Artificial intelligence (AI) has become a part of our everyday lives, from smart assistants to advanced data tools. However, there's a common misconception that AI is here to replace people. The truth is, AI isn't about replacing brilliant people — it's about empowering them to be even better at what they do.
I run four businesses and use these tools every day. They've gone from novelty to plumbing in about eighteen months. What follows is what actually happens when a small UK business puts AI to work: where it pays, where it bites, and what to do about it in the next seven days.
Taking care of the mundane
One of the biggest ways AI helps is by taking care of repetitive and time-consuming tasks. In accounting, AI can handle data entry, reconcile transactions, and flag anomalies. This frees up accountants to focus on higher-value work, such as advising clients and strategising for growth. In creative fields, AI can streamline editing, suggest improvements, or generate basic drafts. The final touch, the human intuition, and the unique flair are things only people can bring to the table.
The important word there is mundane. Not "hard". Not "important". Mundane means the task has a right answer, a repeating pattern, and no real consequence provided a human checks it afterwards. Transaction coding is mundane. Turning a meeting into notes is mundane. Drafting the third version of a job advert is mundane. Deciding whether to take on a client who feels wrong is not, and never will be.
What the mundane is actually costing you — a worked example
Numbers make this argument better than adjectives do. The business below is illustrative — invented, not a client — but the rates are real and the arithmetic is the arithmetic.
Say a small firm has one recurring admin job: pulling information out of emails and attachments and putting it somewhere useful. It takes 45 minutes a day, five days a week, 46 working weeks a year. That's 172.5 hours.
- Costed at the National Living Wage of £12.71 an hour — the rate for workers aged 21 and over from April 2026 — that's £2,192 of wages.
- Add employer's National Insurance at 15% on earnings above the £5,000-a-year secondary threshold. On a salary already past that threshold the extra falls on the whole amount: roughly £2,521 fully loaded, or about £14.62 an hour.
- Now cost the same hours at the owner's rate instead. If you'd otherwise bill that time at £75 an hour, 172.5 hours is £12,937 of capacity you never sold.
Here's the part most AI pitches skip. AI does not make those 172.5 hours vanish. It converts them from doing into checking — call it 15 minutes of review for every 45 minutes of old work. So you get roughly 115 hours back, not 172.5. At the loaded wage rate that's about £1,681 a year of cost removed. At the owner's notional rate it's £8,625 of capacity released.
Against a tool costing £20–£40 a seat per month (£240–£480 a year), the decision isn't close. But notice what made it work: a task that repeats, a rate you can put a number on, and a human still checking the output. Remove any one of those three and the maths stops working.
Enhancing decision-making
AI excels at processing large amounts of data quickly. It can analyse trends, predict outcomes, and provide insights that might otherwise take weeks to uncover. For business owners, this means better decision-making and less guesswork. But again, the final call still relies on human expertise and judgement.
The practical version of this is unglamorous. Instead of a management pack that arrives three weeks late and gets skimmed, you get a plain-English read on what moved and why, on the day the bank feed settles. That isn't clairvoyance. It's removing the lag between something happening and you knowing about it — and lag is where most bad decisions live. That compounding effect is the whole argument behind marginal gains in business.
Collaboration, not competition
Rather than viewing AI as a competitor, think of it as a teammate. Brilliant people paired with AI have access to tools that amplify their skills. Imagine a chef with a perfectly organised kitchen or an architect with an advanced modelling tool — it's about enabling creativity and expertise to shine.
The teams who get this right treat AI like a very fast, very literal junior. It never gets bored, never gets tired, and never tells you it's out of its depth. That last one is the catch, and it's why the next section exists.
Where it goes wrong
Five failure modes, in the order I see them most often.
- Confidently wrong answers. These tools produce fluent text whether or not they know the answer. Rule: nothing generated by AI is ever the last check on a number that goes to HMRC, a client, a lender or a board.
- Automating a broken process. If the process is wrong, AI runs it wrong, faster, and at scale. Fix the process, then automate it. Automating chaos just gets you chaos sooner.
- Data going somewhere you haven't checked. Under UK GDPR you stay responsible for client data you hand to a third-party tool. Before anyone pastes anything, know whether the provider trains on your inputs, where it's processed, and how to switch that off.
- Skill atrophy. If juniors never do the boring version, they never learn to smell when something's off. Keep some manual work in the training path deliberately.
- Tool-shopping before job-defining. Buying software and then hunting for a use is how firms end up paying for six subscriptions and forming one habit.
The UK reality: this is about to stop being optional
Making Tax Digital for Income Tax starts on 6 April 2026 for sole traders and landlords with qualifying income over £50,000. The threshold drops to £30,000 from 6 April 2027 and £20,000 from 6 April 2028 (GOV.UK, checked 28 July 2026). Whichever band you land in, the shape of the year changes: quarterly updates plus a final declaration, rather than one annual scramble.
That is a volume problem, and volume problems are what automation is for. A firm still rekeying bank statements by hand isn't facing a technology decision. It's facing a capacity one.
The human touch matters
No matter how advanced AI becomes, it can't replicate human empathy, creativity, and critical thinking. AI can assist, suggest, and predict, but it's the people who make the real impact — connecting with others, solving complex problems, and bringing ideas to life.
Nobody has ever renewed a contract because the reconciliation was fast. They renew because someone rang them before the problem got big, understood the business, and told them the truth.
Do this in the next seven days
A concrete exercise. Ninety minutes, once.
- List every recurring task you or your team did last week that took more than fifteen minutes. Be specific: "chase unpaid invoices", not "admin".
- Sort each one into three columns. A: has a right answer and a repeating pattern. B: needs judgement, but a first draft would help. C: needs trust, or delivers bad news.
- Automate column A. Draft column B. Never touch column C. That single rule will save you more than any tool comparison.
- Price the biggest item in column A the way I did above: hours a year × loaded hourly cost. If the answer is under £500, leave it alone for now.
- Pick one. Run it alongside the manual version for two weeks and compare outputs before you trust it.
AI is a powerful tool, but it's just that — a tool. It helps people unlock their potential, tackle challenges, and focus on what truly matters. By embracing AI, brilliant people aren't being replaced; they're finding new ways to excel. If you want a hand deciding which of your columns is which, that's much of what my advisory work involves, and it's the first thing I look at when I step in as a fractional COO.