AI in the Trades: The Real Adoption Numbers, Survey by Survey

The Baseline Nobody Wants to Quote

Start with the most boring, most reliable number available. The Census Bureau surveys businesses every two weeks, and its national file puts AI use at 21.8% of US businesses for the two weeks ending July 26, 2026. That line has climbed slowly and unevenly, moving between 17.5% and 21.8% since December 2025. The Federal Reserve's own accounting lands in the same neighborhood a few months earlier: about 18% of firms had adopted AI by the end of 2025, and that was after 68% growth in a single year.

Then there is the number that confuses everyone. The same Fed note reports about 41% of workers say they use generative AI at work. Both figures are true at once. Big companies adopt at high rates and employ most of the workers, so worker-weighted numbers run more than double the firm-weighted ones. When a headline claims 40% of businesses use AI, someone mixed up those two measurements.

And in the building trades the floor sits lower still. NAHB's builder panel found that a majority of single-family builders use no AI at all, with marketing the top use case at 20%, market analysis at 11%, and everything else under 5%. If a vendor deck told you adoption is universal, it was selling you something.

Census puts a number directly on the trades, too. Its sector file shows AI use in construction at 14.9% for the two weeks ending July 26, 2026, against the 21.8% national rate in that same read. The national rate drifted up over the summer; the construction gap did not close. A Census working paper says the same thing in words: sectors built on physical output, construction among them, run at much lower adoption rates than the knowledge sectors, where Information sits near 38%. Construction is not on the front of this curve. It is roughly two thirds of the national rate.

The Size Split Is the Whole Story

Buried in the Census data is the line that matters most for anyone running a shop with a handful of trucks. The recent growth in AI use came almost entirely from bigger companies; in the Bureau's own words, AI use did not change significantly among firms with fewer than 20 employees. Adoption runs 32% at firms with 100 to 249 employees and 37% at 250 or more, while firms with four or fewer people sit under 20%.

Read that gradient as a competitive map. The large regional outfits bidding against you for the same replacement jobs are on the adopting side of the line, and the movement is all happening on their side.

The question for a small shop is whether you are fine with that gap getting priced into your close rate.

The adoption gap is not a technology gap. It is a size gap, and it is widening.

Where the Real Work Shows Up: The Office, Then the Estimate

Construction gives the clearest picture of what firms actually do once they start. AGC of America surveyed 951 firms at the end of 2025 and found 61% use AI or plan to increase AI investment, up from 44% a year earlier. The use cases are the telling part: office and administrative work leads at 45%, then estimating at 23%, design and preconstruction at 20%, and HR at 16%. Nobody starts with the truck. Everybody starts with the paperwork, and the first line-of-business job it touches is the estimate.

The Numbers We Left Out, and Why

There is a second tier of trades AI statistics floating around, and it is the tier that gets quoted most: three in four residential contractors calling AI essential, measurable impact doubling year over year, specific hours saved per week. Every one of those comes from a field-service software company surveying its own customer list and publishing the result as an industry finding.

That is not research. A vendor asking its own paying users about the category it sells will get a flattering answer, and a shop that already bought software is not a representative sample of shops. We are a software company too, which is exactly why we are not going to launder someone else's marketing into this page and call it data. If a number here does not carry a government agency or a trade association behind it, it is not here.

What survives that filter is thinner and more useful. Construction sits at 14.9% AI use against 21.8% nationally. A majority of single-family builders use none. Among the construction firms that have started, AGC's read is that the work lands in the office, not the field. Put those together and you get a trade that is behind, moving, and starting with paperwork.

We have taken the two objections we hear most apart in full elsewhere: the too-small belief in 'We're Too Small for AI' Is Costing 3-Truck Shops Real Money, and the what-does-it-actually-do question in AI Tools for the Trades: Proof Over Hype.

Where the Hours Actually Are

There is no trustworthy independent number for how many hours AI gives a trades business back, so we are not going to hand you one. What the neutral data gives you is the location of the work rather than the size of the prize, and that is the more useful half anyway. AGC's use-case ranking puts office and administrative work first and estimating second, which matches where a small shop's unpaid hours pile up: writing and sending quotes, chasing the follow-up you meant to send Tuesday, answering the phone from a crawlspace, remembering to ask for the review.

That list is the exact ground OptimizeIt is built on. The estimate and quote workflow gets a number in front of the customer while the visit is still warm. The CRM keeps every job and contact in one place instead of a dashboard, a notebook, and your memory. Follow-up automation sends the nudge you never get to. Review and referral asks go out at the moment the customer is happiest. On pricing, with annual billing: Core is $79 a month with 2 seats, Plus is $159 with 4 seats and a dedicated business phone line, and Pro is $239 with 6 seats plus the AI Voice Agent that answers and books when you cannot. Extra seats are a flat $25 a month.

Here is the honest way to size the return. Time yourself for one week on quotes, follow-ups, and repeat phone questions. Multiply the total by your billable rate. That is your number, and it beats anyone's survey average because it is actually about your shop.

The Honest Take

This is not for everyone, and the surveys say so. If you are a solo operator with a full book, a waitlist, and no ambition to grow, the under-20% adoption rate among the smallest firms includes plenty of shops doing fine, and you may be one of them. If your office manager runs a paper system that actually works and nobody in the building wants to change it, forcing software into that shop fails more often than it helps. And if you are hoping AI will run the service call, diagnose the unit, or swing the hammer, no honest number in this post supports that; the neutral surveys above put the wins in the office, not the field. But if you are losing jobs to slower quotes, missed calls, and follow-ups that never went out, the gap between you and the firms on the steep end of the Census curve is already costing you money. That one you can close this month.