Case Study: How One Mid-Market Manufacturer Turned Analog Power Signals Into AI-Driven Operational Gains — Guidewheel — March 18, 2026
Vidcast Interview w/ Lauren Dunford, Heath Evans, and Steve Billock
Artificial intelligence has rapidly moved from experiment to an increasingly mainstream expectation in manufacturing. Industry surveys show rising adoption across operations, maintenance and planning functions. Yet operational returns remain uneven. Many initiatives stall after proof-of-concept, while others struggle to integrate into daily plant workflows.
For mid-market manufacturers, the tension is particularly acute. This segment operates under capital constraints, workforce pressures and persistent margin compression. Raw material volatility, energy costs and global competition leave little room for miscalculated technology investments. AI cannot function as a parallel innovation program. It must reinforce uptime, improve overall equipment effectiveness and reduce operational risk without destabilizing production.
In a recent BizTechReports vidcast interview, Lauren Dunford, CEO of Guidewheel, joined Heath Evans, plant manager at Alleguard’s Greenville, Michigan facility, and maintenance manager Steve Billock to examine how one manufacturer navigated that tension. Rather than beginning with robotics or automation overlays, the Greenville team started with a more foundational question: how to translate analog machine behavior into structured digital signals that both humans and AI systems could rely on.
Here is what they had to say:
BTR: To ground the conversation, Heath, what does Alleguard do, and where do you fit operationally?
Heath Evans, Alleguard: I’m the plant manager at Alleguard in Greenville, Michigan. We do foam packaging for a range of markets. You see it in appliance packaging, like for Whirlpool products. We do foam for office furniture, and about 30 percent of our work is automotive, including interior side impact components made out of expanded polypropylene. We also have a large cooler business. We do kitting for a cooler chain, and we do medical and pharmaceutical coolers. If you walk into a retailer like Dollar General and buy a cooler on the go, you might be holding something we made. You may not have heard of us, but you have probably used products that rely on our packaging.
BTR: Steve, what is your role, and why did this particular technology show up on your radar?
Steve Billock, Alleguard: I’m the maintenance manager. Machine upkeep and preventive maintenance are the name of the game. I’m a certified welder and I’ve done an electrical apprenticeship. When Guidewheel came in, the way they monitor equipment felt second nature to what maintenance teams do. We troubleshoot with electrical information all the time. So the idea that you could take something familiar and turn it into real-time data made sense, even if I was skeptical at first.
BTR: Lauren, describe Guidewheel in the simplest operational terms. What is the product actually doing in plants like this?
Lauren Dunford, Guidewheel: We built a scalable way to give manufacturers visibility into how equipment is running so teams like Heath’s and Steve’s can address production challenges with good data and practical tools. The most common starting point is what I call the electrical heartbeat of the equipment. We use noninvasive sensors that clip around electrical lines and measure power draw. Our algorithms translate fluctuations in that signal into machine states like running, idle, or down. From there, teams can build alerting, dashboarding, and a consistent operational picture that supports improvement work.
BTR: When most people think about AI, they imagine software systems and APIs. This is different. You are starting with an analog signal. Why does that matter?
Dunford: Because the analog signal is universal. Most manufacturing plants have mixed fleets. Different ages, different makes, different models. Even if machines have controls and data, it is often locked inside the equipment and hard to compare. Power draw gives you a common denominator. We instantly capture small fluctuations of that electrical heartbeat, and the system gets smarter at recognizing patterns over time, especially when it has context from the plant floor.
That context matters. Operating schedules matter. If a plant is not supposed to run on Sundays, the system should not treat that as downtime. The goal is to translate raw signal into something actionable, then keep improving interpretation as more equipment is connected and as people provide operating context.
BTR: In practical terms, what changes once that electrical signal is converted into actionable machine data?
Dunford: It changes what is measurable. When you can infer machine state consistently across equipment, you can stop relying on paper logs and end-of-shift reconstruction. You can move to clear visibility, and that changes how quickly teams respond, how they prioritize, and how reliably they can improve.
BTR: Steve, how would you describe the “electrical heartbeat” concept from a maintenance perspective?
Billock: Any time AC voltage is transmitted through a wire, it creates an electromagnetic field. Maintenance technicians know this. If you want to know how many amps a motor is pulling, you put an amp clamp on the wire and you read it. That is basic troubleshooting.
What Guidewheel did was take that signal we use every day and turn it into current data. Every time a press does anything, the motor draws current. The software can capture that and show what the machine is doing based on how that draw changes.
BTR: Are you placing these sensors everywhere, or only on selected equipment?
Billock: We started with a good majority of our production equipment and one air compressor. On each press we used multiple clamps, because interpretation matters. We monitor our hydraulic system, we monitor the overall incoming power to the press, and we monitor something that tells us when the machine is actually producing a part. In our world you have to preheat a press. If you only look at power, preheating can look like production. We wanted a way to distinguish “energized and warming up” from “actually cycling and producing.” We put a clamp on the hoppers so we could see when material is being pushed into the mold, which helps us interpret what the press is doing.
Over time, that became a skill. We did not just install clamps. We developed a working understanding of the signals under different conditions so the data matched reality.
BTR: Heath, let’s step back and frame the problem you were trying to solve. What was the struggle before the technology showed up?
Evans: In order to manage production, you need hour-by-hour, visibility. We didn’t have that when I got here. We were relying on lagging indicators. You didn’t know if you were winning or losing until the end of the shift. The goal was to move away from paper and toward technology that could give us real-time insight without overwhelming the team. We wanted better data going in so we could actually solve problems instead of chasing our tail.
Before this, we were always looking back at what we did. You didn’t know if you were winning or losing until the end of the shift. That makes it hard to manage hour by hour. We wanted live production monitoring and we did not want to be stuck in paper-based processes.
BTR: How did the decision happen? Was it top-down, bottom-up, or both?
Evans: Top management was already working on solutions to getting real time data and they wanted technology instead of paper. One of our owners started working with Guidewheel. We were the beta plant. We trialed it in October for a week, then extended it on one press. We launched broader deployment the week before Thanksgiving. Guidewheel had people onsite to help, and then Steve and his team took over and ran with it.
So it was supported from the top, but it only works if the plant team owns it.
BTR: What did “owning it” look like in practice?
Evans: We approached it as crawl, walk, run. In the crawl-to-walk phase we focused on uptime and downtime occurrences, and we focused heavily on data integrity. If downtime is not tagged properly and coded properly, then you end up operating inefficiently. You can’t solve problems if the data is inconsistent. We wanted good inputs before we tried to do deeper improvement work.
BTR: Can you tell us what changed?
Evans: Our uptime went from low 70s to about 86 percent in February, and that was our third full month with Guidewheel. That is a significant change. On OEE, we were in the mid-70s in the last quarter of 2025. In February we were at 88.9 month to date.
The numbers are meaningful, but what was really important was engagement. We had to get people involved and show them what is in it for them.
BTR: Why is it important for manufacturers to be deliberate about the use cases they choose to launch their AI initiatives?
Dunford: Many companies jump to the most advanced or flashy use cases, like robotics or predictive maintenance, and those can absolutely have value. But there is also a category I would call fixable fundamentals. That is where a lot of the money is made or lost. If you cannot see runtime, downtime, and loss patterns consistently, you cannot manage them. Heath’s improvement from the 70s to the mid-80s is not primarily a predictive maintenance story. It is a visibility, alignment, and execution story, enabled by tools.
Another point is that culture change supported by technology is powerful. Culture change alone can work, but technology can make it easier to sustain because it makes the truth visible to everyone at the same time.
BTR: Heath, did it create a cultural impact?
Evans: Yes. We now know hour by hour, minute by minute. We have screens up for equipment, showing whether it is running or down. We have tablets at equipment for operators to interface and tag downtime. When people can see the target and see performance against the target, most of them will go after it. People come to work wanting to do a good job.
It also created friendly competition. Between operators, between shifts, between supervisors. Not in a negative way. It is about shared visibility and shared goals.
BTR: Steve, how did your team initially react to bringing this technology into the plant?
Billock: Maintenance people like tools that live in their toolbox. We like wrenches and meters and things we understand. When I first heard this was coming, I was hesitant. But once it was in my hands, I got excited because it was familiar. It was basically saying, we are going to put an amp clamp on equipment and make it visible all the time.
The other factor is that it gave my team a reason to look at information in a new way. To get maintenance technicians excited about looking at a screen is rare. But if the screen tells you something meaningful before it becomes a bigger problem, you pay attention.
BTR: Can you give me a concrete example of how the plant is using it day to day?
Billock: Every morning, we review what happened the last two shifts. We make sure downtime is classified consistently. If everyone calls the same problem by different names, you cannot troubleshoot. So we standardize language.
Then weekly, we do a review. We identify the biggest downtime driver and focus on that. I used a metaphor in the conversation, and yes, the point is to tackle the biggest issue in a disciplined way rather than chase everything. We find what hurt us most, we go instance by instance using the tags, and we decide what action to take.
We also use it to recognize what went well. It is not just a negative tool. If we had a strong day, we call that out.
BTR: Has it changed the way you view your role, or how you do your job?
Billock: It really has. You know, I’ve got a three-month-old baby. So every morning we get up, I get my coffee, he gets his bottle, and I pull up Guidewheel on my phone. I can see what happened on the last two shifts. I can compare it to what the supervisors said happened. It is surprisingly accurate.
For me, the point is not that I like checking my phone. The point is that information that used to be locked inside machines and paper logs is now available securely, in a format I can use, to make decisions faster. It means I walk into the plant already knowing where the pain points are.
BTR: Heath, what does that imply about how work changes when information about machine behavior becomes mobile and visible?
Evans: It shortens response times and it changes accountability. Supervisors and team leads can prioritize where to go. If something has been down over a threshold, we get texts or emails. Scoreboards flash. It becomes clear where help is needed.
It also changes the burden of paperwork. Our goal in the next six months is to go paperless with production. We want to connect Guidewheel to our ERP system. Supervisors should not spend 45 to 60 minutes doing paperwork at the end of a shift. They should validate, pass it through, and focus on running the floor.
BTR: Lauren, where does risk management fit, especially cybersecurity? Many manufacturers are wary of connecting operational technology to networks.
Dunford: That concern is real. Manufacturers are targeted, and many mid-market operators are cautious about connecting equipment controls to corporate networks. Our system can be deployed in ways that avoid touching existing networks or equipment controls. It can be effectively separated from the plant network while still sending data to the cloud.
The information is air gapped. The practical point is to reduce perceived exposure and make adoption easier in risk-sensitive environments.
BTR: How does the system learn from individual machines while also benchmarking across similar equipment?
Dunford: Each machine has a unique history, usage pattern, and maintenance profile. Models need to account for that uniqueness over time. At the same time, as more machines across more plants are monitored, benchmarking becomes possible. You can compare similar equipment across sites and learn patterns that improve recognition and accuracy.
Over time, the system can help identify anomalies that correlate with risk, like overheating that could lead to a larger incident. The value is not only prediction. It is earlier detection and better prioritization.
BTR: Heath, does this improve your ability to manage a network of facilities, not just a single plant?
Evans: Yes. At our site we have 14 production machines being monitored. We also have a corporate-level scoreboard where VPs can see performance across plants. The rollout is still in progress across 18 plants, but it is well underway.
That visibility can create constructive oversight. It is not about micromanagement. It is about asking questions, seeing trends, and pushing improvement. It also helps when making the case for capital investment.
BTR: Can you explain that? How does better data affect capital decisions?
Evans: The data gives you authority. We can take screenshots of a press going down at a specific time for a specific reason, with operators tagging it in real time. That can go to my boss, then to his boss. It makes the case that a problem is recurring and costly.
Billock: Before, we had books of paper. If I handed that to Heath, someone could say, how do I know this is not just written because you want a new machine? Now the data is tied to events as they happen. It is harder to dismiss.
BTR: As you look ahead, what is the next phase?
Evans: We want to build on what we have. We want to start entering scrap, adding part numbers and cycle times, and capturing more complete OEE without extracting paper information. The crawl-to-walk phase was about data integrity. The next phase is using that integrity to drive deeper improvement.
Billock: From maintenance, I’m interested in baselining and moving toward earlier indicators. We already saw a hint of that when a pump failure changed power consumption patterns, and after replacement the signal smoothed. If we can establish baselines, we can potentially see problems earlier.
Dunford: And we want to keep the focus on making the data fit into existing workflows. Manufacturing already has strong continuous improvement practices. The technology should support that discipline, not compete with it.
BizTechReports Conclusion
The Alleguard experience underscores a broader reality in mid-market manufacturing. AI becomes operational not through ambition alone, but through disciplined sequencing. By prioritizing universal signal capture, standardized classification and workforce alignment, the Greenville plant strengthened both human judgment and AI enhanced analysis.
In a macroeconomic environment marked by geopolitical uncertainty, margin pressure, capital constraint and rising AI expectations, that phased approach may offer a sustainable path to effective modernization, if not transformation . Rather than attempting full automation prematurely, disciplined observability can generate measurable gains and build credibility.
###