All insights
    Governance4 min read

    AI needs an owner

    By Sascha Cutura

    A quiet walnut meeting table with a single charcoal chair and linen curtains

    AI use in UK businesses has doubled in a year. The government's latest AI adoption survey puts adoption at 32% of UK businesses in 2026, up from 16% in 2025. The financial picture has moved far less. Among adopters, 64% reported no immediate financial effect on operating costs, and only 14% of businesses considered AI tools to be of high or critical importance to their operations.

    Globally, McKinsey's State of AI 2026 found that 37% of respondents attribute any EBIT impact to AI, about the same share as a year earlier. Only around 6% qualify as high performers.

    Usage is spreading faster than value. In our experience, the gap rarely comes down to the choice of model or vendor. It comes down to three management questions that are easy to defer: who owns AI, what state the underlying processes and data are in, and how anyone will know whether it is working.

    Activity without an owner

    Most of the organisations we work with already have AI running in several places. A customer service team trialling a chatbot. Marketing paying for a writing assistant. Someone in finance who built an automation that now quietly runs part of month-end. Each started for a sensible reason, and each is defended by the people who started it. Few of them report to anyone in particular.

    This is how budgets fragment. Spend spreads across a dozen initiatives, none large enough to move a line in the P&L and none subject to a decision about whether it should continue. Gartner expects over 40% of agentic AI projects to be cancelled by the end of 2027, citing escalating costs, unclear business value or inadequate risk controls. Every one of those causes points back to ownership.

    The pattern is familiar in the UK, and we see a close version of it across Europe. Someone capable and curious runs a trial. It works well enough. Then the pilot sits there. Nobody has the budget to roll it out, or the authority to change the process around it. Twelve months later the licence is quietly cancelled, or worse, still being paid for.

    What ownership actually requires

    Ownership is a set of rights. The owner holds the AI budget, decides which initiatives go ahead, and has the authority to stop work that is not delivering. The third right matters most, and it is the one most often missing. A portfolio that never closes anything is a portfolio nobody is managing.

    In a mid-sized organisation, this person is rarely a new hire. It is usually the managing director, the COO or a founder: someone who already carries commercial accountability and can say no to a department head without convening a committee. Larger organisations may formalise the role with a title. The authority matters more than the job title.

    McKinsey's research points the same way. Its high performers are twice as likely as other organisations to say their senior leaders demonstrate commitment to AI initiatives, and twice as likely to have defined processes for measuring the impact of those initiatives.

    That person also needs permission to change the process, which is where most pilots die. Plugging a tool into an existing workflow rarely delivers much. Removing steps, simplifying approvals and fixing the data underneath is what produces the gain, and that requires someone who can say "we will work differently from Monday".

    One number, reported regularly

    Every AI initiative should carry one measure, agreed before work starts and backed by a baseline. Hours spent chasing overdue invoices each week. Days from order to invoice. The share of customer queries resolved at first contact. One number is enough to decide whether to scale, adjust or stop, and it keeps the conversation in the language the board uses.

    A useful test for any initiative: who owns it, which number is it moving, and at what point would you stop it? Where those answers come quickly, the initiative is ready for investment. Where they do not, that is the first conversation to have.

    Most organisations already have the right person to own AI. What they usually lack is a clear mandate: the budget, the right to stop work, and an agreed number to report against. A leadership team can settle that in a single focused session.

    The 90-Day AI Adoption Plan

    A practical guide for leaders, covering people, governance and a first delivery, with a skills plan and checklist.