A super at one of our jobs has been quietly using ChatGPT to clean up his daily reports for nine months. Nobody told him to. Nobody told him not to. He just got tired of writing the same paragraph about weather impact in five different ways every Friday afternoon.
He didn't ask permission. He didn't tell anyone. He just stopped hating Fridays.
That's most of what's happening in construction offices right now.
Why leadership waits
The hesitation in leadership rooms is real. The tools work some of the time. They don't work other times. They occasionally hallucinate a sub's name or invent a spec section. So leadership waits. "Let's see how it plays out." "Let's wait until it's mature."
Mature is doing a lot of heavy lifting in that sentence.
The real risk is falling behind
The real risk isn't that AI is imperfect today. The risk is that the company waiting for it to be perfect is going to wake up two years from now and realize their PMs don't know how to think with it. Their estimators are slow. Their proposal team takes three days to do what their competitor does in a half day. The competitor's people aren't smarter. They've just been at it longer.
Using AI is a skill. Knowing what to ask, when to trust it, where it lies, how to verify it. None of that is learned by reading a vendor's white paper.
Let people use it out loud
The cheapest way to start: stop pretending it's a special tool.
Tell the team they can use it. Out loud. Not "for special cases." Not "after approval." Just—use it on the work that's draining you, and tell us what you find.
When you don't say that, employees use it anyway. Quietly. Without guardrails. Without sharing what works. The risk you were avoiding by going slow is the risk you actually have, just less visible.
Start where the work hurts
Start with the work nobody wants. The Friday reformatting. The proposal cover letter that's a copy of a copy. The five-page subcontract exhibit that needed to be cleaned up two weeks ago. The submittal log that arrives in someone else's format and has to be retyped.
These tasks are not creative work. Removing them is not stealing anyone's job. It's giving an estimator forty minutes back. A PM ninety minutes. A super his Friday.
The emotional shift matters more than people admit. When AI shows up first as relief, it gets trusted. When it shows up first as a top-down mandate, it gets resented. Same tool. Different framing.
A note on ownership
The mistake leadership tends to make is centralizing the AI roadmap. A committee decides what tools are approved. A working group defines use cases. A vendor sells a platform. Six months later, adoption is flat because the people closest to the work weren't asked.
A better move: set the boundaries (what data is sensitive, what requires human review, who owns final outcomes), and then let the PMs, estimators, and supers tell you where the friction is. They know. They've been complaining about the same five tasks for years. Give them permission to fix them.
It's a culture shift, not a rollout
The shift isn't technical. It's cultural.
Trust the team to experiment. Trust them to verify. Trust them to come back and say "this one didn't work." Most of them will. The ones who learn alongside the tools today are the ones who run the company tomorrow.
Two years from now you won't have a choice. The companies that started growing into this in 2024 will look like they have unfair leverage. The companies that waited will be doing what waiting companies always do—running a hiring search for someone else's people.
The super still hates Fridays a little. But he's home for dinner.