Ask a founder of a 30-person firm who owns AI decisions and you tend to get a small pause, then a half-shrug, then a name. “Well, I suppose Sarah in operations has been doing most of it.” Sarah, or whoever it is in your version of the sentence, has been quietly trialling Copilot, working out which AI tools the team can actually use without breaching client confidentiality, writing the unofficial prompt library on the side of her desk. Nobody told her to. Nobody is paying her for it. Nobody has said it is her decision when staff push for a new tool.
That is the position a lot of owner-managed firms find themselves in. The AI question arrived through the side door, mostly via Microsoft, Google and the vendors who quietly added AI features to existing software. The technical reality has moved on. The governance reality has not. And the gap between the two is now wide enough to be a leadership problem.
What is an internal AI lead in a small firm?
An internal AI lead in a 20-to-50-person firm is the named, accountable owner of how the business chooses, uses and governs AI tools. The role is typically carved out of an existing operations, finance or client-delivery position, with explicit time, a small budget, decision rights on tools and prompts, and a standing line into the founder. Half a day a week, not a job title from big tech.
The distinction between an AI champion and an AI lead matters here. A champion advocates and experiments, often on their own initiative, with no formal mandate. A lead owns outcomes, makes decisions about which tools the firm uses, sets training standards, and reports on what is and is not working. Practitioner analysis of the two roles suggests champions are the right call when fewer than three use cases have shipped and stuck. Once experiments accumulate and staff are pulling in different directions, the case for a named lead gets harder to ignore.
Why does it matter for your business?
Three reasons stand on their own. Regulators no longer accept “we did not know staff were using it” as a defence, and the Information Commissioner’s Office expects firms of any size to have a named owner for AI and data protection decisions. Founder time is the second reason, AI decisions otherwise pile up on your desk. The third is staff clarity on what is allowed.
On the regulatory point, the EU AI Act, though not directly binding on UK-only firms, signals the direction the UK is travelling in, with proportionate but explicit accountability for smaller organisations. If a client complaint, a data incident, or a regulator question arrives, you need to be able to say who decided this, not who happened to be using it.
On the founder-time point, AI adds a new category of decisions every week, vendor pitches, prompt standards, staff requests for licences, training questions, data-residency choices. If those keep landing on your desk, you have answered the delegation question by default and made yourself slower in the process. On the staff point, in the absence of a named owner people either avoid AI entirely or use whatever they find on the open web, including the tools you would rather they did not.
Where will you actually meet this decision?
You meet it the moment someone asks for budget to buy a tool, hands in a piece of work that was clearly AI-drafted, or quietly admits at a 1:1 that they have been pasting client data into ChatGPT. You meet it when a client questionnaire asks how you govern AI use, or when the unofficial AI champion starts to feel taken for granted.
The recognition dimension is worth pausing on. In small firms the de facto AI champion is often a senior operations manager, a practice manager, or a senior administrator. Those roles are disproportionately held by women, and the informal AI work tends to be done on top of the day job rather than instead of it. Research on senior women in leadership consistently shows higher burnout rates and a pattern of being overlooked when informal contributions are not formalised. If your firm is relying on goodwill to hold the AI line, the fairness question and the retention question are the same question.
When to formalise the role, and when to wait
Formalise when three or more AI experiments have shipped and stuck, when staff in different functions are asking different vendors for budget, when the same governance question keeps reaching you, or when a client or insurer has started asking how you handle AI. Wait when the activity is one person tinkering with one tool and no client data is involved. In that case, invest in the champion and set a review date.
If you do formalise, keep the shape proportionate. Half a day a week of protected time. A clearly bounded tools budget, often a few thousand pounds a year for licences and training. Decision rights on which vendors the firm uses and which prompts go into the shared library. A standing quarterly report to you on what is working, what was retired, and where the risk sits. A line in their objectives, a line in their pay review.
The trap to avoid is appointing someone with the title but none of the authority. Practitioner data on founder dependency shows that nominal delegation, where the role exists on paper but every decision still routes through the founder, makes the bottleneck worse rather than better.
What you should actually do this quarter
If you recognise your firm in the opening paragraph, the first move is a conversation with the person already doing the work. Ask them what they have been doing, what they would do if they had time and authority, and what they would want in exchange for taking the role on formally. Design the role around that conversation rather than around a job description you found online.
PwC’s analysis of AI ROI in mid-market firms found that the companies delivering real financial returns were those that gave AI clear leadership and governance foundations, not those that bought the largest stack of tools. The pattern is consistent with what UK SME surveys are reporting on the AI confidence gap, the constraint is rarely access to technology and almost always the absence of someone owning the decisions.
For many 30-person firms, the first AI hire is already on the payroll. The decision is whether to name them, fund the half-day a week, and put the authority behind the title, or to keep relying on goodwill that the person doing the work is gradually running out of. If you would like to talk through what the role looks like for your firm specifically, Book a conversation.
