Artificial intelligence is changing construction faster than most people realize—but it's not about replacing skilled workers. It's about giving contractors better tools to work smarter, communicate more effectively, and solve some of the industry's biggest challenges. In this episode of Building With Metal, Christian Rosas, President of Nexus Builders, joins us to discuss how AI is already making an impact on job sites.
We explore practical applications for contractors, the ongoing labor shortage, using iPads to simplify onboarding, overcoming language barriers, and why successful companies will view AI as a powerful assistant rather than an autopilot.
Christian Rosas: "I think most of it is coming from putting AI in the driver's seat rather than in the passenger seat, where they're able to oversee things that are really organic in construction, that AI is just not ready to take over."
Christian Rosas: "You can't just throw AI at the problem and so it can go away. Sometimes that could be a huge mistake."
Christian Rosas: "I still want it to be me."
Christian Rosas has an analogy that stuck with everyone in the room.
"I think most of it is coming from putting AI in the driver's seat rather than in the passenger seat, where they're able to oversee things that are really organic in construction, that AI is just not ready to take over."
Rosas runs Nexus Builders, a metal framing contractor working almost entirely in self-storage. About 90 percent of the company's work is installation. He joined the Building With Metal podcast to talk with host Kathi Miller about how his company uses AI, and, just as usefully, where it has let him down.
He is not an AI skeptic. "I'm pro AI," he said. "It's more of the how we use it that's been very, either very helpful or just not good at all."
Ask him what contractors get wrong at the start, and the answer is not a tool choice. It is scope.
"There's a big novelty with AI," Rosas said. "One of the first mistakes is just implementing it, where it becomes overwhelming, where you're not using it properly."
The fix is research before rollout. Figure out where you actually need help. Then figure out whether AI is ready for that particular job. "You can't just throw AI at the problem, and make it go away," he said. "Sometimes that could be a huge mistake."
The wins at Nexus are not glamorous. Repetitive administrative work. Estimating support. Small tasks he can check quickly.
Excel is his favorite example, partly because it is so ordinary. Like a lot of people, Rosas thought he was good with spreadsheets until he watched someone who was really good.
"I feel like I'm a lot better with formulas just because I use AI for that," he said. "If I'm trying to get somewhere and it's just kicking back and it's not working for me, that's where I'm able to use those formulas, and it helps guide me, whether it's through screenshots of my issues that I'm having."
The result is cleaner, more automated sheets instead of sprawling ones held together by hope.
He also uses AI as a sounding board rather than a source. He writes the idea. He runs it past AI. He brings it back and looks hard at what came out. He stays the creator.
The bigger shift at Nexus happened in the field, not the office.
The company has been building 3D models of its projects for a while, then putting them on iPads so crews can see the building before they build it. The models include step-by-step instructions for putting the building together.
For a foreman bringing on someone new, that changes the conversation. "It's very easy, for the foreman, to actually show them on the iPad, 'Hey, this is what we're building. This is the envelope. This is what a base track looks like. These are the studs. Here are the dimensions and whatnot.'"
Miller pointed out how much has changed. Five years ago, a job site had one or two sets of drawings in a central location, and people walked back and forth to look at them. Now the instructions are in everyone's hands.
Rosas says the quality has improved along with the speed. His crews still read prints, and the company still runs a program to teach them. The models just get people there faster. A 2D plan and a 3D model of the same assembly, side by side, close the gap.
He is quick to add that this is not only for the inexperienced. "I've been doing this for quite a while, and it's still very helpful for me."
The crews took to it faster than he expected. "We tried it once years ago, and I remember hearing that on the next project, they were already asking, 'Are we gonna have a 3D model on this one?' So they really loved it."
Does any of this solve construction's labor problem? Rosas does not think so, at least not directly.
What it does is widen the pool. Metal framing for self-storage is a specialty in its own right. A framer who has spent years on hospitals or schools carries real skill, but cannot walk onto a storage job and know how it goes together. Visual tools and software shorten that ramp.
"I think with the help of AI and software, and iPads out in the field, it makes it a lot easier so that you don't feel the shortage as much."
The communication gap on his job sites is not only about language. It is about vocabulary.
Every component in metal framing has a proper name. Out in the field, nobody uses it. Crews invent their own terms, and they do it in both English and Spanish, so two people speaking the same language can still be talking past each other.
Nexus is building a shared reference: a list of components with the names people actually use, images of what each one looks like, and an explanation of what it does. Rosas treats it as ongoing training rather than a one-time document, because new people keep arriving.
AI helps at the edges here too. He put a tool on an iPad so he could translate documents for crews in the field as communication came through.
Marketing is where Rosas has been burned.
AI-generated images and posts have a sameness to them, and he noticed his own output drifting into it. "I started seeing this looks like this company and that company," he said, describing LinkedIn posts that all shared the same rhythm.
That sameness costs engagement. It is also, in his view, a bigger problem with younger employees, who spot the pattern and read it as a lack of human involvement.
So he changed his process. He writes his own thoughts first, even if they're rough. He runs the draft through AI for structure and cleanup. Then he takes the last pass himself and cuts anything that does not sound right.
"I still want it to be me."
Miller has run into the same limit with images. AI is useful for generating a concept she can hand to a graphic artist, which saves everyone two or three rounds of drafts. But asking it to make one small edit to a finished piece produced changes she never requested. Photo and image editing, she said, is still not something she trusts.
Rosas expects AI to stop being a separate destination.
The standalone tools are still there, but the software his company already runs is absorbing AI directly. Procore. SketchUp. Revit. Part of his job now is learning what those built-in features can do.
"I don't think we're gonna be seeing AI as... It's just gonna be a normal day-to-day now, whereas you're working inside your software, it's just gonna be there."
Miller has watched that happen already. Friends recently told her they don't use AI. She pointed out that they do, every time they run a search.
For a contractor who has not touched any of this, Rosas keeps it simple. Start with a repetitive administrative task. Something small enough to check and clear enough to measure. Get a result you can point to, then build from there.
"Before just throwing AI at everything, I think you need to implement it somewhere where it works for you."
Miller suggests going one step further and starting outside the business entirely. Pick a tool; most have free versions, and plan a trip. Tell it where you are going, what you want to see, and when you need to leave. The itinerary comes back in seconds.
Play with it first. Then decide what it is good for at work.
McElroy Metal manufactures metal roofing and siding products for residential, commercial, agricultural, and industrial applications. Headquartered in Bossier City, Louisiana, with plants and service centers across the United States, McElroy supplies contractors, architects, designers, and building owners with metal panel systems, accessories, and the engineering expertise to use them well.