Grow the book without hiring another pair of hands

By Jason Sibley

The problem

You've got more work than the last quarter, and probably more than the one before that. But another salary, with recruitment costs, onboarding time, National Insurance, and the risk of getting the hire wrong, isn't something you can commit to yet. So the work either gets squeezed into evenings and weekends, or it gets turned away.

This is the everyday reality for the vast majority of UK businesses. At the start of 2025 there were an estimated 5.69 million private sector businesses in the UK, and 5.64 million of them, 99.2%, were small businesses with fewer than 50 employees. Three-quarters of all UK private sector businesses have no employees beyond the owner at all. If you're trying to grow without adding a pair of hands, you are not the exception. You are most of the UK economy.

What people get wrong

They think "using AI" means picking one big tool and hoping it fixes everything. It doesn't work like that in practice. Office for National Statistics data from its Business Insights and Conditions Survey shows that even among UK businesses with 10 or more employees that have adopted AI, the average number of AI technologies in use per business has only crept up from around 1.4 to around 1.6 since late 2023. Adoption has been "relatively shallow". People pick up one tool for one job, not a company-wide overhaul. That's not failure. That's the realistic, sustainable pattern.

They assume everyone else has already sorted this out. They haven't. ONS figures show that 28% of UK businesses with 0–9 employees reported using at least one AI technology by June 2026, compared with 49% of businesses with 250 or more staff. Separate research from Enterprise Nation's Tech Hub, a partnership with Google, Sage, Dell Technologies and Square, surveying 1,320 UK micro and small businesses, found only 21% use AI regularly, and just 6% have it embedded into daily work. Only 34% of one-person businesses use AI at all. If you haven't got round to it yet, you're in the majority, not behind some curve everyone else has already climbed.

They wait for confidence before they start. Enterprise Nation's research also found real barriers: 53% of small businesses cite cost, 46% cite a lack of skills, and 37% cite a lack of time as reasons they haven't adopted AI. Those are legitimate constraints, but waiting until you feel fully confident before trying anything tends to mean never starting, because the confidence comes from doing it, not from reading about it first.

They pick the flashiest use case instead of the most repetitive one. The most common reported use of AI among UK businesses, at every size band, is improving business operations, not some dramatic new product or service, but making the existing grind faster. Close to 60% of adopting businesses report using AI for this. That's the boring, valuable target: the task you already do every week that doesn't need your judgement, just your time.

They let one bad first attempt put them off for good. A clumsy first draft from an AI tool, or a summary that missed the point, is often treated as proof the whole idea doesn't work for a business like theirs. In practice this is closer to trying a new supplier once and writing them off after a single late delivery. The more useful response is a better brief or a different tool, not abandoning the category entirely. Enterprise Nation's own findings note that among early adopters, 65% said their use of AI increased over the past year, which suggests the pattern is iteration, not one perfect result on the first try.

What this looks like in an ordinary week

Take a small business with three people: the owner, someone doing a mix of admin and customer service, and one person out doing the actual client-facing work. Growth has meant more enquiries, more invoices, and more scheduling, but not enough extra volume to justify a fourth salary.

The realistic version of "absorbing the repeatable third" here isn't a grand initiative. It's the admin person using an AI assistant to draft a first-pass reply to routine enquiries before personalising and sending them, the owner using the same kind of tool to turn rough meeting notes into a clean list of action points before the next catch-up, and invoice reminders running automatically through whatever accounting software they already use, rather than someone manually checking who's overdue each week. None of this is dramatic. All of it, done consistently over a couple of months, adds up to genuine extra capacity, the kind that used to require a fourth person, or a lot of unpaid evening hours.

A practical frame: absorb the repeatable third

Not all of your work is equally suited to AI help, and pretending otherwise leads to disappointment. A useful way to sort it:

The repeatable third. Bookkeeping admin, drafting first-pass replies to routine enquiries, writing a first draft of a proposal you've written variations of a dozen times before, summarising a meeting into action points, chasing an overdue invoice. This is where AI buys back hours, because the task doesn't change much from one instance to the next and the judgement required is low.

The judgement middle. Deciding which lead to prioritise, how to handle an awkward client conversation, pricing a custom job. AI can draft options here, summarise context, and flag what you might be missing. The decision still needs you.

The relationship top. The actual client call, the site visit, the negotiation, the thing your customers are paying for because it's you. This doesn't get automated, and trying to hand it to AI is usually where "AI feels fake" complaints come from.

The businesses making real headway aren't trying to automate everything. They're identifying the repeatable third, often just two or three specific weekly tasks, and building a habit of letting AI take the first pass, while judgement and relationships stay firmly with a person.

Be honest about the boundary between the repeatable third and the judgement middle, because it moves depending on your business. A first-draft reply to "what are your opening hours" sits firmly in the repeatable third for almost anyone. A first-draft reply to a complaint about a botched job sits in the judgement middle even though it looks, on the surface, like the same kind of task, because the cost of getting the tone wrong is much higher, and because it usually needs specific knowledge about what actually happened. The categories aren't about task type; they're about how much genuine judgement each instance actually requires.

The one decision to make this week

Don't decide "should we use AI." Decide: which single repeatable task, done at least weekly, will we hand a first draft to AI for over the next two weeks?

Pick something low-risk and high-frequency: invoice chasing, meeting notes, first-draft review replies, or supplier research. Not something client-facing on day one, and not something irreversible. Give it two weeks, keep a human reviewing every output, and then decide whether it earned a permanent place in how you work. That's a much smaller, much saner decision than "should we change how the business works", and it's the one that Enterprise Nation and ONS data both suggest is how the businesses currently pulling ahead actually got started: one task, done consistently, before moving to the next.

A soft word before you go

If you want a second pair of eyes on which task to start with, My AI Apprentice runs SMB AI sessions built around exactly this kind of decision, and a lighter AI sales & marketing desk if the repeatable task sits in that part of the business. We also do free website reviews if growth is being held back by a site that isn't pulling its weight. Entirely optional. The frame above still works without us.