As manufacturers rush to embrace AI and analytics, one truth remains: Smart factories depend on clean, connected and trusted data.
Modern manufacturing runs on data — but much of that data still lives in silos. Machines, sensors and control systems generate terabytes of information every day, yet most of it never reaches the people or systems that can use it. For companies chasing digital transformation, the problem isn’t collecting more data. It’s connecting what they already have.
“We continue to see plant floors that still have a mix of new and legacy systems,” said John Lee, senior manager of manufacturing intelligence at Matrix Technologies. “Even some of the greenfields don’t have the latest and greatest. What matters is creating an architecture that brings all of that information together.”
The patchwork problem
Every piece of industrial equipment seems to speak its own language. Older programmable logic controllers (PLCs) communicate through proprietary protocols; newer sensors send data via Ethernet or wireless gateways. The result is a patchwork of disconnected islands.
Lee’s team helps manufacturers standardize those connections. The goal isn’t just visibility — it’s data integrity, ensuring that information is accurate, contextual and secure as it moves through the organization.
“Some of the assets and devices aren’t even connected to Ethernet,” Lee said. “We help bring them up to speed, make recommendations and ensure the data is reliable.”
Protocols such as OPC UA and MQTT are proving crucial. MQTT, in particular, uses a broker model that reduces network load and simplifies communication across systems.
“It cleans everything up and doesn’t overload the network so much,” Lee said. “There are a lot of advancements helping us do that today.”
Cybersecurity in a connected world
The more equipment that comes online, the greater the risk. Many industrial systems were designed decades ago, long before cybersecurity became a plant-floor concern. That leaves glaring vulnerabilities when legacy machines are suddenly connected to enterprise networks or the cloud.
Lee emphasizes a layered approach. His teams segment operational technology (OT) networks from information technology (IT) systems, using managed switches, VLANs and edge devices that restrict data to outbound-only traffic. “We have to have the layers of security with firewalls in place and architect that in a certain way,” he explained.
This architecture keeps production systems isolated while still allowing data to flow securely to analytics platforms or remote dashboards.
From data collection to decision intelligence
Once the structure is in place, the payoff comes in faster, smarter decisions. Real-time visibility enables manufacturers to act before small issues snowball into costly downtime.
“When you’re looking at data even hours later or in weekly reports, there’s no time to take action,” Lee said. “Real-time data empowers operators to be involved in decision-making and really accelerates the value of change.”
He stressed that automation doesn’t eliminate the human element. Operators remain critical for context and judgment.
“They’re still some of the most valuable decision-makers,” he said. “These tools just enable and help.”
That blend of human expertise and digital insight is what’s pushing factories toward predictive operations, where data doesn’t just describe what’s happening, but anticipates what will happen next.
Predictive maintenance and the rise of AI
Artificial intelligence (AI) and machine learning (ML) are beginning to reshape maintenance and production strategies. By modeling equipment behavior and correlating variables such as vibration, temperature and pressure, AI can identify early warning signs of failure.
“You can monitor vibrations, temperatures and other variables to predict failures well in advance,” said Lee. “Some OEMs are even offering this as a service now.”
These tools not only reduce downtime but also extend asset life and improve planning. Yet the success of AI still hinges on clean, connected, contextual data — the very foundation manufacturers are building now.
Building for scalability, not technical debt
Many plants are still in the early stages of digital transformation, and Lee warned against rushing toward flashy analytics before fixing the fundamentals.
“You want to be open and agile and be able to move with change, so you’re not painting yourself into a corner,” he said.
That means investing in scalable infrastructure, standard protocols and a cybersecurity posture strong enough to support future growth.
For Lee, progress isn’t just about technology — it’s about people seeing new possibilities.
“When we bring a solution in and those light bulbs start clicking, people really start seeing the value,” he said.
The path to smart manufacturing may be paved with sensors and software, but its foundation is still human insight and the discipline to make data worth trusting.
Make sure to check out the first episode of the Ctrl+Alt+Mfg podcast, where hosts Gary Cohen and Stephanie Neil discuss insights from WTWH Media’s State of Industrial Automation Report – Spring 2025.
