Surveys of small business AI adoption keep surfacing the same number: roughly three quarters of the firms not using AI say they see no applicable use case for their business. Not too expensive. Not too risky. Simply nothing to point it at.
That is worth taking seriously rather than dismissing, because it is often a reasonable conclusion drawn from bad examples. The same research finds the leading barriers are lack of understanding and no in-house expertise — ahead of privacy and ahead of unclear return. The gap is not appetite or budget. It is knowing where to look.
Why “no use case” is usually a search problem
The public examples are chatbots, image generation, and writing marketing copy. If you run a plumbing firm, a dental practice, or a six-unit rental portfolio, none of those describe your Tuesday. Concluding it is not for you is a rational response to the examples you were given.
The candidates in a small business are almost always duller and further from the front of the business than the marketing suggests.
Where they actually hide
Three patterns account for most of what we find worth doing at this size:
- Turning messy input into structured data. Invoices in fifteen layouts, emailed job requests, handwritten notes photographed on site. This is unglamorous, it is genuinely hard by traditional means, and it is where the technology is strongest.
- Reading things nobody has time to read. A year of maintenance notes, support emails, or inspection reports contains patterns that matter and that nobody will ever sit down and summarise.
- First-pass drafting where a human still signs. Not publishing. Drafting — the tenant notice, the quote follow-up, the summary — with a person reviewing before it goes anywhere.
Notice what these have in common: none of them is customer-facing, and none removes a human from the decision. That is not a coincidence. At this size, the reliable value is in preparation, not in autonomy.
Where it depends on you
Whether any of that clears the bar in your business depends on volume and on what an error costs. The same research finds that firms which do adopt overwhelmingly stall at shallow use — a large majority report having no actual strategy behind it, and only a small minority get past experimenting. Buying a tool is not the hard part. Knowing which of your processes is worth pointing it at, and what happens when it is wrong, is.
There is also an honest answer we give reasonably often: your volume is too low for this to pay, and the same effort spent on a form or a template gets you most of the benefit with none of the maintenance.
How we would walk you through it
We start from your work, not from a capability list. An hour going through what actually arrives in your business each week and in what shape — the emails, the documents, the photographs, the notes — is usually enough to tell whether there is anything here worth pursuing.
You get back a short written view: the candidates if there are any, what each would cost to stand up and to run, what it would break if it got something wrong, and a clear statement where the answer is “not yet”. We would rather be the firm that told you to wait.
That assessment sits in AI consulting. When the honest answer turns out to be a rule rather than a model — which it often is — it moves to workflow automation, which is cheaper and more predictable.
Describe one thing that arrives in your business every week in an inconvenient format. That single example is usually enough to say whether there is anything here for you.
A question worth answering
What arrives in your business every week that a person has to read and retype somewhere else? That, not the chatbot, is where the question starts.
