AI Isn't Coming for Your Superintendents. It's Coming for the Job Description.
There's a stat making the rounds in every corporate leadership newsletter right now: Gartner predicts that by the end of this year, 20% of organizations will use AI to flatten their structure, wiping out more than half of today's middle management roles in the process.
If you run a construction company, your first reaction to that is probably: sure, but that's a software company problem. That prediction is coming out of orgs full of analysts and coordinators pushing spreadsheets between meetings. My PMs are out running jobs. My supers are on-site solving problems that don't exist in a slide deck. That's not us.
I'd push back on that, and I'd do it from the field, not from a research report.
Here's the part of the construction industry data that doesn't get quoted as often: RICS found that 45% of construction organizations report no AI capability at all right now, with another 34% still stuck in early pilot/exploration. Meanwhile ServiceTitan's 2026 industry report found the share of contractors seeing measurable AI results more than doubled year over year, from 17% to 38%. Read those two numbers together and you get the real picture: adoption is accelerating fast, from a very low base, in an industry that mostly hasn't touched it yet. That's not "this doesn't apply to us." That's "we're earlier in this curve than almost anyone, which means the leadership decisions are still ours to make instead of ones we're reacting to."
That's the distinction that actually matters, and it's the one most of the Gartner commentary is missing entirely.
The headcount framing is a distraction
There's a Forbes piece from earlier this year making the case that companies cutting middle managers to "fund AI" are making a mistake. I agree with the conclusion, but I think it's answering the wrong question. The real question was never "does AI let us run with fewer leaders." For a construction company with a real org chart, that question barely applies. You're not overstaffed on leadership. You're usually thin on it. Nobody's sitting around wondering who to cut.
The question that actually matters is: what does AI change about what the leadership you already have is spending their time doing?
Because it does change that. I've watched it change in my own work. A huge share of what a construction manager does day to day isn't leadership. It's translation and paperwork. Turning a schedule slip into a look-ahead update. Turning a field conversation into a daily report. Turning three subcontractors' conflicting updates into one coherent picture for the owner. That work matters, but none of it is why you promoted that person into a leadership seat. You promoted them because they can read a jobsite, read a room, and make a call under pressure. AI is genuinely good at the first category, the translation and reporting layer, and it is nowhere close to good at the second.
What actually gets freed up, and what it needs to be spent on
This is where most of the AI-and-construction conversation goes wrong. It treats the win as "your super files reports faster now." That's real, but it's not the win. The win is what that time gets reinvested in, and that's a leadership decision, not a software decision. If a superintendent gets two hours back a week because reporting and schedule-tracking got faster, and nobody redesigns what he does with those two hours, you haven't changed his job. You've just given him more time to do the same job slightly less stressed. That's a fine outcome. It's also a wasted one.
The companies that actually get something out of this aren't the ones that adopted a tool first. They're the ones who decided, on purpose, what their leaders should be doing more of once the paperwork stopped eating their week: more time walking the job with a foreman who's about to lose a good hire. More time actually training the PM who's technically capable but has never been taught how to have a hard conversation with a sub. More time noticing the two crews that are about to collide over sequencing before it becomes a change order fight. That's not a tooling decision. That's a leadership redesign decision, and it has to be made deliberately or it doesn't happen at all. The time just gets absorbed by whatever's loudest that week, same as it always has.
I've been on the personal-use side of this, not the org-wide side: using AI tools to compress the reporting and admin grind in my own managerial work, on my own time, with public tools, nothing employer-specific about it. What I noticed wasn't "wow, this is fast." It was how much more obvious the actual leadership gaps became once the busywork stopped filling the space where they used to hide. That's the uncomfortable part nobody puts in the adoption stats.
The actual choice in front of you
Nobody is going to hand a construction company a memo saying "redesign your leadership roles now." There's no HR department pushing this, no board asking for an AI strategy deck. It's going to arrive quietly, tool by tool: a scheduling assist here, a reporting shortcut there. Each one on its own will look like a minor efficiency win. The org-chart-level question, the one Gartner's prediction is actually pointing at, doesn't ask itself. Somebody has to.
You can let that stay accidental, or you can decide on purpose what your leadership layer should look like once the paperwork stops being the job. One of those is a strategy. The other is just hoping it sorts itself out.
That second option has a worse track record than most owners assume.
- Gartner: 20% of organizations to use AI to flatten structure by 2026 (BW Businessworld)
- "Why Companies Cutting Middle Managers To Fund AI Is A Mistake" (Forbes, May 21 2026)
- RICS: Builders slow to adopt AI despite perceived benefits (Construction Dive)
- ServiceTitan 2026 Commercial Specialty Contractor Industry Report