Dan·25 October 2026·6 min readUse AI to Grow Your Business, Not to Cut Your Team
Every few weeks someone asks me the same question, usually a bit quietly: "So how many people can AI replace?"
I understand why. Most of the headlines are about job losses, and if you run a small business, wages are your biggest bill. But I think it is the wrong question, and answering it is the least imaginative thing you can do with the most capable tool most of us have ever been handed.
Cutting a job with AI saves you one salary. Using AI to take on the work you used to turn down is worth far more.
What AI actually did in my business
I have been building with AI every day for over a year. Not one of the wins I would point to is a job removed. Every one of them is work I simply could not have taken on before.
A product explainer that an agency would quote thousands for, and weeks of turnaround, took an afternoon. I did not sack an animator. I never had one. Now I can offer films to clients I would have had to turn away.
A data job that would have been days of copy and paste finished in two minutes. Nobody lost their role. The person who would have done it spent those days on work a customer actually notices.
I now run a club's WordPress site by talking to it. Fixtures, news and pages that sat on a volunteer's to do list for weeks happen the same day.
In each case the team stayed the same size. What changed is how much it could do.
What the research says
This is not just my experience. The best real world evidence points the same way.
The study worth knowing about
Researchers at Stanford and MIT (Brynjolfsson, Li and Raymond, published by the US National Bureau of Economic Research) followed more than 5,000 customer support agents given an AI assistant. Productivity rose by about 14 per cent on average, and by 34 per cent for the newest and least experienced staff. Customers were happier and staff were more likely to stay. The AI did not replace the agents. It spread the know how of the best ones to everyone else.
UK small businesses are catching on fast. Simply Business's 2026 SME Insights Report found 47 per cent of small business owners now use AI tools, more than double the 22 per cent a year earlier. A Novuna Business Finance survey found AI use was far more common among small firms that were growing significantly (59 per cent) than among those that were shrinking (23 per cent). That is a correlation, not proof, but it is a telling one.
And plenty of businesses are still on the sidelines. Enterprise Nation found only 21 per cent of small businesses use AI regularly, and just 6 per cent have embedded it in daily work. That gap is the opportunity.
Cutting versus growing
- Same output, one fewer person
- The know how walks out of the door
- Everyone left is nervous and quietly resists the tools
- Customers see no difference, or a worse one
- A one off saving, then you are back where you started
- Same team, much more output
- Experienced staff teach the AI and check its work
- People adopt the tools because the tools make them better
- New services you could never staff before
- Growth that compounds every year you keep at it
The cutting route has a hidden cost too. The people who understand your customers, your quirks and your mistakes are exactly the people who make AI useful. In every project I have done, the big jumps in quality came from someone who knew the job saying "that is not right". Remove them and you are left with a fast tool and nobody to tell it when it is wrong.
Where this argument has limits
I would be selling you something if I said AI never replaces work. It does, and pretending otherwise does nobody any favours.
How to point AI at growth
Not the work you do. The jobs you say no to, the services you never offer, the follow ups that never happen. That list is where AI earns its keep.
They know what good looks like, so they get the most out of it and catch its mistakes. Then let them show everyone else.
Let AI do the copying, drafting and checking. Keep humans on the decisions, the relationships and anything a customer will read.
Track jobs completed, response times, new services launched. If those are climbing with the same team, it is working.
If you want to see the more technical side of how we keep AI reliable, read let AI do the hard work, use determinism for consistency.