Marc Tolleson | Shareholder, LBMC Audit · Charlie Apigian | Shareholder, LBMC Data & AI Leader
Key Takeaways:
- Start with your most expensive operational problem. Not with technology.
- Data quality is the real risk. Most manufacturers have more data than they trust. Fix that first.
- AI works best as a tool for your best people, not a replacement for anyone.
- Sequence matters: assess your data, build a real-time operational picture, then run one focused pilot before year-end.
- Manufacturers who have done this well are not more automated. They are less frantic.
I was sitting with a plant manager in West Knoxville last fall when he put it plainly: “I know I’m supposed to be doing something with AI. I just don’t know what.”
That’s the conversation I keep having with manufacturers across this region — job shops with 30 employees, multi-plant operations with hundreds. Everyone knows AI matters. Nobody’s sure where it fits.
I’m a manufacturing auditor. My job is to understand how operations work and tell people what I see clearly. I’m not a technology salesman, and this is not a technology pitch. What I can offer is what I’ve been hearing from manufacturers I work with every day, and what I’ve learned from a colleague who understands this space better than almost anyone I know.
Charlie Apigian leads our Data and AI practice at LBMC. When my clients started asking questions I couldn’t answer, Charlie was the first person I called. He’s been in rooms I haven’t, and he’s seen what works and what doesn’t across manufacturing operations all over the Southeast. What follows is our shared thinking on where AI is creating real value today and how to start moving without wasting a year figuring out where to begin.
What I Hear Every Time I Sit Down with a Client
The problems keeping manufacturers up at night are not new. They’re familiar ones that keep getting more expensive.
Unplanned downtime that would have cost a fraction if someone caught it earlier. A quality issue that passed two checkpoints before anyone flagged it. A Friday operations report that should take 10 minutes but eats half the day because the numbers are living in four different places. A maintenance technician with 30 years of institutional knowledge who just told you he’s retiring in the spring.
None of those are technology problems. They’re operational problems. But they are exactly the kinds of problems that a more data-driven approach can help address, if you can get there without spending a fortune on something that doesn’t fit your business.
East Tennessee manufacturers have built something real here. The relationships running through this business community — with suppliers, customers, and workforce — are genuine competitive advantages. When manufacturers worry that AI will erode that, I take it seriously. What I’ve seen, though, is something different. The right approach to AI does not replace the judgment your people have built. It gives them fewer things to fight through so they can spend more time on the work that requires them. A human-centered AI strategy starts with that same idea: technology should support people and performance rather than become the strategy itself.
Where AI Is Creating Real Value Right Now
Charlie Apigian:
When Marc started pulling me into these conversations, what struck me was how consistent the frustrations were. Nobody was asking about AI in the abstract. They were asking about specific pain they were already living with. That is exactly the right starting point.
Five Areas of Real Impact for Manufacturers
- One trusted operational picture: Connect your ERP, production floor, and reporting into a single live view. Business intelligence and analytics can bring disconnected data sources together through dashboards, KPI tracking and real-time reporting. When leadership works from the same numbers, decisions get faster and better. This is the foundation everything else is built on.
- Catching problems before they escalate: Real-time dashboards and anomaly alerts let the right people act before failures happen. One operation cut unplanned downtime significantly — without buying new equipment — just by surfacing signals earlier.
- Preserving experienced workers’ knowledge: When your best technician retires, what he knows shouldn’t leave with him. AI tools that capture decision logic and make it available to the next person are among the highest-value investments in manufacturing today.
- Predicting equipment failures: A planned maintenance window costs a fraction of an unplanned line stoppage. Predictive analytics can help manufacturers identify maintenance needs before they disrupt operations. Sensor data and maintenance history are often already in your systems. The question is whether you’re using them.
- Getting ahead of supply chain disruptions: The manufacturers who came through recent supply chain chaos best had the most visibility and the earliest signals. AI-assisted forecasting gives you more time to respond — and that lead time is often the difference between a manageable problem and a production crisis.
From Someone Who Has Done It
“We spent two years saying we needed to get smarter about our data. What we actually needed was someone to tell us we weren’t an AI problem. We were a data quality problem. Once we got honest about what our systems actually knew, the path got a lot clearer. We started with a dashboard. Then we ran one predictive maintenance pilot on the line that was killing us most. Six months later we had a real business case and the confidence to do more. Don’t skip the data cleanup. Everybody wants to skip it. Don’t.”
— Operations Director, mid-size contract manufacturer, Southeast
A Practical Path to a Use Case by Year End
The sequencing matters more than the speed.
Three Steps to a Working Pilot
- Assess your data — Get honest about what your systems actually know. Evaluate your ERP, production, maintenance, and quality data for gaps and inconsistencies. Most manufacturers find things here they did not expect.
- Build one real operational picture — Create a centralized, live view of production, downtime, quality, and inventory. Give your leadership team one trusted version of what’s happening. Better data-driven decision-making starts with giving leaders information they can actually trust and use. This step alone changes conversations.
- Run one targeted pilot — Pick the operational problem that costs you the most and run a focused pilot there — typically predictive maintenance or production forecasting. Prove value before scaling anything. This is achievable by next quarter if you start now.
The manufacturers who get this right are almost never the ones who spent the most. They are the ones who started with a specific problem, proved something worked, and built from there.
The Thing Nobody Wants to Talk About
Data quality determines whether an AI investment works or doesn’t. Technology is not the risk. The data is.
Inconsistent downtime codes. Inventory counts that stopped matching physical reality six months ago. Maintenance logs that look different depending on who filled them out. Quality records living in someone’s personal spreadsheet. These are not unusual situations. They are what we find in most manufacturing environments when we look honestly.
Before you invest in AI capabilities, invest in understanding what your data says right now. That work will pay dividends regardless of what technology choices you make later.
Five Questions Worth Sitting with Before You Start
Assess where you stand.
- Do your core systems connect to each other, or does someone reconcile them manually every week?
- When your leadership team gets a report, do they trust the numbers or verify them first?
- Where are your most expensive surprises coming from, and could earlier information have changed how you responded?
- What would you lose if your three most experienced people left in the next year, and is any of that written down?
- Have you asked your frontline where the daily friction is? They know.
Why This Matters Here
What makes Knoxville-area manufacturers competitive is not primarily their equipment or their software. It is the relationships they have built with the people who work for them, the suppliers they depend on, and the customers who trust them.
The manufacturers who have adopted these tools thoughtfully are not more transactional. They are less frantic. Their people have better information and fewer fires to fight.
If your procurement manager spends her Tuesdays manually chasing supplier data, she is not having strategic conversations with your most important vendors. If your plant manager is spending Friday mornings rebuilding reports from scratch, he is not on the floor with his team. The time AI gives back gets spent somewhere, and the manufacturers using it well are intentional about where.
Charlie and I are not here to tell you AI will transform your business overnight. What we will tell you is that manufacturers ignoring it entirely are going to find themselves at a real disadvantage, and the ones chasing it without a clear problem to solve are going to waste a lot of money.
The right path starts with an honest conversation about where your operation actually stands. LBMC’s AI and Data Transformation team works with middle-market organizations to assess AI readiness, strengthen data foundations and identify practical use cases that can produce measurable results. If you’re asking these questions, that conversation is worth having.
About the Authors
Marc Tolleson, Shareholder, LBMC Audit
Marc works alongside manufacturing clients across East Tennessee and the broader Southeast. He is known for asking the questions that get to the operational reality behind the numbers. Marc is based in Knoxville.
Charlie Apigian, Shareholder, LBMC Data & AI
Charlie leads LBMC’s Data and AI practice, working with middle-market manufacturers to build data-driven operations grounded in practical use cases. Charlie is based in Nashville.
