From "tried ChatGPT once" to a weekly habit

By Jason Sibley

The problem

You know the story because you've probably lived it. Someone mentioned ChatGPT at a trade show, or your nephew showed you on his phone, or you read a piece (maybe even one on this site) telling you it would change how you work. So you opened it up, typed something in, and got back three paragraphs of nothing in particular. Bland. Generic. The sort of copy any competitor could have written. You closed the tab and went back to doing the quote by hand.

That's not a personal failing. It's what happens when a new tool gets tried once, under no real pressure, with no plan for what happens next. A hammer used once and put in a drawer doesn't prove anything about hammers. It proves you hadn't decided what you were building.

Small firms don't have a training budget or an IT department to nudge them back to the tool a second time. If it doesn't click in the first five minutes, it's gone. And that's roughly what's happening across the country right now, at scale.

What the numbers say

Three different bodies have measured AI use among small UK firms recently, and it's worth being honest that they don't agree with each other, because they're not measuring the same thing.

SME Magazine reported on 9 June 2026 on research from Enterprise Nation's Tech Hub, working with Google, Sage, Dell Technologies and Square, called "UK SME Digital and AI Adoption: The state of play in 2026". It surveyed 1,320 micro and small businesses and found only 21% use AI regularly, and just 6% have it embedded into daily work across the business. Just over half, 57%, described themselves as highly or moderately digital, which tells you the gap isn't really about being anti-technology. Among firms that had adopted AI, 65% said their use had grown over the past year. But adoption is very uneven by size: only 34% of one-person businesses use AI at all, against 68% of firms with 50 to 249 staff. It's also uneven by region: London sits at 30% and Scotland at 27%, while the North East is at 10% and Northern Ireland at 6%.

The Office for National Statistics published its own figures on 20 July 2026, in "Artificial intelligence in UK businesses: 2023 to 2026". It found 28% of businesses with 0 to 9 employees use at least one AI technology, against 49% of businesses with 250 or more staff. Among the businesses that do use AI, though, the smallest ones use it more intensively: 17% of 0-9 employee AI users describe their use as extensive, against 9% of the big firms. Among all employees, 55% report using AI for work or education, but only around 35% of businesses with 10 or more staff report using at least one AI technology themselves. So plenty of people are using AI at work before their business has formally decided anything.

Then there's DSIT's AI Adoption Research (DSIT is the government's science and technology department), updated on GOV.UK on 13 February 2026. It's based on 3,500 telephone interviews with firms employing at least five people, run between February and May 2025. It found around one in six (16%) used at least one AI technology, with micro businesses at 14%. But of those who had adopted it, the habit was strong: 80% use it at least weekly, with 53% saying they use it constantly and 27% at least once a week.

Why do the numbers range from 6% to 21% to 28%? Because "regular use", "any use" and "embedded into daily work" are different bars, and the surveys cover different sizes of business. Don't get hung up on which figure is "right". The pattern underneath all three is the same: most micro firms haven't formed a habit, but the ones who have, use it a lot.

Why the habit doesn't stick

The Enterprise Nation research names cost (53%), lack of skills (46%), privacy and data concerns (38%) and lack of time (37%) as the main barriers. DSIT, asking what stops or stopped businesses adopting AI, found something slightly different at the top: 71% cited a lack of identified need, meaning they simply hadn't worked out what they'd use it for, ahead of limited skills and expertise at 60%. The ONS's earlier analysis found similar themes: difficulty identifying business use cases, plus cost and a lack of expertise.

Put that together and the picture for a micro firm is fairly clear. The tool usually works well enough. What's missing is someone deciding "this is the one thing I'll use it for, every week, on this day". Without that, you're paying for something with no clear return. You never build the skill, because you never repeat the task. And the data worry never gets settled, because nobody looks into it properly.

A weekly habit that fits a micro firm

Pick one recurring task. Not five. One.

Say you run a joinery firm and every Monday you're chasing quotes that went quiet. Or you're a florist and every Friday you've got a backlog of customer emails to answer. Or you run a small online shop and you need one social post drafted a week. Any of these works. What matters is doing it every week.

Here's how to set it up:

1. Choose the task and the day. Monday quote follow-ups, Friday customer email round-up, whatever fits your week.
2. Start on the free tier. You don't need to pay until it's properly part of your week.
3. Write one prompt that does the job reasonably well, and save it in a note on your phone or desktop. Don't rewrite it from scratch each time.
4. Never paste in customer personal data you wouldn't happily put in an email to a stranger. If you're not sure what that covers, our plain-English guide to AI and business data walks through it.
5. Give it 20 minutes. That's the whole slot.
6. Keep a two-line log, in your own words, of what it saved you that week. Not a spreadsheet. Two lines.

That log matters more than it sounds. It's the thing that turns "I suppose that was useful" into a decision to keep going.

What to do about each blocker

Cost. Start free. Don't pay for anything until the habit is weekly and you can point to the log and say it's worth it.

Skills. You don't need a course. One saved prompt, reused and slightly improved each week, will get you further than a training session you attend once and forget.

Privacy. Fair worry. Deal with it properly rather than avoiding the tool altogether. See the data guide linked above before you start.

Time. Don't create a new slot in your diary for this. Attach it to a task you're already doing, on a day you're already doing it. The habit rides on the back of an existing one.

The one decision to make this week

Not "should I use AI". That question is too big and too vague to act on, which is part of why DSIT found most non-adopters simply hadn't identified a use case. The decision is smaller: which one task, on which one day, will you try for four weeks running. Write it down. Save the prompt. That's it.

If you want a slightly longer run at this, Pick one AI workflow and stick with it for 30 days goes through the same idea in more depth.

A soft word before you go

If you've read this far and still aren't sure which task to pick, My AI Apprentice runs SMB AI sessions for exactly that, and a free website review is there if your site is the thing eating your week. No pressure either way. The four-week test above works fine on its own.