Episode 8: AI in Manufacturing: Where It Actually Works (and Where It Doesn’t)

In this episode of The Manufacturing Evolution, Ron Schlegel breaks down the practical side of AI in manufacturing—cutting through the hype to focus on where it actually delivers results on the shop floor. Drawing from his deep background in process improvement and quality systems, Ron explores how manufacturers can use AI and smart automation to strengthen consistency, reduce waste, and improve operational efficiency.

He walks through real-world applications, showing how these technologies can integrate with existing manufacturing processes rather than disrupt them. More importantly, Ron emphasizes the critical role of people and process—highlighting how the People–Process–Technology framework ensures that innovation drives measurable outcomes, not just complexity.

Whether you’re looking to improve quality, streamline workflows, or take the first step toward digital transformation, this episode delivers practical insights and grounded strategies to help manufacturers scale with confidence.

Welcome to The Manufacturing Evolution: AI, Ops & The Future of Work. I’m Ron Schlegel, and today we’ll unpack AI in manufacturing: where it truly works, where it doesn’t, and how to implement it without derailing your operations today.

Let’s cut through the hype. Define success using real, measurable results—like efficiency, quality, waste, lead time, and changeovers.
Then focus on what actually matters: fix your biggest bottlenecks first, look at your processes next, and only then bring in tools—don’t chase shiny solutions before solving the basics.

One: predictive maintenance for rotating equipment and critical utilities. Monitor vibration, temperature, amperage, and pressure at the edge to predict bearing failures, cut unplanned downtime, and right-size spares—without bloating maintenance calendars.

Two: computer vision for surface, assembly, and packaging inspection. With controlled lighting and fixturing, AI flags scratches, misalignments, seal issues, and label errors—boosting first-pass yield, cutting rework, and speeding feedback upstream.

Three: AI can help with scheduling and sequencing. It can learn patterns like changeovers, setup times, labor limits, and due dates to create better schedules.

This helps reduce delays, excess inventory, and last-minute rush orders.
But the key is—your team stays in control. AI suggests the plan, and your planners review and approve it before it goes live.

Four: Use AI to spot issues early. It can monitor multiple signals at once and detect changes like tool wear, process drift, or sensor problems.
Pair this with simple control charts so your team focuses on real issues, not false alarms.
The goal is to catch problems early and prevent defects before they move further down the line.

Fifth, use AI to optimize energy use. It can predict demand and adjust how equipment like compressors, ovens, and HVAC systems run.
This helps reduce energy peaks, lower costs, and improve efficiency per unit.
At the same time, it takes pressure off older systems and keeps everything running more smoothly.

Six: demand forecasting and inventory optimization. Blend history, seasonality, promotions, and forecasts to stabilize flow. Right-size safety stock, cut stockouts, and give suppliers clearer signals—reducing expediting and fire drills across purchasing and production.

Seven: Use automation and AI assistants to handle routine tasks. They can create work instructions, quality documents, and reports, and help with things like reconciling data and paperwork. This reduces manual, repetitive work so your team can focus on more important decisions and problem-solving.

Where doesn’t AI work well? When data are noisy, processes lack standard work, samples are tiny, or the problem is unclear. Models learn chaos. Stabilize, clarify, and instrument first; the algorithms can wait.

Another weak spot: artisanal work with human-driven, uninstrumented variability. Vision also struggles without controlled lighting and fixturing, or when SKUs change constantly without a retraining plan. In those cases, fix the process—go lean before AI.

Start with your data. Make sure you’re capturing the right information—add sensors where needed and connect your systems so data flows properly. Keep everything consistent, like labels, timestamps, and units. Most importantly, assign ownership and clear processes so your team trusts the data and actually uses it to make decisions.

Integrate without disruption. Favor edge deployments, non-invasive sensors, and APIs. First, run models in parallel with current checks and human-in-the-loop reviews. Only then automate responses. That avoids surprises and builds shop-floor confidence.

Focus on fixing your process before bringing in AI. Organize your work areas, reduce changeover time, and create clear, consistent ways of working.

Once things are stable, AI can learn the right patterns and help improve performance—without constant firefighting.

Measure results with small, quick pilot projects. Set clear check-ins at 30, 60, and 90 days, and track key metrics like efficiency, waste, changeover time, and lead time. Focus on real savings—like reduced downtime, lower material use, and energy costs—as well as improvements like fewer rush orders. Before you start, define what success looks like so you know when to scale—this helps avoid getting stuck in endless testing.

Talent and upskilling: operators become sensor-savvy problem solvers; technicians handle edge devices; quality stewards labeling and validation; IT/OT secures data and pipelines. Provide microlearning, hands-on labs, and mentorship so capability grows alongside projects—not behind them.

Practical first steps: map your value stream; run a data and tech gap check; pick one line or cell for a focused pilot; and schedule weekly reviews with ops, maintenance, quality, and IT.

That’s today’s tour of what works, what doesn’t, and how to implement AI safely. If this helped you think in people-process-technology terms, share it with a manufacturer who needs a practical path forward. Thanks for listening.