What AI is actually good at in a small business, and what it is not

There is a lot of noise about AI and very little of it is written for someone running a six-person company. So here is the split as we see it, from actually using these tools day to day.

Genuinely good at: the first draft of anything

Listing descriptions, follow-up emails, job post wording, the awkward reply to a difficult review. Not because the output is finished, but because editing something is faster than starting from nothing.

The win is not quality. It is that the task stops getting postponed.

Genuinely good at: turning a mess into a structure

Paste in rambling notes from a site visit and ask for a scope of work. Drop in twelve customer emails and ask what people keep complaining about. Give it a transcript and ask for the follow-up actions.

This is the most underrated use and almost nobody does it. The work is tedious, the output is checkable at a glance, and the stakes are low if it is slightly off.

Genuinely good at: making one thing into five

One filmed answer becomes a caption, a short post, an email paragraph and an FAQ entry. One blog post becomes a month of social. You are not creating new claims, just reformatting something you already said.

Not good at: anything where being wrong is expensive

Pricing a job. Quoting lead times. Interpreting a contract. Telling a patient what their symptoms mean.

These tools produce confident, fluent, plausible answers, and fluency is not accuracy. A wrong number delivered confidently is worse than no number, because nobody thinks to check it.

The danger is not that it gets things wrong. It is that it gets things wrong in a tone that sounds exactly like getting them right.

Not good at: sounding like you

Default AI writing has a texture people now recognise instantly, and in a local business that texture is actively harmful. Your advantage over a national competitor is that you sound like a person from here who has done this a hundred times. Generic polish throws that away.

Not good at: knowing what it does not know

It will not say "I am not sure." It will fill the gap. For anything load-bearing, you need a human who knows the subject checking the output, which means AI saves time on things you can verify quickly and saves nothing on things you cannot.

The practical test

Before using it for a task, ask: if this came back subtly wrong, would I notice?

If yes, use it freely. You are the safety net.

If no, do not use it there yet. Not because the technology cannot, but because you have no way of catching the one time in twenty it invents something.