No eight words strike more fear in a manufacturing leader’s heart than, “The board wants to see our AI strategy.”
As part of my role as co-founder of an industrial maintenance startup, I’ve spent hundreds of hours over the past 12 months speaking with frontline manufacturing teams across the country. This demand for an AI strategy is something I hear everywhere I go.
Manufacturing leaders are feeling the pressure; in fact, a recent study of 1,200 IT decision-makers found that two-thirds identified FOMO, or “fear of missing out,” as a significant factor in their company’s decision to adopt AI technologies.
Podcast: What awaits manufacturing in 2025
While I believe it’s true that every sector will be reshaped by AI within the next five years—and that there’s a significant first-mover advantage to be had—I also see a troubling pattern emerging. Most of the people tasked with bringing AI strategies forward within their organizations don’t really understand what AI is or how it works. This lack of understanding leads to what I call the “AI trap,” where teams rush to adopt AI without having the right foundation, leading to failed or ineffective implementations.
When we talk about AI in manufacturing, we’re often referring to machine learning—a subset of AI that uses algorithms to learn from data and make predictions or decisions. While true artificial intelligence that “thinks” like a human doesn’t exist yet, these machine learning systems are already transforming industries. This wouldn’t be an issue if so many companies weren’t scrambling to implement AI initiatives without the proper foundation in place.
When under pressure to demonstrate progress on the AI front, many leaders default to purchasing an off-the-shelf solution and plugging it into their existing systems. Then, they wait for the magic to happen—but often, the magic never materializes.Instead, one of three scenarios occur:
- The AI fails spectacularly, leading company leaders to dismiss its potential entirely.
- The AI generates seemingly plausible but inaccurate predictions, resulting in poor decisions and skepticism about AI’s viability.
- The AI produces results that seem reasonable but lack actionable insights, leading to wasted resources and frustration. In all three cases, the outcome is the same: resistance to further AI initiatives is created. These companies become inoculated against the very technology that could make them more competitive.
To avoid these pitfalls, manufacturing leaders must first understand the two primary types of data at play inside their operations: machine-data and human-generated data.
See also: Six ways to incorporate AI into your manufacturing operations
Machine data, often referred to as operational technology (OT) data, comes from sensors and control systems embedded in manufacturing equipment. This data stream provides information about machine performance, output, and conditions. Many companies have already invested heavily into IoT sensors and SCADA systems to monitor equipment in real-time.
Machine data is only half the picture
Human-generated data, the operational data from frontline teams, is equally important but often overlooked. This includes maintenance records, quality control observations, shift handover notes, and the stream of daily decisions made by experienced operators and technicians. This type of data contains crucial insights and context that machines alone cannot capture.
For example, a sensor might suggest a machine is running hot, but an experienced operator knows this machine tends to run warmer on humid days, and it’s not an immediate cause for alarm.
Podcast: AI, automation and computer-aided manufacturing
In reality, truly effective AI systems in manufacturing need both machine data and human insights, working together. Machine data provides the “what” and “when,” while human-generated data provides the “why” and “how.”
When you combine machine and human data in the same system, you can then start to identify powerful connections between the two. You might discover, for example, that a certain operator adjustment consistently leads to higher output quality, or that machine failure coincides with a specific maintenance task being skipped.
These insights can only be drawn when your data platform can link the “what” and “when” machine data with “why” and “how” insights from your people. This contextual integration enables you to move from merely monitoring your operations to truly understanding and optimizing them.
Manufacturing leaders who recognize integration are already benefitting
They’re not waiting for the perfect AI tool to appear on the horizon.
Instead, they’re focusing on capturing clean, comprehensive data about their operations, and organizing it in a structured and consistent way. This strong data foundation is something that many manufacturers I speak with still don’t have. Instead, their operational data is incomplete and scattered across siloed systems, paper records, and inside the heads of their veteran employees. And this is the bottleneck for most teams, because you can’t just skip to the AI part. You need to build the data foundation first.
Episode 4 of (R)Evolutionizing Manufacturing: The Art of the Possible and The Art of the Practical
Effective data capture requires team effort. The workflows for collecting data need to be realistic and grounded in the everyday realities of frontline workers. Software implemented to enable this data capture must make employees’ lives easier, not harder. Far too often, solutions are selected by IT or executive buyers, and not enough consideration is given to the end-user experience.
If software is painful to use, or not designed to be used where the work actually gets done (away from a desk), then it won’t get used. Humans have a knack for finding workarounds, whether it’s entering the bare minimum of information, using the software inconsistently, or inventing a parallel system of Post-It notes and spreadsheets. The result? Frustrated employees and bad data.
To truly benefit from AI, particularly the breakthroughs right around the corner, manufacturers need to build a solid foundation of data first. Instead of rushing into AI implementation, the focus should be on capturing and organizing both machine and human-generated data in a sustainable way.
Only then can AI deliver the transformation that manufacturers seek.
