AI for small business
Most advice on AI for small business is either breathless or dismissive. The reality is narrower and more useful: AI is very good at producing first drafts and summarising long text, and unreliable at anything requiring verified fact.
Knowing which side of that line a task sits on is the whole skill.
Tasks where it pays immediately
- Proposals and quotes. Draft the narrative; you set the price and the scope.
- Customer emails. Especially the ones you keep putting off.
- Summarising documents. Contracts, tenders, supplier terms.
- Job adverts and briefs.
- Marketing copy. Social posts, service descriptions, listing text.
- Spreadsheet problems. The formula that nearly works.
- Images. Social graphics and article images instead of stock.
Tasks where it costs you
- Anything with figures you have not checked. Prices, tax rates, deadlines, statistics.
- Legal or tax advice. It produces confident, plausible, occasionally wrong answers in exactly the areas where wrong is expensive.
- Customer-facing output sent unreviewed. The failure is visible and reflects on you, not the tool.
- Technical work you cannot evaluate.
- Anything requiring knowledge of your business it does not have.
Starting in a week without a project
- Day 1. Pick the writing task you most dislike. Have AI draft it. Edit and send.
- Day 2. Take a long document you have been avoiding. Summarise it. Check the summary against two sections.
- Day 3. Bring the spreadsheet problem you gave up on.
- Day 4. Draft one marketing piece — a service description or a social post.
- Day 5. Review the week. Keep what saved time; drop what did not.
What fails is the opposite approach — trying to redesign a process around AI before establishing where it actually helps.
Reviewing output properly
A workable rule with three tiers:
- Internal, low stakes — a light read is enough.
- Customer-facing — full read, and rewrite the opening and closing in your own words.
- Anything with a figure, date or commitment — verify against a source before it leaves.
Almost every AI failure that reaches a customer is a tier-three item reviewed as though it were tier one.
What it costs and whether it is worth it
Judge it against the time it returns. A business producing four proposals a month, spending two hours each, that gets to a usable draft in twenty minutes, recovers most of a working day monthly. That is a straightforward calculation.
If you write rarely and read few long documents, the honest answer is that a free tool is enough and you should not pay for more.
Frequently asked questions
What is the best use of AI for a small business?
First drafts and summarising. Proposals, customer emails, marketing copy, and reducing long contracts or tenders to their requirements. Those return time immediately.
Where does AI cost small businesses money?
Unverified figures, legal or tax questions, customer-facing output sent unreviewed, and technical work you cannot evaluate. Confident wrong answers are most likely exactly where being wrong is expensive.
How should I start using AI in my business?
One task at a time over a week — the writing job you dislike, a document you have been avoiding, the spreadsheet problem you gave up on. Keep what saves time. Do not redesign a process around it first.
How much should I review AI output?
Lightly for internal low-stakes work, fully for anything customer-facing with the opening and closing rewritten in your own words, and verify against a source for anything containing a figure, date or commitment.
Is AI worth paying for in a small business?
Judge it on time returned. Four proposals a month at two hours each, reduced to twenty minutes of drafting plus editing, recovers most of a working day. If you write rarely, a free tool is enough.
Will AI replace staff in a small business?
It changes what a person spends their day on rather than removing the person. It drafts and summarises; deciding, checking and dealing with customers remains human work.