Where should a small business actually start with AI?

The most common question I get about AI isn’t “how does it work?” It’s “where would it actually help me?” That’s the right question. And the honest answer is that it almost never starts with the tool.

Most weeks someone sends me a list of AI tools they’ve bookmarked and asks which one to buy. I get it. Open any feed and there are ten more by lunch. But picking the tool first is like buying a ladder before you know which wall needs painting. The tool is the easy part. Knowing where it earns its keep in your business is the whole job.

So here is how I go about it with a client, and how you can do a rough version yourself before you spend anything.

Start with your week, not the technology

Forget AI for a minute. Think about a normal week and ask where the time actually goes. Not the interesting work, the other stuff. The tasks you’d hand to someone in a heartbeat if you could.

Three kinds of work are worth a second look:

• Things you repeat. The same reply to the same customer question. The same summary written a slightly different way each Friday.
• Things that are mostly text. Quotes, emails, reports, notes, product descriptions. Anything where you stare at a blank page and fill it.
• Things that are slow because they’re fiddly, not because they’re hard. Reading a long thread to find one decision. Pulling numbers out of five emails into one place.

Write down five of those. That list is worth more than any tool comparison, because it’s specific to you. AI is good at exactly this kind of work: repetitive, text-heavy, and annoying rather than difficult.

A few examples that are dull on purpose

When people picture AI in a business they tend to picture something dramatic. In practice the wins are usually boring, and boring is good, because boring is what happens every single day.

Customer questions. A lot of what lands in your inbox is the same ten questions. AI can draft the answer in your tone, pull the right detail from your own documents, and leave you to press send. You stay in the loop, you just stop typing the same thing over and over.

First drafts. A quote, a follow-up email, a job ad, a product description. Getting from nothing to a rough draft is the part that eats time. Going from a decent draft to a good final version takes minutes, and you keep the last word.

Summaries. A long meeting, a fifteen-email thread, a document you don’t have time to read properly. A summary in thirty seconds beats reading the whole thing in twenty minutes, and you can always dig into the bits that matter.

None of these replace you. They take the runway work off your plate so you get to the part only you can do.

Do the maths before you commit

Here’s the test I use, and you can use it too. Take one task from your list and put rough numbers on it.

Say answering routine customer emails takes you an hour a day. That’s five hours a week. If AI drafts them and you just review and send, you might get that down to an hour total across the week. Four hours back, every week, on one task. That’s the kind of number that tells you where to start, and it beats any feature list.

If the maths is thin, move to the next task. You’re not looking for the cleverest use of AI. You’re looking for the one that quietly gives you the most time back.

Where it’s worth being careful

Two honest cautions, because I’d rather you hear them from me.

Don’t paste sensitive customer data into random free tools. Where your data goes, and whether it’s used to train someone else’s model, matters. For most of my clients that means choosing tools that keep data inside the EU and don’t train on your inputs. It’s a question worth asking a vendor before anything else, and I write more about it in a separate piece.

And check the output at the start. AI is confident even when it’s wrong. Give it a week of your review and you’ll quickly learn what it’s reliable at and what it isn’t. After that you can loosen the reins on the safe stuff.

The honest version

You don’t need an AI strategy. You need one task where it clearly saves you time, done well enough that you trust it. Then the next one. The time you save grows task by task, and you learn as you go instead of betting big on day one.

If you’d rather not work out the “where” alone, that’s the part I help with most. We look at your week, find the spots where AI actually pays off, and then I build the thing. Advice, implementation, or both.

Want to find out where it’d help in your business? Get in touch and we’ll have a look together. You can read more about how I work with AI here.