How small businesses are actually using AI right now
AI has settled into a handful of genuinely useful routines for small businesses. Here is what owners are actually doing with it and what is worth the cost.
AI has settled into small business life without the drama the press releases promised. Here is what owners are actually using it for, what is working, and what is not.
The conversations among small business owners have changed. A year ago they were asking whether they should pay attention to AI. Now they ask each other what they are actually doing with it.
What is worth the subscription fee. What turned out to be hype. The answer, in most cases, is that AI has settled into a handful of useful routines without transforming anything as dramatically as the press suggested.
That is not a failure story. It is just what adoption actually looks like. Small business AI has become a quiet workhorse rather than the revolutionary force it was marketed as.
Writing and drafting
The most common use, across almost every type of small business, is writing. Not replacing writers, but handling the parts of writing that owners have always found tedious. Drafting reply emails to difficult customers. Writing job adverts.
Producing a first version of a product description. Cleaning up a meeting summary. AI is fast at all of this. It does not get it right first time, but it gets you to a usable starting point far more quickly than a blank page.
The pattern is consistent. A builder in Edinburgh uses Claude to draft responses to complaint emails. A marketing consultant in Bristol runs every client deliverable through ChatGPT for a quick proofread before it goes out. Choosing between the main tools is covered in the ChatGPT vs Claude vs Gemini comparison if you are still deciding which to try.
A small accountancy firm in Manchester uses it to produce first drafts of client-facing letters, which a junior member of staff then checks and personalises. None of these involve AI working unsupervised. All of them save real time.
Social media and content
Social media has become another reliable application. Producing content for Instagram, LinkedIn, or Facebook is one of those tasks that feels quick but routinely takes longer than it should. AI tools can produce a week’s worth of caption drafts in ten minutes.
They tend toward a generic cheerfulness that needs toning down. But the bones of the content are usable, and editing is faster than originating. Small retailers, cafes, salons, and fitness businesses have taken to this particularly well.
Many do not have the budget for a social media manager but know they cannot ignore the channels. For them, small business AI has become the de facto content assistant.
Meetings and admin
Meeting notes are a growing use case that surprised a lot of people. Tools like Otter.ai and Notion AI can record a meeting, transcribe it, and produce a summary with action items.
The whole thing takes roughly the time it takes to make a cup of tea. For solo operators and small teams constantly moving between client calls, this kind of passive capture has become genuinely valuable.
The transcripts are imperfect, especially with strong accents or overlapping speakers. But the reduction in mental load is real. Knowing you do not have to remember every decision made in a call changes how much attention you can give to the conversation itself.
HR and administrative documents are an area where AI has earned its keep quietly. Writing a staff handbook or drafting a policy used to mean hiring someone or producing something you were not confident in.
AI does not replace proper legal advice. Any documents touching on contracts or redundancy should still be reviewed by a professional. For the operational layer of running a small team, it makes documents easier to get into roughly the right shape before an expert reviews them.
Finance and research
The accountancy angle is subtler than people expected. AI cannot file your tax return or replace your accountant. What it can do is help you understand what your accountant is saying. It can draft questions to send them.
It can organise records into a format that is easier to work with and summarise financial reports into plain language. For small business owners who find numbers intimidating, that translation layer has been surprisingly useful. Anyone wanting to go deeper will find the Investing Basics section a useful complement.
Research has also shifted. Before AI search tools became widely used, looking into a supplier or reading about an unfamiliar regulation meant a long afternoon of browser tabs. Checking competitor pricing was the same.
AI search compresses the initial research phase significantly. You get to the right questions faster, even if you still need a professional to answer them properly. The UK government’s guidance on AI adoption for SMEs is worth reading alongside practical experimentation.
What the costs actually look like
The cost picture is simpler than the marketing suggests. Most small businesses are paying between ten and thirty pounds a month for a capable AI tool. ChatGPT Plus, Claude Pro, and Gemini Advanced all sit in that range. For the hours saved on drafting, research, and admin, the return is not complicated to calculate.
The free tiers are worth knowing about too. They are capable enough for occasional use. But the paid tiers are meaningfully better for anything you want to rely on regularly. For most small businesses, the paid subscription pays for itself within the first week of genuine use.
What is not working yet
What is not working, or not working reliably, is AI that runs without human oversight. The idea that you can point AI at your customer service inbox and trust it to respond without review is possible in theory. In practice it is risky.
The outputs are confident in a way that does not match their reliability. A wrong answer delivered with authority is worse than a slow human response.
The small businesses getting the most out of AI are using it as a drafting layer, not a decision layer. That distinction matters. Understanding what can go wrong with AI agents helps explain why unsupervised AI use in customer-facing roles carries real risk.
What this means for you
If you have not started experimenting with small business AI tools, the clearest place to begin is your biggest time sink. If it is email, start there. If it is social media, start there.
Pick one tool, pay for one month, and apply it to one problem. The learning curve is not steep. Within a few days you will have a clear sense of whether it is earning its subscription.
The wider picture is straightforward. AI is not doing anything extraordinary for most small businesses right now. It is doing ordinary things faster. That turns out to be quite enough to justify paying serious attention to it.
One thing worth tracking is how quickly these tools improve. The AI tools available to small businesses today are already more capable than they were twelve months ago. The ones available in twelve months will likely be more capable still.
Getting comfortable with the current versions now means you will be better placed to use whatever comes next. The businesses that wait for the perfect moment to start will always keep waiting.