Probably not yet. Most small businesses do not need to hire an AI lead in their first year with AI. Appoint an AI steward from inside instead: someone the team already asks for help, with about two hours a week, a direct line to the owner and three duties. Hire, or buy outside help, only when that work outgrows the time or AI starts acting on your systems.
Sources re-read November 18, 2026. The role description and the decision table below are free to copy.
That answer surprises owners who have just watched a competitor post a job ad with "Head of AI" at the top. It should not. The official figures show that almost nobody your size is hiring for AI, and the firms getting somewhere are growing the skill inside.
Do small businesses actually hire AI specialists?
Very few do: the evidence says small firms build AI capability by training the people they already have, not by recruiting.
The clearest picture comes from the UK's Office for National Statistics. Its July 20, 2026 article on AI in UK businesses, drawn from the Business Insights and Conditions Survey, found that recruiting new staff with AI-related skills "rises from around 2% among the smallest businesses to 10% among those with 250 or more employees." Among businesses with 10 or more employees that named a lack of AI expertise as a barrier, around 62% said they were training or retraining existing staff instead. Training has not caught up with use, though: only 11% of businesses with 10 or more employees reported that more than half of their workforce has received AI-related training. The survey behind those figures ran from June 5 to June 28, 2026, and it found AI in small firms as well as large ones: 28% of businesses with 0 to 9 employees reported using at least one AI technology, against 49% of those with 250 or more.
The British Chambers of Commerce tells the same story from the other end. Its March 18, 2026 research with Atos and the University of Essex found 54% of UK firms actively using AI, up from 35% in 2025. Around 94% of the firms it surveyed were SMBs (SMEs in the UK and Europe), and 95% of the SMBs using AI said it had had no impact on the size of their workforce over the past year. AI arrived in these businesses without a single new hire to carry it.
The American picture matches. A Goldman Sachs 10,000 Small Businesses survey published March 17, 2026, of 1,256 US business owners in the program, surveyed from January 27 to February 4, 2026, found 76% currently using AI. But in the release's words, "Only 14% say AI is fully embedded in their core operations," and 73% said they would benefit from additional access to training and implementation resources.
Read those three sources together and a pattern appears. Use is wide. Depth is thin. And what owners say they lack is not a new employee with "AI" in the job title. It is someone in the building who makes the tools part of how the work gets done.
What does an AI lead actually do in a 20-person business?
In a small business the AI lead's job is not building AI; it is keeping three things in order so everyone else can use it well.
Picture the work that needs doing. Someone has to know which tools people are allowed to use, and which jobs those tools are used for. Someone has to hold the rules: what never goes into a chatbot, and what a person must check before anything leaves the building. And someone has to notice whether any of this is paying off, so the owner can decide what to do next.
None of that needs a computer science degree. It needs a person who is close to the daily work, trusted by colleagues and willing to say "I don't know, let me find out." A large company might need engineers to build models, connect data and run infrastructure. A 20-person firm buying Microsoft Copilot, ChatGPT or Gemini licenses does not. It needs a keeper, not a builder.
That is why I use a different word for the role.
What is the Steward Rule?
The Steward Rule is what I suggest a small business does before it hires anyone to lead AI: appoint an AI steward from inside first. Choose someone the team already asks for help, close to the daily work, and give them protected time, a direct line to the owner and three duties: keep the list of approved tools and the jobs they are used for, keep the rules on what never goes in and what a person checks, and keep the evidence of what AI is saving. Hire from outside only when the steward's work no longer fits the time you gave it.
Each part of that definition does a job.
Someone the team already asks. Every business has this person. They are the one colleagues lean over to when a prompt goes sideways. They are rarely the most senior and not always the most technical. What they have is the one thing a new hire takes months to earn: people already bring them their questions.
Protected time. A steward with no time is a volunteer, and volunteers drift back to the work they are measured on. Two hours a week is a sensible start for a business of 10 to 50 people, and something else must come off their plate to make room. Write it into their role.
A direct line to the owner. The steward keeps things in order. The owner still decides. The Readiness Test is three questions an organization must be able to answer consistently, from the top to the front line, before it can call itself AI ready: What is AI for here? What may I do with it? Who decides when it goes wrong? The steward can draft answers to the first two. The third stays with leadership, however good the steward is.
The list. One page naming the approved tools, the jobs each is used for and the person who uses it. If you pointed AI at one job first, as I suggest in how a small business should start using AI, the list begins with that job and grows one line at a time.
The rules. The steward keeps your AI policy current and understood. If you do not have one yet, the One-Page Five gives the questions it must answer. In a regulated practice the rules already have a professional body behind them; the AI policy for an accountancy practice and the AI policy for a small law firm show what the steward would be keeping.
The evidence. Once a month, the steward tells the owner what AI saved, what it got wrong and what people are asking for next. This is the duty most often skipped, and the one that turns a hunch into a decision. For a simple way to keep that evidence, one job at a time, see the Before Number Rule.
Take a hypothetical 22-person heating and plumbing firm. The office manager already fields every "how do I get it to do this?" question. The owner makes her the AI steward, takes supplier invoice chasing off her desk to free two hours a week, and books a 30-minute call on the first Monday of each month. By the third call she can say which jobs AI is doing, which rule nearly got broken and where the engineers want help next. The owner can then decide with evidence, not with whatever he last read.
What goes in an AI steward's role description?
Three duties, the hours, what comes off their desk, and who checks. A steward without a written role is a favor, and favors get dropped in a busy month. Here is one page to copy. Fill in the square brackets and give it to the person you have chosen.
AI steward: role description for [Business]
Who: [Name], alongside their role as [job title].
Time: [two] hours a week, protected. To make room, [task] comes off their desk.
Reports to: [Owner], in a 30-minute call on the first [Monday] of each month.
Duty 1. Keep the list. One page naming each approved AI tool, the jobs it is used for and who uses it. Nothing joins the list until [Owner] agrees.
Duty 2. Keep the rules. Keep our AI policy current, hand it to every new starter on their first day, and answer "am I allowed to?" questions or take them to [Owner].
Duty 3. Keep the evidence. Bring three things to the monthly call: what AI saved, what it got wrong or nearly got wrong, and what people are asking for next.
What the steward does not decide: spending, any new tool that touches customer or staff information, and what happens when something goes wrong. Those stay with [Owner].
Who checks: [Owner] reviews the list and this page every [six] months. If the work has taken more than the hours above for three months running, we look at the role again.
The hours and the review period are my suggestions for a business of 10 to 50 people. Change them to fit. Keep the three duties and the line about what the steward does not decide.
Should you appoint a steward, buy outside help, hire or wait?
Appoint a steward if you have more than a handful of people using AI tools; the other three options are for the businesses either side of that. The table sets the four choices against the size of the business and the amount of AI work. The bands are my working suggestions, not official thresholds, and the ONS figures above do not set any.
| Your situation | What to do | What it takes | Move on when |
|---|---|---|---|
| Fewer than 10 people, one or two AI tools, nobody asking for more | Wait on the role. The owner keeps the list and the rules | An hour a month of the owner's time | Staff start using tools you did not choose |
| 10 to 50 people, off-the-shelf tools in daily use | Appoint a steward from inside | About two hours a week of one person's time, with something taken off their desk | The work runs over the time for three months in a row |
| A one-off job the steward cannot do: connecting AI to your accounts system, a policy for a regulated activity, training the whole team | Buy outside help for that job, with an end date | A fixed fee or a day rate. The steward still owns the list | The same outside job comes back every month |
| The steward is at a day a week or more, agents are acting on your systems, or AI is used in decisions about people | Hire, or contract a part-time specialist who reports to the owner and works to the steward's list | A salary or a retainer, and a job description written from what the steward actually does | You are there |
Verdict: most businesses with fewer than 50 people belong in the second row, and none of them should skip it on the way to the fourth.
What about a fractional AI officer?
A fractional AI officer is a specialist you pay for a few days a month, and it is the third or fourth row of the table, never the first. Several of the pages that rank for this question in November 2026 are written by consultants who sell exactly that, so read them knowing it. One of them, Lilach Bullock, says herself that most small businesses with fewer than 20 people do not need one yet. I agree, and I would add a condition for everyone else: bring nobody in from outside until you have a steward and a list for them to work to. An outside specialist with no list spends the first paid days finding out what your office manager already knows.
The Mistake I See Most Often
The mistake I see most often is handing AI to whoever seems good with computers, with no time and no authority, and calling that a strategy.
It usually goes like this. The owner asks the youngest person in the office, or the one who set up the Wi-Fi, to "look after the AI stuff." Nobody takes anything off their desk. Nobody tells the rest of the team that this person now holds the rules. Six months later the owner wonders why nothing has changed, and starts looking at job ads.
The second version of the mistake is the job ad itself: hiring a title before you know the work. An outside hire arrives into a business with no list, no rules and no evidence, and spends months discovering what a steward would already know. Worse, it tells everyone else that AI belongs to someone else.
The businesses that do well treat AI capability as something the team grows. One business I have worked with shows the other way: theidol.com. In April 2023 I spoke about AI in marketing at a business event in Penrith. A year later their marketing team was using AI for content writing, and the managing director asked me back to show teams across the business what was possible. The spark came from outside. The capability grew inside, in the people doing the work.
When should a small business hire an AI lead?
Hire when the steward's work outgrows the time you gave it, when AI starts acting on your systems, or when your use touches decisions that carry legal weight.
Three signals tell you the moment has come.
The work outgrows the time for three months running. When the steward's two hours become a day a week, month after month, you have evidence of demand. Now you can write a job description from what the steward actually does, not from someone else's ad.
AI starts connecting to your systems or acting on its own. Chatbots that draft text need a steward. Agents that log into your accounting software, send emails or update customer records need someone who understands access, permissions and what happens when it goes wrong. Before you let any agent act for you, run it through the Errand Test. If you plan several such errands, technical skill is now a genuine gap.
Your AI use touches decisions about people. Screening job applicants or judging whether a customer is creditworthy already sits under employment, equality and data protection law in the UK and the US. In the EU, both appear on the list of high-risk uses in Annex III of the EU AI Act, which names AI systems used "to analyse and filter job applications, and to evaluate candidates" and AI systems "intended to be used to evaluate the creditworthiness of natural persons or establish their credit score." That calls for specialist governance advice. It may not call for a full-time hire.
Even then, the hire works to the steward's list. It does not replace it. The person who knows which jobs AI does in your business, and who the team trusts with their questions, is worth more after a specialist arrives, not less.
What should you do this week?
Appoint your steward this week, and give them the three duties in writing. Here is how a business of 20 can do it without a project.
- Ask the team one question: "When you get stuck with AI, who do you ask?" The name that comes up most is usually your steward.
- Find out what is already in use. People are almost certainly using AI tools nobody approved. The Daylight Rule shows how to surface them safely before you decide anything.
- Write the three duties into the steward's role: keep the list, keep the rules, keep the evidence. Name what comes off their desk to make room.
- Tell the whole team. One message from the owner: this person holds the list and the rules, and questions go to them.
- Book the first monthly call for 30 days from now, and ask for three things: the list, the rule that nearly broke, and one thing AI saved.
Then give your steward what they need to teach others. A steward cannot pass on skills nobody gave them, which is where training that actually changes what people do comes in. The steward is also the natural person to hand every new starter the rules on their first day; the Three Firsts shows how to bring a new hire in.
Before you hire anyone for AI, see what you can clear from your own desk first, using the Waiting-On-Me Rule.
Where to go next
If you want a clear answer to "what is AI for here, what may people do with it, and who decides when it goes wrong" before you appoint a steward or write a job ad, that is exactly what my AI strategy and governance work is built for. If you want to see where your business stands first, take the free Business AI Readiness Scorecard: 12 questions across six areas, about five minutes, and the report includes a 90-day plan with three priorities. And if you are planning a leadership day or a conference session on who should own AI in a growing business, it is one of the subjects I cover in keynotes for business audiences.
Sources and further reading
- Artificial intelligence in UK businesses: 2023 to 2026, Office for National Statistics, July 20, 2026 (Business Insights and Conditions Survey, fieldwork June 5 to 28, 2026).
- Half of SMEs Using AI, With Limited Headcount Impact So Far, British Chambers of Commerce with Atos, March 18, 2026.
- Annex III: High-Risk AI Systems, Regulation (EU) 2024/1689 (the EU AI Act), regulation dated June 13, 2024, via the EU AI Act Explorer (Future of Life Institute).
- Small business AI survey press release, Goldman Sachs 10,000 Small Businesses, March 17, 2026 (1,256 US participants, surveyed January 27 to February 4, 2026).
- Lilach Bullock on fractional AI officers for small businesses, modified November 12, 2026, read November 18, 2026. One of the pages that ranks for this question; the author sells the service.
Dan Fitzpatrick is The AI Educator: a keynote speaker, Forbes contributor and five-times bestselling author who has trained more than 150,000 educators in more than 30 countries. More about Dan.



