The Truth About Adopting AI in Manufacturing

Most of the manufacturers we talk to have been told they need AI. Usually by a vendor, sometimes by their kids. They're not against it. They have a dock that's a day behind, a rep who spends her mornings keying orders, and a clear memory of the last software rollout.

Jace Preston is Director of Manufacturing at Capital Steel & Wire. Capital Steel started working with us in August 2026. In September, our co-founder Josh Ross interviewed him on stage at Tech Week Grand Rapids about how it's gone so far. The video is above. This is the written version, organized around the questions we get asked most.

Table of contents

• About Capital Steel and Jace
• Where do you start?
• Will my people use it?
• What else did you look at, and why Scalable?
• What results are you seeing?
• How do we start?

About Capital Steel and Jace

Capital Steel & Wire is a cold-finished bar mill in the Lansing area with three manufacturing locations. The company turns 25 this year. Jace's father Dustin started it from the ground up with Kelly Preston, who is president today. The board has six directors. Four are family.

Jace came up through sales before moving to operations. He knows how long an order takes to process because he used to process them.

The interview took place September 17, 2026 at Smarter Manufacturing Starts Here, a Tech Week Grand Rapids event we co-hosted with Prescott AI, in partnership with The Right Place and the Michigan Manufacturing Technology Center. The audience was mostly people who run plants. Jace had no script.

Where do you start?

Capital Steel started by listing their bottlenecks.

We originally looked at all of our bottlenecks in our process. Not necessarily, "Hey, let's throw another body at that." It was more just, "Hey, this isn't working. It's taking more time." Those were the areas that we wanted to address.

He calls this work the slack. Sending the paperwork, doing the busy work, the hours that pile up around the job itself. It also scales with the business. More orders, more shipments, more steel coming in the door, and the slack grows with them.

They picked order entry first. Jace knew from his sales days how much typing and clicking it takes to get from a customer's email to an order in the ERP, and that process hasn't changed much in 10 to 20 years. Time saved there goes back into selling: "make more phone calls, make more sales, or get in front of more customers."

Receiving is next. Today a truck comes in, material gets unloaded and set down, and can't get stored away until it's been entered and verified in the ERP against what was purchased.

Currently we're touching it and leaving it in a spot for three hours. Got to come back, remember every tag number, everything that came in, all the paperwork, before it gets received in.

If you're looking to find a specific use case for AI, consider using the same approach as Jace and the team at Capital Steel: find the bottlenecks that scale with volume and are just slack, not the skilled work you hired your people for.

Will my people use it?

Jace was worried about this too, but he understood this was a change management problem more than a problem with the tools. There is usually some fear around adopting AI, and on stage he called the job displacement conversation "always the tricky one."

He did two things. First, he addressed the elephant in the room and told his people what the leadership team was thinking: this is the vision, this is what we're going to be doing, and it's not here to take any jobs away. It's here to help the team and make the process more efficient.

Then he got our team onto the floor with his shipping and receiving manager to walk the process and run live demos for the people who do the work.

From there they've seen the live demos, talking with them, just kind of let their guard down, and they got super, super excited about it. That was the big thing, getting you guys on the floor.

Josh asked how he's preparing the dock crew for go-live on receiving. Jace's answer was to keep talking with them. Let people say what they're worried about. Talk about where the company is headed in ten years. Hear their issues before anything changes. He expected more resistance than he got: "There hasn't been as much pushback as I originally thought." He credits that to the team having met us on their floor before anyone asked them to use anything.

He was also clear about what the agents don't touch. Capital Steel is always going to need people running machines, managing pressures and angles, getting steel bars in straight. "There's some feel to that that an AI can't bring."

What else did you look at, and why Scalable?

There are three ways to go.

Replace your ERP with one that has AI built in. If you were already planning a new ERP, look at these. If you weren't, know what you're signing up for. You migrate your data, retrain every department, and change how you work to fit the software. To make a change like this takes time and money from your team. And a vendor with thousands of customers is not going to work with you to understand and adapt to how your shop operates.

Build it yourself with Claude or ChatGPT. You can get a long way with one motivated person. We've written about how. You also become a software company. You own the ERP integration, the uptime at 6 a.m. when a truck is waiting, and the knowledge if that person leaves. Jace had been dropping "little nuggets" about AI to the executive team for a while and "they maybe didn't go over super well at first." The tools were already there. Nobody owned the rollout.

Work with a partner. This is what Capital Steel did.

The moment we met Josh, I was like, this is a perfect match. This is exactly who we need. We need a young company like this that's growing its own business to grow with us. So we see it more as a partnership, and it's been an absolute game changer for us from the jump.

Josh asked how much of the improvement came from new technology versus changing the process. Jace said it was his favorite question of the day.

We don't have to reinvent the wheel. We don't have to change our process. We're just making it more efficient and streamlining the time it's going to take to get that task complete.

Half of Capital Steel's board was sold after the first meeting. The other half came around when Josh and Gabe visited the plant and they could "see it, feel it, hear it from the horse's mouth." If you have skeptics on your team, invite them to that meeting.

What results are you seeing?

Josh asked about KPIs. Jace said what most people won't say on stage: "It's still ongoing. We're so new to this whole AI sector."

Capital Steel started in August. Order entry is live. Receiving is being built. He doesn't have a return-on-investment slide yet, and we wouldn't trust one from anyone a month in.

What they track is time. They knew how long each bottleneck task took before, and they watch how long it takes now.

What they've seen so far, in Jace's words: "game changer from the jump," and the rollout has "been easy for us." Reps key fewer orders. The receiving agent will tie into the ERP so material can be "received in, tagged, put a red bow on it, and forgotten about until the next time we have to ship it out."

If you want some tangibles to measure at your shop, here are a few places where we've seen the biggest time savings:

• Sales: Minutes per order, from the customer's email to entered in the ERP.
• Receiving: Time from truck arrival to material received in the system.
• Accounts Payable: Time spent per week matching documents (invoices, purchase orders, and packing slips) together and entering invoices into the ERP.
• Quoting: Number of hours it takes to get a quote out to the prospect.

How do we start?

Jace's plan for the next year is "knocking down those small dominos." One process at a time. Order entry, then receiving, then whatever has the most slack after that.

Working with us looks like that. We come to your shop. We walk the process with the people who do it. We pick one task together and agree on how to measure it. We build an agent that fits how you already work, and we demo it to your team on your floor before it goes live.

You don't need a strategy or a budget line to call us. Capital Steel didn't have either. You need a bottleneck and someone willing to walk us through it.

Every shop runs differently, and a tool built for the average shop ends up fitting nobody. That's why we start on your floor.

Want to sit down and figure out what makes sense for your shop? Book 30 minutes with us. Bring the bottleneck.