0

Manufacturing leaders know that AI holds enormous promise for their operations. Boards are asking about it, investors are talking about it and as usual, operations teams are under pressure to deliver it.

The potential is real. So is the problem.

AI runs on data, and for most manufacturers today the data that matters most simply doesn’t exist in a usable form. It’s sitting on a clipboard, scrawled on a whiteboard or in a supervisor’s notebook, or locked in a paper form stacked on a workstation at the end of a shift. The machines are generating signals and data too, but they’re flowing into disconnected OT systems that the person standing right in front of the machine can’t see or access.

This is the fundamental gap in the digital factory conversation. We’ve spent decades automating machines and investing in ERP systems, MES platforms and analytics dashboards. But the human layer of the operation (the decisions made, the tasks completed, the problems observed by frontline workers every single shift) has largely been left behind and left on paper. The vast majority of data generated from critical frontline operations never makes it into any system. It gets cleaned off whiteboards and boxed up in archive rooms at the end of every shift, every day.

The Cost of the Paper Black Hole

When critical operational data lives on paper, your visibility into performance is permanently incomplete. You can see outputs and outcomes, but you can’t see whether the work that drives those outcomes was actually done. Whether checks were completed, standard work followed, compliance requirements met.

Most companies fill that gap with people. Analysts, quality managers and continuous improvement consultants manually collecting data from paper records, trying to reconstruct a picture of what happened and why. It’s retrospective, incomplete, and by the time it reaches someone who can act on it, the moment has already passed.

This isn’t a new problem. But the reason it persists in 2026 is worth being honest about.

We all know the manufacturing success stories, the continuous improvement hall of famers. But what we continually seem to ignore is that Toyota, Motorola and Procter & Gamble didn’t build world-class operations by treating frontline workers as a cost. They built them by treating frontline workers as the

most valuable source of operational knowledge in the business, and designing systems to capture and act on that knowledge every day.

Then the automation revolution arrived, and that philosophy quietly inverted. Workers became a variable cost to cut, not a resource to equip. The frontline became the place you automated away from, not the

place you invested in. Technology was deployed to replace people, not enable them.

The results of that shift are everywhere today: paper-based processes, disconnected workers, and a widening gap between what machines can see and what the humans running them actually know. You can’t build an innovative manufacturing culture on that. You certainly can’t build a credible AI strategy on it.

The philosophy has to change. That means more than buying new software. It means recognising that the people on your floor are not a legacy problem to be automated around. They’re the most underutilised data source in your operation. Give them the right tools and they become the foundation everything else is built on.

Connected Worker Platforms: The Foundation Layer

The good news is that the technology to fix this is already in everyone’s pocket. Modern connected worker platforms make it straightforward to digitize frontline work: checklists, quality inspections, maintenance requests, downtime records, all completed on a mobile device, captured in real time and structured from the moment it’s submitted. No clipboards. No paper stacks. No data that disappears at end of shift.

The immediate payoff is process discipline. Work done digitally is done consistently. Checks can’t be quietly skipped. Photos and video replace subjective verbal reports. Corrective actions are owned, tracked and closed. Standards update everywhere, instantly.

But the bigger value compounds over time. Every task completed, every issue raised, every frontline observation becomes a data point. Weeks and months of this builds an operational dataset most manufacturers have never had before.

Then you connect it to machine data, IoT feeds and environmental sensors, and the picture gets genuinely interesting. Bearing failures that correlate with humidity levels. A 30% spike in manual handling injuries when temperature drops below 8 degrees. Cost-saving initiatives that look good on paper until you see the downstream quality impact. Leading indicators, lagging outcomes and machine signals combined tell a story none of them could tell alone.

Putting the Data to Work

With this foundation in place, the AI conversation changes entirely. Teams are working with continuous, structured data from both the human and machine layers of the operation. The models have something meaningful to work with.

To start, AI surfaces patterns that no human analyst has the bandwidth to find: leading indicators of quality failure, correlations between frontline activities and OEE, the right information routed to the right

person before the problem escalates. Frontline workers get AI they can actually use: real-time answers to troubleshooting questions, SOPs and compliance requirements, without leaving the floor.

Further out, agentic AI starts closing loops automatically. A machine stop triggers a maintenance alert with the relevant repair SOP already attached. A performance variation adjusts the production schedule and notifies the planning team. The operation responds to what’s actually happening, not to last week’s data in this morning’s meeting. And critically, the frontline worker isn’t cut out of that loop. They’re the ones with the context to act on it, faster and with better information than ever before.

The manufacturers that benefit most from AI won’t be the ones with the most sophisticated models. They’ll be the ones that built the right foundation first: digitizing the frontline, connecting the human and machine layers, linking it all to the enterprise outcome data that’s already there. Everything else follows.

Getting Started

The path forward is a change in philosophy as much as a technology decision.

Every dollar going into AI models right now is only as valuable as the data underneath them. AI capabilities are no longer a differentiator on their own. They’re a commodity. Your competitors have access to the same models, the same platforms, the same tools. What they don’t have is your data. And right now, while most of the industry is still debating AI strategy, the window to build a data advantage is open. It won’t stay open forever. The manufacturers moving now are quietly accumulating months of structured operational data that late movers will never be able to recreate. That gap compounds every single day.

One thing is for sure, what was true back when Toyota changed the game is true today. Factories that win won’t be the ones that eventually bought the best technology or AI. They’ll be the ones that built the strongest foundation: the culture, the patterns, the institutional knowledge, the real-time picture of what’s actually happening on the floor, captured shift after shift, in structured form, at scale.

Your competitors can copy your technology stack. They can’t copy that.

In the AI race in manufacturing, that’ll be the difference maker.