Research note · September 2026 · about 6 minutes · this is the thinking behind what Cutting Hedge builds
Being curious about AI and handing over money for it are two very different things. You can try a free chatbot, sit through a demo of an “AI receptionist”, and still decide it isn’t worth it. That’s usually the right call — until something is solving a problem you’d actually pay to get rid of.
The short version
- 46% of US small employers use AI — but only 7% of those have it fully built into how they work.
- Businesses with staff pay for AI more often than sole traders, and office-type industries far more than hands-on ones.
- There’s no reliable figure for what Irish tradespeople will pay. Anyone quoting one is guessing.
- The things worth paying for are narrow: fewer lost enquiries, quotes that get chased, getting paid faster.
Lots of trying, not much committing
The US Federal Reserve’s 2026 report on small employer firms, based on a 2025 survey of 6,525 businesses, found 46% already using AI and another 15% planning to start within a year.
But look at how deep it goes.
7%
of AI users had fully integrated it. About half were still experimenting; 44% had partly integrated it.
The most common uses were writing or marketing (83%), personal productivity (61%) and planning (51%). The top problems were accuracy (46%) and getting tools to fit how the business works (43%). Small businesses try AI long before they commit to it.
What businesses actually pay for
Surveys tell you what people say. The JPMorganChase Institute looked at what they spend: payments from 4.6 million US small businesses to more than 500 AI tools between 2019 and 2025.
- Newer businesses take it up far faster. Businesses started in 2025 reached 10% paying for AI in about six months; for those started in 2019 it took more than six years.
- Businesses with employees pay for AI more often than sole traders, and the gap has widened since 2023 — from 5.4 to 10 percentage points.
- Knowledge-heavy industries lead. By 2025, 39.3% of information-sector businesses were paying for AI, against 5.4% in transport and warehousing.
That study only counts paid tools, so it misses free ones. But the pattern matters for trades. A one-person, hands-on business can have a real need and still have less time, less tidy data and less patience for a complicated setup than an office-based firm.
Construction firms: interested, not reckless
The Associated General Contractors of America’s 2025 outlook survey drew 1,109 responses from US construction firms. Of the 965 who answered the technology question:
| Planning to increase spending on… | Share |
|---|---|
| Artificial intelligence | 44% (1% decreasing, 55% no change) |
| Estimating software | 35% |
| Project management software | 35% |
| Service management software | 12% |
Their biggest IT headaches were keeping data secure (41%), the time it takes to set up and train on something new (39%), staff resistance (36%), communication between site and office (32%) and getting their own systems to talk to each other (28%). Interest is real — so is the worry about the effort.
What’s likely worth paying for first
Not “innovation.” Something narrow. The offers that make sense solve a problem that keeps happening, don’t make you change how you work, show a result you can see, and leave the judgement with you.
| What it does | The problem | How you’d know it’s working |
|---|---|---|
| Missed-call reply | Enquiries vanish while you’re on a job | Conversations recovered, quotes sent, jobs booked |
| Quote follow-up | Quotes go out and never get chased | More quotes answered, more work won |
| Enquiry sorting | Time lost on unsuitable or half-finished enquiries | Better details before you ring back |
| Voice notes to job records | Job notes are patchy or late | Complete records, fewer things missed on the invoice |
| Customer update drafts | Customers keep ringing to ask how it’s going | Fewer interruptions, happier customers |
| Posts from finished jobs | Good work never makes it online | Regular posts, and enquiries that mention them |
None of these should replace your expertise. A system can draft, organise or remind. You approve prices, technical statements, safety information, dates and anything else you’re committing to.
What you’re actually paying for
If you already use ChatGPT, you don’t need to pay someone for AI. What might be worth paying for is everything around it: working out which job to fix, writing the messages you’re happy to send, setting what it must never say, connecting it to your phone or forms, and checking the first month’s results with you.
That fits the evidence. Accuracy and fit are the biggest problems for businesses already using AI, and Irish business leaders say they want practical tools, templates and real examples. The value is in making it safe and useful for your business, not in the software itself.
How to buy without taking a big risk
Because nobody knows the right price for this yet, small steps make more sense than a big package.
| Step | What it should give you |
|---|---|
| A look at how you work now | One clear recommendation, in writing, for the job most worth fixing |
| A short trial of one thing | A fixed scope, a set length (say 30 days), your approval on anything that goes to customers, and a clear way out |
| Ongoing support | Only once the trial has shown a result — covering upkeep, message changes and a regular look at the numbers |
Before you pay for anything, ask:
- What result would make this worth the money — and how will we measure it?
- What happens if it gets something wrong in front of a customer?
- What would make me cancel, and how easy is that?
- What information do I have to hand over, and where does it go?
A one-person business with irregular enquiries shouldn’t be sold the same thing as a contractor with several crews and an office. One recovered emergency job might easily cover the cost for a plumber; a business with a long renovation sales cycle might wait months to see it pay back.
What this research can’t tell you
Every spending figure here is from the US. The Federal Reserve survey is a convenience sample of businesses with staff, not trades specifically. The JPMorganChase data counts only paid tools on its own business bank accounts. The AGC survey measures what construction firms plan to spend, not what they spent.
I couldn’t find a single trustworthy figure for what an Irish tradesperson would pay for any of this. So there isn’t one in this post.
Tell me what would be worth it
I’m running a short survey asking local tradespeople which problem they’d most like fixed, what they already pay for, and how they’d prefer to pay for help if they wanted it. It takes 6–8 minutes, and leaving your contact details is optional.
Keep reading
- 10 practical ways AI could help Irish trades businesses
- Why an AI cash-flow assistant may matter more than an AI chatbot
- Could AI help small contractors win more repeat work?
- AI-assisted quoting: useful shortcut or costly mistake?
Sources
- Federal Reserve Banks — 2026 Report on Employer Firms (Small Business Credit Survey, fielded September–November 2025)
- JPMorganChase Institute — Understanding the use of AI among small businesses (April 2026)
- Associated General Contractors of America — 2025 Construction Outlook National Survey Results (PDF)
- ICT Skillnet and AI Ireland — AI Readiness Pulse: what Irish businesses are really saying about AI in 2025