Cornerstone article
The AI Leadership Gap in UK Professional Services
What the AI leadership gap in UK professional services looks like, why it sits at partner level, and what good AI governance looks like in practice.
By Gwendolyn Smythson, HR and Leadership Practice Lead, The AI Consultancy (London) Ltd.
Published · Reading time approximately 11 minutes.

The AI bottleneck inside UK professional services firms in 2026 is not a technology problem. It is a leadership problem, and the data from 2024 and 2025 is clearer on this point than most partnerships have internalised. This article sets out what the leadership gap in UK professional services actually looks like, why the leadership layer is the bottleneck rather than the tooling, what governance decisions only senior leaders can make, what good AI leadership looks like in practice, and a direct next step.
The framing throughout is to describe what senior leaders are accountable for. Nothing in this article claims regulatory endorsement on behalf of Learn AI; it describes obligations that already sit with the partnership, the executive committee, or the board under UK law and professional rules.
What does the AI leadership gap in UK professional services actually look like?
The AI leadership gap is the distance between what senior leaders at UK professional services firms know about AI and what they are operationally ready to decide about AI inside their firm. Three figures from 2025 research set the shape of the problem.
Fifty-five per cent of C-suite leaders in UK professional services rate themselves as the least prepared group in their firm for AI adoption (Skills England and the Department for Science, Innovation and Technology workforce research, 2025). The figure is a self-assessment by the people who sign off the AI budget and the AI risk appetite. It is not a statement about the workforce.
Eighty-one per cent of partners at UK legal firms are rated least prepared for AI adoption inside their own firm (Law Society AI readiness survey, 2025). Inside the narrowest definition of UK professional services, the pattern is the same as in the broader C-suite figure, and it is sharper.
Sixty-one per cent of professional services firms have abandoned at least one AI project because of skills gaps at leadership level (Skills England and the Department for Science, Innovation and Technology workforce research, 2025). That figure captures what happens after the pilot has been approved. Firms are not standing still on AI; they are buying tools, running pilots, and then running into a wall that is located above the team that was running the pilot, not below it.
Read together, the three figures describe a market where the partnership has been aware of AI for at least two years, has authorised spend, has run pilots, and has hit a leadership-level wall at the point where the pilot needs to become a firm-wide position. The wall is at the top of the firm, not at workforce level, and two independent 2025 research strands have put numbers on it inside twelve months.
Why the leadership layer is the bottleneck, not the technology
The technology works. The tooling exists. The sector-specific use cases are well-documented. The firms that have abandoned AI projects have not abandoned them because the tool did not perform; they have abandoned them because the partnership could not agree on the decisions that the pilot surfaced.
Four structural reasons explain why the leadership layer is where the wall sits.
First, governance authority in UK professional services firms sits with the partnership or the board by law and by professional rule, not with the IT function. The Solicitors Regulation Authority expects solicitors to maintain competence and supervise delegated work, and the named fee-earner cannot delegate that obligation to an IT team. The Information Commissioner's Office expects the data controller to know what personal data is going where; the controller is the firm, represented by its senior leaders. ICAEW, CIMA, and ACCA each expect their members to evidence professional competence on any work that bears the member's name. None of those obligations can be discharged by the IT partner, the external vendor, or the cloud provider.
Second, risk appetite on AI cannot be delegated. Someone in the firm has to decide what the firm will and will not do with AI in front of clients, what it will and will not put AI-assisted output its name to, and what it will commit to the regulator if asked. That decision sits with the partnership or the executive committee. It does not sit with the team running the pilot.
Third, budget decisions on AI sit at the top of the firm. An AI pilot that costs £20,000 to run and £80,000 a year to embed requires a partner-level or board-level sign-off in a firm of any size. If partners do not have a defensible view of which pilots to embed and which to kill, the sign-off decision defaults to "not yet", and the pilot joins the 61% abandonment figure.
Fourth, the firm's competency framework and the hiring and promotion rubric sit with the partnership and the people function. If AI fluency is not written into what "good" looks like for an associate, a management accountant, or a senior consultant, the behaviour does not land inside the firm regardless of how much is spent on tools.
The workforce can be trained into AI use inside a month. The leadership position can take six months to form, or it can take one well-prepared half-day session to accelerate. The difference is not the technology.
What AI governance decisions can only senior leaders make in a professional services firm?
Four AI governance decisions in a UK professional services firm belong to the partnership, the executive committee, or the board and cannot sit below that level without creating a control weakness.
The first is the firm's written position on AI. The position is the answer the firm gives when a regulator, a client, or an insurer asks what the firm is doing with AI, on what basis, and with what controls. It is produced at partner level, signed off at partner level, and reviewed at partner level. It is not a marketing document; it is a governance document. Firms without one have a disclosure exposure they have not quantified.
The second is the named owner for AI inside the firm. The owner is a person, not a committee. The owner sits at partner level or executive-committee level and is accountable for keeping the firm's written position in step with the tooling estate, the regulatory environment, and the firm's risk appetite. Firms without a named owner typically have three or four overlapping owners, which is equivalent to having none.
The third is the firm's risk appetite on AI-assisted output leaving the firm. The decisions here are concrete: which categories of work can be AI-assisted and filed, which can be AI-assisted with mandatory review, which are AI-prohibited. The decision is sector-specific and the decision is narrow enough to put on one page. Firms that have this page circulate less, rework less, and escalate less.
The fourth is the budget line for AI governance itself, separate from the budget line for AI tooling. Governance costs include policy drafting, training, the named-owner time, the review cycle, and the evidence-capture process. Firms that put the governance spend in the same line as the tool spend systematically under-invest in governance, because the tool spend always looks more concrete on the finance report. Separating the lines protects the governance spend through the budget cycle.
None of the four decisions is new to professional services firms. They are the same decisions the firm already makes about audit methodology, data handling, client money, or any other firm-wide capability. The firms that have got AI right have recognised the pattern and declined to treat AI as a special case that sits outside the firm's existing governance muscle.
What does good AI leadership look like in practice inside a UK firm?
Good AI leadership in a UK professional services firm is visible and boring. It has the same shape as every other well-governed firm-wide capability, adapted to the specific risk profile of generative AI.
A one-page written AI position, signed at partner or executive-committee level, circulated to every fee-earner or qualified professional in the firm, and referenced in the firm's staff handbook. The position states which tools are approved for which categories of work, which are prohibited on which categories, and where to go with a question. It is reviewed quarterly.
A named AI owner at partner or executive-committee level, with defined time and authority. The owner's name appears on the AI position document. The owner is accountable for the quarterly review and for keeping the document in step with the regulator's evolving position.
A verification step, on the record, for every AI-assisted output that leaves the firm or posts to a regulated system. The verification is procedural: a named reviewer, a date, a checklist item for each relevant regulatory anchor. The verification evidence is retained alongside the output for the same period as the output.
A measurable indicator that the partnership reviews alongside standard firm KPIs. The indicator can be simple: share of matters or engagements where AI-assisted output passed verification without rework, share where verification failed, share of AI-adjacent complaints or near-misses. The indicator does not need to be complex to be load-bearing.
A training pass, specific to the firm's sector and to the work the firm actually does, delivered to the partnership or the executive committee before it is delivered to the workforce. Leadership goes first. The 55% and 81% figures above are measures of what happens when leadership goes last.
None of this requires a dedicated AI team, a six-figure transformation programme, or a technology overhaul. It requires a partnership that has made the decision to treat AI as a governance question rather than a tooling question, and a half-day to collapse the decision into a written position.
How to close the leadership gap in your firm
The Executive AI Briefing format exists specifically to accelerate this.
Half a day, six to twelve senior leaders, sector-specific, vendor-neutral, with governance treated as part of the work rather than as a disclaimer. The output is a partnership-level written position on AI, the named owner, and the first draft of the four governance decisions above. The session is designed to produce the artefacts the partnership will need anyway, compressed into a defensible half-day rather than spread across six months of partnership meetings.
The Executive AI Briefing for UK partners and directors is built around the three sectors Learn AI serves at launch: UK law firms, UK finance and accounting teams, and UK consultancy and advisory firms. The regulatory anchors that apply in each sector are integrated into the session rather than bolted on at the end.
If the partnership is currently debating whether AI warrants a serious investment of firm time, the Briefing is designed to collapse that debate into a defensible position in a single session. If the partnership has already authorised pilots and is now deciding whether to embed or kill them, the Briefing is designed to surface the four governance decisions the pilot has generated and produce a written partnership position on each.
If the partnership is not ready for a Briefing yet, the AI Readiness Assessment produces a fifteen-minute diagnostic covering current maturity, realistic sector-specific use cases, governance gaps mapped to the firm's regulatory anchors, and a recommended next step. The diagnostic is free, takes fifteen minutes, and is a sensible preparatory read before a Briefing rather than a replacement for it.
The AI leadership gap in UK professional services is not closing on its own, and the 55%, 61%, and 81% figures above show that the gap is concentrated where the decisions are made. The firms that treat closing it as the job of the partnership, not the IT team, are the firms that will not be re-running the same AI pilots in 2027.