Eleven things small businesses are actually doing with AI — and the one most owners still aren't.

The setup

Most owner-operators I talk to have one of two reactions when AI comes up.

Reaction one: a faint grimace, the kind people make when their accountant says the word "BAS". They've heard about it. They know they should probably do something. They have done absolutely nothing, and would prefer to keep it that way until 2028 at the earliest.

Reaction two: a wild gleam in the eye, usually after a forty-minute YouTube binge, in which they explain that ChatGPT is going to replace their bookkeeper, their dispatcher, their compliance officer, and possibly their wife by the end of the quarter.

Neither reaction is correct. Both are very Australian.

Here's what's actually happening, with real numbers attached. Intuit's 2026 Australian SMB report — published last month — found that regular AI use among Australian small and medium businesses climbed from 40 percent in July 2024 to 69 percent by January 2026. Daily use went from 9 percent to 28 percent. Of the businesses using it, 79 percent report measurable productivity gains. Forty-three percent say it's contributed to higher revenue.

That isn't a tech-blogger fantasy. That's the cleaner who runs commercial sites west of Cairns asking ChatGPT to rewrite his quote template at 9pm on a Sunday so he can sleep.

So. What are people actually doing with it.

The eleven

I'm not going to give you tool names or recipes. Every list of "the 17 best AI tools for small business" reads like it was written by someone who has never met a small business, and they all recommend the same three things their affiliate links pay them for. The point isn't the tool. The point is the job.

Here are the eleven jobs where I keep seeing AI land in regional Queensland SMBs in 2026, ranked roughly by how unglamorous they are:

1. Writing the email you've been avoiding for three weeks. Pricing rise to a long-term client. Bad-news to a supplier. The follow-up to the apprentice's parents. AI gets you a draft in 90 seconds. You edit. You send. The email gets written. Bookkeeping for the soul.

2. Turning the voice note into something legible. You're driving back from a site. You record three minutes of "yeah so the issue is the bracket's the wrong gauge and Steve reckons we need to". AI turns it into a one-page brief, a quote variation, or a toolbox talk before you've pulled into the yard.

3. Quote and tender drafting. Not the price — the price is yours. The prose around the price. Scope sections, exclusions, assumptions, schedules. The bits that used to eat your Sunday.

4. The Centrelink-style admin form. Grants. RTO paperwork. Workplace health and safety questionnaires from primes. Insurance renewals that ask the same eleven questions in slightly different language. AI ingests last year's answers and proposes this year's. You read, correct, sign.

5. Reading the contract that just arrived. Not replacing the lawyer. Replacing the bit where you don't read the contract at all because it's 47 pages and you're tired. AI flags the indemnity clause, the rise-and-fall absence, the auto-renewal you missed. You then decide what to do with the lawyer's time.

6. Building the procedure that should already exist. The one your insurer asks for. The one the apprentice would have read on day one if it existed. AI takes a video of you doing the thing, plus a transcript, plus three voice notes, and writes a draft SOP that's 80 percent right. You fix the 20.

7. Customer follow-up that actually happens. Quote sent, no response, three weeks later, crickets. AI drafts the nudge. You read it, send it, win the job, wonder why you didn't do this in February.

8. Reading your own P&L. You've stared at the Xero dashboard for two years. You know the top line. You don't know which job, which crew, which day of the week is actually killing you. AI ingests the data, finds the patterns, tells you in English. You then decide what to do about it.

9. Toolbox talks and inductions. Five minutes of content, in your voice, dropped into a workflow that runs every Monday morning, in the language your crew actually speaks.

10. Compliance scanning. The latest WHS amendment. The new heavy vehicle ruling. The Fair Work order that came in three weeks ago and applies directly to your subcontractors. AI watches. You don't have to.

11. The thing you've been "going to write" for two years. The capability statement. The website copy. The LinkedIn page. The customer one-pager. AI gets you 70 percent of the way there. The 30 percent that's left is the bit that needed your judgement anyway.

The one most owners still aren't doing

Here is the use case that almost nobody in regional Queensland is doing properly, and it's the one with the biggest dollar return:

Setting AI up to do the same job, the same way, every week, without being asked.

Almost every owner-operator using AI in 2026 is using it like a vending machine. Walk up, push a button, get a thing. That's fine. That gets you back four hours a week.

What gets you back a person is the next step: standing AI up as a piece of infrastructure that runs on a schedule, reads from your actual systems, produces the same output to the same standard every single time, and lands on your phone before you've finished your coffee.

That is not a tool. That is an operating system change.

It also isn't something you do on a Sunday afternoon between dinner and the news. It needs scoping, governance, data plumbing, and someone accountable for the result. Doing it well is the difference between an owner who now has AI, and an owner whose business has AI.

That's the line worth being on the right side of.

What I'd suggest

If you've read this far, you're already past the first reaction. You're somewhere in the middle — interested, slightly suspicious, not sure where to start.

The honest answer is: start with one of the eleven. Pick the most painful one. Spend a week on it. See whether the output is good enough to keep, or only good enough to mock at the pub.

If it's good enough to keep, the next conversation is about the twelfth. The infrastructure one. The one with the real money behind it.

The vending machine is fine. The operating system is better.

John McKnight is Director of Northern Vector Pty Ltd, an operations consultancy based at South Mission Beach serving North Queensland. Three decades across Defence, civil construction, mining, mobile maintenance, transport and logistics.

Sources: Intuit QuickBooks — AI Adoption in Australia 2026 Impact Report · IBS Intelligence — AI adoption surges among Australian SMEs, May 2026