Panasonic Connect: AI's real value in factories is supporting workers, not replacing them

Scott Zerkle, associate director of technical operations at Panasonic Connect North America, identifies a fundamental challenge reshaping electronics manufacturing: the explosion of complexity in modern products. Modern vehicles alone now contain 60 to 100 or more sensors, with some models exceeding 200. This proliferation of electronic components translates directly to tighter pitch spacing, smaller parts, and placement tolerances far more demanding than production lines were originally designed to handle. Even a deviation that was once acceptable can now result in a defect; an incorrect feeder load or unverified changeover can cause real disruption across multiple component variants.

Panasonic Connect, a business-to-business technology company within the Panasonic Group, addresses these challenges through its "Gemba Process Innovation" strategy, which aims to connect people, machines, and production data to improve productivity, quality, and operational resilience on the factory floor. Gemba itself means the actual place where work happens—the factory floor—and the goal is to capture what happens there in real time through a digital system rather than relying on paper records or operator memory. Zerkle explains that component miniaturization is outpacing what older lines were built for, with parts now placed at spacing measured in tens of microns. This demands real-time correction, where placement and paste volume adjust as conditions shift mid-run, and printing, placement, and inspection must work together so defects are caught at the step where they occur, not afterward.

When asked where AI delivers the greatest practical value today, Zerkle is candid: predictive maintenance and defect detection using machine data to catch issues before they cause downtime or scrap. But he is equally direct about where expectations run ahead of reality—the idea that a factory can be run by AI alone is "neither realistic nor the right objective." The core constraint, he argues, is that "AI is only as good as the data it receives," and most plants still run legacy systems that don't talk to each other, creating fragmented data. The real opportunity is to use AI to support people on the floor, taking on repetitive analysis so operators can focus on complex decisions that require human expertise. Manufacturers face competing pressures to improve quality, increase throughput, and remain resilient in the face of supply-chain disruptions, and Zerkle frames automation's real contribution as consistency across what he calls the "5Ms": huMan, machine, material, method, and measurement. Automated systems that monitor these five together can catch deviations before they evolve into defects or missed deadlines and correct them in real time.

High-mix, low-volume production, increasingly common across many sectors, used to mean accepting slower changeovers in exchange for flexibility. Today that tradeoff is shrinking through greater automation and process intelligence. Manufacturers are reducing setup time by verifying materials and feeder positions before a run starts and allowing operators to switch between configurations simply by selecting a different program rather than through manual reconfiguration. Reliability is just as important as speed, especially given workforce shortages and the growing challenge of relying on experienced technicians who know the process by memory alone. Looking ahead over the next decade, Zerkle predicts the biggest shift will be convergence: AI, sensing, and automation feeding off the same data instead of running independently. Many plants today have these capabilities but tend to work in isolation, each system flagging its own deviations without the context needed to understand why they occurred. The factories that benefit most will be those that evolve and get smarter from their own production history rather than those that simply invest in the newest technology. The competitive advantage, Zerkle concludes, won't come from adopting more tools—it will come from connecting them into a smarter manufacturing ecosystem.

The article presents a vision of manufacturing transformation rooted not in replacing human workers but in augmenting their capabilities through better data and decision support. Zerkle's central claim—that AI is only as good as the data it receives—identifies a critical gap: most plants still run legacy systems that don't communicate with each other, creating fragmented data that limits AI's potential. This framing matters because it shifts the conversation from automation-as-job-killer to automation-as-enabler, addressing a real concern many manufacturers face when adopting new technologies.

The underlying pressure comes from product complexity. As vehicles, devices, and industrial equipment incorporate exponentially more sensors and processors, the manufacturing systems that build them face dual demands: components are physically smaller and more densely packed, requiring tighter placement tolerances measured in tens of microns, yet production must remain flexible enough to handle high-mix, low-volume runs. This is where Zerkle sees the next decade's competitive advantage: not in adopting individual new tools (AI, robotics, sensing), but in connecting them into a coherent system that learns from production history. Factories that continuously optimize based on their own data will outpace those that simply purchase the latest technology.

No comments yet. Be the first to share your thoughts!