What separates a stable manufacturing operation from one that constantly reacts to problems? It is not just the technology. It is the discipline behind how that technology is understood, maintained, and improved. The best teams I have worked with do more than respond quickly. They prepare deeply. They document clearly. They build systems that prevent problems from returning.

In this issue of FRAME, we examine what that discipline looks like in real-world manufacturing.

We begin with a realistic look at artificial intelligence and where it is already delivering value on the plant floor. From vision systems to robotic cell design and data analytics, AI is moving from concept to tool, and the shift is worth understanding. We then shift to the fundamentals of troubleshooting. Not the heroic kind that happens after a breakdown, but the kind that starts before anything fails. Preparedness, access to tools and documentation, and building the right relationships all play a part in recovery speed. We continue with a clear and practical explanation of Six Sigma. Too often misunderstood, Six Sigma remains one of the most powerful frameworks for reducing variation and improving quality when applied with intent and clarity.

Finally, we explore a mental model that separates good engineers from great ones: the ability to see the system. The professionals who consistently solve the right problems are not just those with deep technical knowledge. They are the ones who step back, ask better questions, and connect local symptoms to systemic causes.

A Realistic Take on AI in Manufacturing

Artificial Intelligence is no longer theoretical in manufacturing. It has quietly moved from abstract promise to practical deployment. While the early hype cycles produced more noise than results, we are now seeing credible tools emerge that solve real problems.

I have been skeptical of AI in the past. Early models struggled to generate coherent images or text, and many of the so-called solutions lacked any grounding in operational reality. But over time, the progress has become difficult to ignore. Last week at Automate, one of the largest industrial automation trade shows, it became clear that AI is beginning to play a legitimate role in our ecosystem. Nearly every major OEM, vendor, and integrator had something to show involving machine learning, computer vision, or generative tools.

Here is where the progress is most evident:

Machine Vision is More Capable and Easier to Deploy

Vision systems used to require hours of fine-tuning. I remember working through Cognex spreadsheets and struggling with EasyBuilder to get a reliable result. Today, AI-enhanced vision systems are far more adaptive. The underlying models can learn and generalize across a wider set of use cases, significantly reducing configuration time. This unlocks broader adoption in quality control, defect detection, and sorting.

AI is Accelerating Robotic Workcell Design

Vendors are releasing design tools that use AI to suggest robot cell layouts, select grippers, and optimize motion paths. These capabilities reduce engineering time and improve first-pass success rates. For high-mix or short-run production environments, this can provide a meaningful edge in deployment speed and flexibility.

Copilots are Emerging for Industrial Use Cases

While AI copilots are widely used by software developers, we are starting to see early-stage tools designed for manufacturing workflows. I spoke with several companies that are testing copilots for troubleshooting, documentation access, and technician guidance. These are not mature systems yet, but the intent and investment are clearly there.

Plant Data is Finally Being Interpreted, Not Just Collected

Many manufacturers have invested in collecting and contextualizing data. The next challenge is drawing conclusions from that data. A new wave of tools applies machine learning to structured plant information, identifying trends, detecting anomalies, and producing actionable insights. Several platforms I reviewed at Automate are doing this with surprising precision.

Like many of you, I am now taking a deeper look at how AI can be applied responsibly and effectively in industrial environments. I am not interested in vague promises or abstract visions. I want to understand the foundations so I can separate signal from noise.

If you are in manufacturing and still treating AI as a distant trend, I would encourage you to reconsider. The technology is not going to replace your operations, but it will reshape how certain decisions are made, how systems are built, and how work gets done.

Here are a few resources I am currently studying to deepen my understanding:

• AI Engineering: Building Applications with Foundation Models by Chip Huyen

• Build a Large Language Model (From Scratch) by Sebastian Raschka

• Ultimate AWS Certified AI Practitioner AIF-C01 by Stephane Maarek

As always, I am open to conversations with others who are exploring this space. We need more grounded voices in manufacturing to guide these tools toward real outcomes.

How to Actually Get Better at Troubleshooting

Troubleshooting is one of the most talked-about yet least taught skills on the plant floor. Everyone agrees it is important, but very few know how to develop it consistently across a team. You can only learn so much from theory. At some point, real learning requires standing in front of equipment, under pressure, and working through a problem. That said, there are steps that can significantly improve your ability to troubleshoot. These steps lead to faster response times, lower mean time to repair (MTTR), and better overall performance.

What Troubleshooting Really Depends On

Troubleshooting is not just a reaction to failure. It is the process of identifying the root cause, but what often goes unnoticed is how much of that process happens before anything breaks. Most teams lose time not because the failure is complex, but because they are unprepared when it happens.

Consider these questions:

• Do you have complete and current documentation for the systems you support?

• Do you know where the latest versions of the software and backups are stored?

• Do you have access to the right hardware tools and software licenses?

• Do you understand how the equipment is laid out and how the system is designed?

• Do you have a clear idea of who to contact when things go wrong?

Each of these questions directly affects how quickly and effectively you can respond to a problem.

Case Example: A New Case Packer on the Floor

Imagine your facility just installed a new case packing machine. The commissioning is done, and the asset is now in the hands of operations. What should you do now to prepare for the first time it fails?

Documentation is Non-Negotiable

The first thing I ask for during any troubleshooting engagement is documentation. This includes electrical drawings, panel layouts, mechanical schematics, and network diagrams. Without this information, even basic diagnostics take longer than they should. If documentation is not available, someone needs to take responsibility for building it. No one should be supporting equipment blind.

Hardware, Software, and License Planning

Downtime is often extended because someone is missing a cable, the right software version, or a working license. Each critical system in your facility should have a clearly documented list of what is needed to go online, including the type of controller, the software package, the version, and who owns the license.

I have seen factories lose multiple shifts of production because they discovered mid-failure that the only licensed software was installed on a laptop that was no longer in the building. This kind of delay is avoidable.

Create a shared resource that lists the software and tools needed for each asset in the facility. Make it easy to update and easy to access.

Building Relationships That Matter

As a young engineer, I believed I needed to know everything myself. Over time, I learned that being resourceful is more important than being all-knowing.

When equipment fails, the people who built it usually know more than those operating it. They have seen rare faults, edge cases, and integration issues that never made it into the manuals. In many cases, solving the issue means picking up the phone and calling someone who has seen it before.

Building these relationships before you need them makes all the difference. That might include staying in contact with the OEM, attending trade shows to meet product experts, or having a trusted systems integrator on call. In every facility I have worked with, the fastest troubleshooters are the ones who know who to call.

Pre-Troubleshooting is Real Work

Troubleshooting begins long before the problem. The most effective teams prepare in advance. They know their systems, keep documentation organized, and have a support network they can rely on.

If you want to improve the way your team handles downtime, start with the basics:

• Audit your documentation

• Review your software and licensing situation

• Identify the key people who can help you when a failure occurs

• Practice walking through systems before they go down

The real work happens before the failure. If you ignore these steps, troubleshooting becomes guesswork. If you take them seriously, your facility becomes faster, more capable, and less vulnerable.

What Six Sigma Really Means in Manufacturing

Six Sigma is a disciplined, data-driven methodology for reducing variation, eliminating defects, and improving the repeatability of complex processes. It was originally developed at Motorola in the 1980s and later adopted at scale by companies like General Electric and Honeywell. Today, its tools and philosophy are embedded in thousands of manufacturing organizations around the world.

Yet despite its widespread adoption, Six Sigma is often misunderstood. In my experience, professionals who have not studied statistics rarely grasp what Six Sigma actually means or how to apply it in the context of production lines, quality checks, or troubleshooting efforts.

This section is not a certification course. It is a practical explanation designed to help you speak confidently about Six Sigma with your colleagues, collaborate more effectively with engineering teams, and think more critically when making data-driven decisions.

Foundations of Six Sigma

Imagine standing on a production floor where your team is producing 1,000 popsicles every minute. Your job is to make sure that every one of those popsicles meets a strict quality standard based on sugar content. But there is a challenge: the test for sugar content is destructive. Once you sample a popsicle for testing, you cannot put it back into inventory.

This leaves you with two competing constraints. First, you want to minimize the number of units you discard as a result of testing. Second, you want to minimize the number of defective products that reach your customers. Your job is to balance these risks intelligently using a structured approach to sampling and decision-making.

At this point, the most important question to ask your leadership is: how many defects are we willing to accept? If the answer is "as few as possible," then the standard you are aligning with is likely Six Sigma.

What Six Sigma Means Statistically

The term Six Sigma comes from statistics and is based on the behavior of normal distributions, also known as bell curves. In any stable process, the majority of output will cluster around a central average, or mean. Sigma refers to the standard deviation, which is a measure of how much variation exists in that process.

Here is how coverage works in a typical normal distribution:

• One Sigma from the mean captures approximately 68.2 percent of all output

• Two Sigma captures about 95.4 percent

• Three Sigma includes roughly 99.73 percent

• Six Sigma captures 99.99966 percent

A Six Sigma process is so stable and consistent that the probability of a defect occurring is only 3.4 per one million opportunities. This assumes a small, long-term drift in the process mean, which is built into the Six Sigma model.

In practical terms, Six Sigma is not a guarantee of perfection. Instead, it is a quality benchmark that says: we have reduced process variability to the point where defects are extremely rare and statistically predictable.

Why This Matters on the Plant Floor

Six Sigma thinking is especially useful in high-volume production environments where:

• Defects are expensive or dangerous

• Testing is destructive, slow, or difficult

• Minor deviations can cascade into large quality issues

• Continuous improvement requires real structure, not just experience or guesswork

Using Six Sigma tools and principles allows your team to make smarter decisions about inspection frequencies, control limits, sampling strategies, and corrective actions. It supports better conversations between operators, quality teams, engineers, and managers.

It also encourages a shift away from reactive thinking toward preventive control. Instead of asking "what broke," Six Sigma pushes you to ask "why did the variation occur, and how do we prevent it in the future?"

Good Engineers Fix Problems. Great Engineers See the System.

One of the most powerful shifts I have observed in manufacturing professionals as they grow is this:

• They stop focusing only on isolated tasks, machines, or issues.

• They start seeing how the entire system operates.

This shift is not about experience alone. It is about learning to think across time, across departments, and across causes. It is the difference between someone who reacts well and someone who builds processes that eliminate the need for reaction in the first place. This is what separates good from great. And in many organizations, it is what separates those who remain stuck in daily firefighting from those who get trusted with transformation initiatives.

The Limits of Local Thinking

In most manufacturing environments, the pressure to solve problems quickly is high. When a machine goes down or a KPI slips, teams are expected to respond fast. That usually means solving the most visible issue.

• A sensor is not reading correctly, so we replace it.

• A label is misaligned, so we adjust the printer.

• A pallet is jammed, so we re-teach the robot position.

None of these are wrong. But all of them may be incomplete. Local thinking optimizes the symptom. Systems thinking identifies and redesigns the pattern. Let me give you a few real examples from the field.

Example 1: The Filler That Was Not the Problem

At a beverage facility, the team kept missing their production targets. The assumption was that the filler was the bottleneck. Technicians requested better nozzles, faster controls, and changes to the cleaning cycle.

But when the process was mapped, it turned out the problem was upstream. Ingredient availability was inconsistent. The batching system was releasing the wrong flavors out of sequence, which delayed changeovers and starved the filler for extended periods.

No amount of local optimization at the filler could overcome the variability in the supply schedule. The root of the problem was in planning and materials coordination.

The team that saw this not only fixed the issue, but also redesigned how batch instructions were released from ERP to the floor.

Example 2: The Alarm Everyone Ignored

At another plant, operators regularly ignored an HMI alarm that warned of high temperature in a hydraulic system. When asked why, the response was simple: “It always goes off, and nothing happens.”

The alarm had been hardcoded years ago, and no one had revisited the logic. Over time, the system drifted. A sensor degraded, the control loop lagged, and the temperature occasionally did creep too high, leading to a major pump failure.

But because the alarm became noise, no one connected the dots.

A true systems thinker would have asked:
Why does this alarm keep triggering?
What was the original design intent?
Has the process or control system changed?

They would not just fix the hardware. They would revalidate the purpose of the system, check the alarm thresholds, and update the logic to reflect the actual risk.

What Systems Thinkers Do Differently

Systems thinking is not just about being clever. It is about developing structured awareness. It is a deliberate effort to step back and ask bigger questions:

• Where is the true constraint in this process?

• How do inputs, workflows, and decisions interact over time?

• Which performance metrics are competing with each other?

• What incentives or assumptions are embedded in this setup?

Great engineers, operators, and managers train themselves to notice misalignments between departments. They see how quality metrics might compete with throughput. They understand how procurement choices affect downtime. They recognize that a new automation project might fail not because of the code, but because no one aligned on changeover protocols.

This way of thinking is not abstract. It is tactical.

How to Build This Skill

You do not need a systems engineering degree to think this way. Here are a few habits that develop systems awareness:

1. Start asking "Why is this happening here and now?"
   Not just "what failed," but "why now, why this asset, why this pattern?"

2. Use process mapping regularly
   Even simple diagrams of material flow, information handoffs, or decision points can reveal inefficiencies no one is talking about.

3. Involve people across silos
   A maintenance issue might be rooted in procurement delays. A production miss might be due to a change in forecast logic. Invite other functions to the discussion.

4. Treat KPIs as interconnected, not isolated
   If OEE goes up but quality complaints spike, something else is broken. Good metrics balance trade-offs.

5. Look for recurring failure themes
   Not just repeated downtime on one machine, but patterns that show up in different forms across the operation.

Why This Matters for Your Career

The professionals who rise in manufacturing are not just technical experts. They are the ones who can think across departments, across time horizons, and across goals.

• They connect frontline issues to executive strategy.

• They see how capital investments ripple into training, maintenance, and change management.

• They can translate from machine behavior to system behavior.

• And most importantly, they are trusted to solve problems that matter.

If you want to accelerate your growth in manufacturing, start by asking better questions. Step back. Zoom out. And train yourself to see the system, not just the symptom.

This is not just how you solve problems. This is how you become someone others trust to solve the right ones.

Conclusion

Excellence in manufacturing does not come from a single innovation. It is built through consistent habits, clear thinking, and disciplined execution. Whether you are adopting AI, improving troubleshooting processes, applying Six Sigma, or learning to see the full system, the most impactful changes are often the result of small, intentional steps taken over time.

This issue is a reminder that modern tools are only as effective as the structure behind them. Documentation, preparation, collaboration, and the ability to ask better questions are what drive meaningful and lasting improvements.

As always, I encourage you to reflect on how these ideas apply to your work. Share them with your team. Start a conversation. The real value of this newsletter comes from how these insights are used, challenged, and turned into action on the floor.

If you have thoughts, questions, or examples of your own, I would love to hear them. Let’s keep learning and building together.

Get in Touch with Us

• Vladimir Romanov

• Joltek