there might only be four boxes not five, and so on.

So you want to go and count that inventory again and do it on a certain cadence—maybe every month you have to go through and confirm that what you put in the location is what's at your location is correct.

And as most of these parts in the warehouses are international, if something goes wrong it's very costly for us to get back to the States, it can take us literally six weeks to ship it back over from Asia.

With warehouses spanning 240,000 to 500,000 square feet, manual cycle counts using "guys or gals in overalls with clipboards" were consuming 30 days and requiring three full-time staff to complete - stretching to 45 days for the largest facility. The stakes couldn't be higher, especially given the international nature of the parts supply chain.

Enter the robots

Chase had explored various solutions, including drones, when serendipity struck. Reading "the only newspaper I read," The Times of London, he discovered an article about UK-based Dexory - a specialist in AI and robotics for warehousing and logistics.

Dexory's approach combines software and hardware, starting with a data visualization platform designed to automate inventory tracking and provide real-time supply chain insights. But the real innovation happens on the warehouse floor with new robotic colleagues that GE Appliances staff have affectionately nicknamed 'Dex,' 'Barney,' and 'The Count' (from Sesame Street).

These Autonomous Mobile Robots (AMRs) each feature an extendable 12-meter tower equipped with sensors, cameras, and laser-based distance measurement hardware. The result? Robots that can systematically trundle through warehouse aisles, scanning entire rack locations simultaneously within seconds.

The data capture is huge - up to 14 meters of warehouse racking scanned per pass, generating four gigabytes of data per meter traversed. All of this feeds into the broader stock visualization system that Chase and his team rely on for decision-making.

Beyond time savings

The robots deliver more than just speed improvements, Chase notes. The comprehensive data capture has enabled GE Appliances to build precise digital twins of its entire warehouse operation - creating virtual replicas that mirror the physical facilities in real-time.

This digital transformation eliminates dangerous practices like sending workers up forklifts to manually count inventory stacked as high as 32 feet. The system also identifies empty spaces automatically, enabling more efficient space utilization and "re-conformations."

These digital twins now serve as the foundation for GE Appliances' next phase of warehouse evolution. Chase envisions integration with existing warehouse management systems to unlock even greater inventory and material control. The goal? AI-driven insights that could determine optimal ordering patterns - perhaps suggesting that certain parts only need weekly rather than daily replenishment, or recommending Tuesday deliveries over Monday arrivals to optimize space utilization.

The transformation represents more than operational efficiency - it's about creating a data-rich environment that allows for predictive rather than reactive decision-making.

Chase concludes:

We have the data which we didn't have before, so we have the real opportunity to go out there and do a lot of enhancements.