Practical AI
A day in the life of an SME owner-manager in 2026: when AI helps, and when it has no business being there
People talk about AI in the future tense, like a wave about to break. I prefer to look at it in the present, in an ordinary owner-manager's day. Not "business transformation", not "strategic vision": the real day, with its emails, its calls, its evening quotes. Because that is where, in the gaps, AI genuinely helps today. And it is also where, at certain moments, it has no business being.
8.40am, the email that has been stuck for three days
There is always that awkward email you keep putting off: a firm but polite chaser to a slow-paying customer, a reply to a complaint, a quote to draft. You think about it in the shower, you never write it. It is the first genuine use, the one I recommend everyone start with: give an assistant the context, get a decent draft in two minutes, correct it, send it. No magic, just unblocking. The blank page, the one that costs days of delay, disappears.
11am, the site meeting write-up
The meeting is over, and a clean write-up needs to go to the team and the client. Before, that happened in the evening, or never. Dictate your notes as they come, ask for a clear write-up, read it back: five minutes instead of forty. Same thing for a quote quickly structured from three scribbled lines. Added together, these small bits of unblocked writing and summarising add up to a good half hour a day recovered, two and a half hours across the week. It does not transform the business. It gives you back your evenings.
The time an owner-manager gets back on average just from writing and summarising, once they have got into the habit. Not on one big spectacular task, on a scattering of small moments you had stopped even noticing.
1pm, the question you kept wondering about but never looked up
Lunch wolfed down, a technical question lingers: a point of regulation, contract wording, the difference between two statuses. Before, it got shelved "for when I have time", which means never. An assistant roughs it out for you in two minutes. With one caveat I repeat to every client: whatever it gives you, especially a figure or a reference, should always be checked at the source before you act on it. It clears the ground, it does not make the call.
3pm, the unhappy customer on the phone
Here, AI gets put away. An angry customer wants to feel someone on the other end, not a quick, well-turned reply that misses the point of their anger. This moment is entirely yours, and it must stay that way. The worst thing you could do to yourself is let a polite machine "handle" a complaint: you save three minutes and lose a customer who, until then, was merely annoyed. Presence cannot be delegated.
5pm, the decision that commits you
Whether to hire, whether to invest in a machine, whether to change supplier: AI can help you prepare these decisions, list out the arguments, make sure nothing is missed. But making the call, no. It will not carry the consequences, you will. I say it often: the machine takes the repetitive, you keep everything that carries a cost if it goes wrong. The day a provider tries to sell you the opposite, be wary.
Useful AI is not where it gets shown off. It is in the five minutes of writing you had stopped even noticing.
The trap of the "all-AI" day
An owner-manager once told me he wanted an assistant to draft every one of his customer replies, morning to evening. Within two weeks, a loyal customer told him his messages had gone "oddly smooth", impersonal. He had saved time and lost his voice. That is the real risk of putting AI everywhere: certain slow moments in the day, a handwritten note, a call just to check in, a coffee with a supplier, are not wasted time to be optimised away. That is often where the business actually happens. Automating them is a bad trade dressed up as a productivity gain.
That is an honest day: AI fits into four or five precise spots, it disappears in three others, and it is exactly that balance that makes it help rather than hinder. If you want to know exactly where your hours go before touching anything, start by counting, for a week, what you are really losing. And to place these uses from the simplest to the most advanced, the three levels of AI will stop you believing you need an over-engineered system to get started. The real first step often costs no more than the price of a subscription and an hour spent learning to use it well.