A business with fewer than ten people is ready for AI when the way it uses AI lives in the business and not in one person's head. Pick one repeat job, give it one owner, and write down how it is done, so the work carries on if that person is away for two weeks. The Small Room Test below checks exactly that.
Most advice on AI for small business is written for a firm of fifty, with a steering group, a training plan and someone whose job title includes the word "AI". A firm of six has you, a bookkeeper, and three people who each do three jobs. This article is for that end of the scale.
How many very small businesses use AI?
Fewer than larger ones, in every official survey I opened this week. The surveys count differently, so read each number on its own and do not set them side by side.
In the UK, the Office for National Statistics published wave 165 of its Business Insights and Conditions Survey on October 8, 2026. Fieldwork ran from September 21 to October 4, 2026, with 9,933 responses from a sample of 38,592 businesses. Some 30% of businesses reported using at least one type of AI technology. Among businesses with fewer than 10 employees the figure was 29%, against 38% for those with 10 or more. The ONS labels these official statistics in development, drawn from a voluntary survey.
In the US, the Census Bureau's Business Trends and Outlook Survey story of May 26, 2026 covers data from December 14, 2025 to May 3, 2026 from a nationally representative sample. It put AI use across all US businesses at 19.8% as of May 3. Less than 20% of firms with four or fewer employees reported using AI. Use rose among firms with at least 20 employees over the period, but not significantly among firms with fewer than 20. The Census Bureau changed the wording of its AI question in November 2025, so older releases do not compare cleanly.
The UK government's AI Adoption Research, published on January 28, 2026, interviewed 3,500 UK private-sector businesses with five or more employees by telephone between February 12 and May 2, 2025. It reported 16% of businesses using AI: 14% of micro businesses, 23% of mid-sized and 36% of large. That is spring 2025, so treat it as a snapshot of the gap, not today's level. Its most useful finding for this article is the reason given. Across all businesses, not only the smallest, 71% named a lack of identified need as a barrier, and among businesses with no plans to adopt AI, 81% did.
Here is my reading, and it is mine, not the surveys'. The gap in numbers is real but it is not the story. The story is that the owner of a six-person firm often cannot see a job for AI in a week that is already full. Nobody is telling them it is dangerous. Nobody has shown them where it fits.
Why does AI work differently in a firm of fewer than ten?
Because four things that larger firms take for granted are missing, and each changes what a sensible first move looks like. This is how I read the pattern, not a finding from the surveys above.
- One person is several departments. The owner who writes the quotes also chases the invoices. AI in that week is a personal habit before it is a company program.
- Knowledge lives in a head. Which prompt works, which past quote to paste in, which details must be checked against the site notes: usually one person knows, and nobody has written it down.
- There is no second reader. In a large firm a mistake passes through several hands. In a small one it can go from the person who made it to the customer in a single email.
- Nobody has time to be "the AI person". The champion model that works in a firm of 200 asks someone to take on a role on top of a full job. At six people that is the owner, who is already the busiest person in the building.
What is the Small Room Test?
The Small Room Test is one question for any business with fewer than ten people: if the person who knows how you use AI were away for two weeks, would the work carry on the same way? If not, the practice lives in a head and not in the business, and the first job is to write it down before you add anything new.
I offer it as my suggested way of thinking, not as research. The name comes from the size of the room. Everyone in it can hear everyone else, so you do not need a committee or a policy manual to share what you know. You need one page that a second person can follow, and a second person who has actually followed it.
A job passes when three things are true. Someone other than the usual user can name what goes in, what comes out and what gets checked. They can run it, once, in under an hour, using only what is written. And the job still gets done the same way whichever of the two is doing it.
What does a failed Small Room Test look like?
Take a hypothetical seven-person landscaping and garden maintenance firm. The owner writes every customer quote in the evening with the help of an AI assistant such as ChatGPT or Microsoft Copilot. The quotes are good, faster than they used to be, and customers reply. Nobody else in the firm has seen how it is done.
Then the owner takes a fortnight's holiday in August. Within three days quotes stop going out. The office manager can write a quote, but not the owner's way. She does not know which past quotes to paste in as examples, which figures the owner always checks against the site visit notes, or that customer addresses are never typed into the assistant. The owner comes back to a pile of unanswered enquiries.
That is a failed Small Room Test, and the fix is small. The owner writes one page with four parts: what the job is, what goes in, what gets checked before it is sent, and what must never go in. The office manager runs the job once with the owner beside her. They agree a date four weeks away to decide whether to keep the habit, change it or drop it. There is no policy and no new tool. Only after that does the firm look for a second job to try.
Which of the Seven Dimensions come first when there are fewer than ten of you?
The Seven Dimensions of AI Readiness are the seven things an organization must have in place before its use of AI can be called ready rather than merely widespread: purpose, permission, ownership, capability, protection, foundations and evidence. Your readiness is your lowest score across the seven, not your average. The full scoring is in How Ready Is Your Business for AI?.
At this size, purpose, permission and ownership are one conversation among the same handful of people, and you can have it this week. The Small Room Test then exposes the next two: capability, because it sits in one head, and protection, because what must never go into a chat window is rarely said aloud. Foundations means the few documents the job relies on, and evidence means one number before and after. The Before Number Rule covers that last piece.
What I Tell Business Owners
I tell owners of very small firms to resist copying what larger firms do. A firm of 200 needs a policy, a steering group and a training plan. A firm of six needs a weekly habit, one written page per job, and an agreed answer to what customer details never go into a chat window.
I have delivered more than 500 keynotes in 30 countries, and the principle behind my advice has not moved: outsource the doing, not the thinking. In a small room that means the owner stops being the only person who can do the AI-assisted job, without stopping being the person who decides what it is for. If you are the one using AI most, AI for Business Owners' Own Work shows how to audit your week. For the question of rules, Does a Small Business Need an AI Policy? explains why a short written rule can be enough. And if your staff are already using their own accounts, start with shadow AI.
What should you do this week?
You can run the whole check in five short sessions, and none of them needs a project or a consultant.
- Monday: write down every way AI is already used in the business, including on people's own phones and accounts.
- Tuesday: pick the one use that happens every week and matters to a customer.
- Wednesday: ask the Small Room Test question about it. Who else could do this exactly as it is done now?
- Thursday: write the page: what goes in, what comes out, what gets checked, and what never goes in.
- Friday: have a second person run it once while you watch. Diary a date four weeks ahead to keep, change or drop it.
If you outgrow ten people, the problem changes from one head to many roles. Who Is Left Behind on AI in Your Team? picks up from there, and Is Your Business Data Ready for AI? helps with the documents your page will rely on.
Where to go from here
If you want to know where your business actually stands before you spend another month experimenting, take the Business AI Readiness Scorecard, which takes about five minutes. If you are choosing a speaker for a chamber of commerce, trade association or industry event, my business keynotes are built for audiences where most of the room runs a very small firm. And if your team is using AI but nobody has shown them how to do it well, that is what my AI training for teams is built for.
Sources and further reading
- Office for National Statistics, Business insights and impact on the UK economy: 8 October 2026, wave 165, fieldwork September 21 to October 4, 2026, UK, published October 8, 2026.
- U.S. Census Bureau, Large Firms With at Least 20 Employees Biggest AI Users, Business Trends and Outlook Survey, data December 14, 2025 to May 3, 2026, United States, published May 26, 2026.
- Department for Business, Innovation, Science and Trade and Department for Science, Innovation and Technology, AI Adoption Research, UK, published January 28, 2026 (fieldwork February 12 to May 2, 2025).
- Skills England, AI Skills for the UK Workforce, published October 29, 2025, includes an Employer AI Adoption Checklist for self-assessment.
Dan Fitzpatrick, The AI Educator, is a Forbes contributor, international keynote speaker and bestselling author who has trained more than 150,000 people across 30+ countries. About Dan.
