Articles

AI for a small business: 7 things you can do today

No strategy project, no new software. Seven places where AI genuinely saves a small company time — and one where it does not.

2 min readBeginnerAI in business
AI for a small business: 7 things you can do today

Most writing about AI for business opens with the word "transformation". This is not that article. Here are seven things you can do this week without anybody implementing anything.

1. First drafts of quotes#

You keep the pricing logic in your head and you have ten similar quotes from last month. Give the model three old quotes as examples plus the new customer's brief — you get a draft to correct rather than one to write.

2. The emails you keep postponing#

Those three awkward replies sitting in the inbox for a third day. Describe the situation in two sentences and have it drafted. You edit. Two minutes instead of ten.

3. Summaries of documents#

A long contract, a new regulation, a ten-page tender document. Have it summarised with the deadlines and obligations listed. Verify it — but you are reading three pages instead of reading ten between the lines.

4. Draft replies to customer enquiries#

Not automatic replies. Drafts a human reviews. The difference matters: an auto-reply damages trust, a draft saves time.

5. Cleaning up data#

The messy spreadsheet where names are in three formats and half the addresses are missing a postcode. This is exactly the work nobody wants and AI does quickly.

6. Writing down procedures#

You know how something works. Talk it through to the model and have it write the instructions for a new hire. The first version takes minutes.

7. A first pass at competitors and market#

Not for making the decision — for starting it. You get a structure and the questions worth researching properly.

And one thing AI will not do#

It will not replace a system you do not have. If quotes, invoices and customer records are spread across three spreadsheets and an inbox, AI does not fix that — it just rewrites the mess faster.

First you need one place where the data lives. Only then does AI start to multiply anything.

Related articles