By Sarah Luxford – Partner, Digital, Data & Technology (DDaT) – GatenbySanderson
London Tech Week always carries a certain charge, and the 2026 edition was no exception. As a proud south Londoner, Croydon born, I’ll happily own my London bias, but the energy across Olympia was genuinely palpable: a new conversation every few minutes, old colleagues turning up exactly where you’d hope, and a clear sense that the UK is not just talking about technology leadership but committing serious capital to it.
The capital was the headline. The Prime Minister opened with a £400 million sovereign compute strategy and a now-familiar challenge: that Britain should shape the AI revolution rather than be shaped by it. Microsoft pointed to a $30 billion commitment to UK AI infrastructure and skills. AWS reaffirmed £8 billion of investment. The Mayor announced a £12 million programme to help SMEs adopt AI. The Office for Quantum confirmed a £2 billion package.
By any measure, a remarkable show of confidence in UK technology. Yet the thing I found most thought-provoking all day was not a number with a pound sign in front of it. It was a question hidden inside one.
The 64/24 question
AWS’s Alison Kay put up a single statistic that reframed the morning for me. Nearly two-thirds of UK businesses, some 64%, have now adopted AI. But only 24% are using it for anything advanced. Most remain at the level of chatbots and document summaries. As one speaker put it, half of organisations own a smartphone but use it only for phone calls.
That gap shows up everywhere, public and private alike. And the barrier to closing it was named explicitly throughout the day. Not money, and not compute. Skills. Almost half of UK businesses cited the skills gap as their single largest obstacle. The encouraging counterpoint is that 67% of workers say they want to develop AI skills. The appetite is running well ahead of the capability.
This is the quiet truth beneath the investment announcements. The capital is becoming the easy part. The harder, more interesting question is whether organisations have the people, and the leadership, to turn any of it into transformed services.
From information work to intelligence work
What makes this a leadership question, rather than only a training-budget one, came through in two of the day’s standout sessions.
Microsoft’s Darren Hardman described a shift from information work to intelligence work: away from task execution and towards directing AI-enabled processes, with AI as collaborator rather than tool. Aravind Srinivas of Perplexity offered the metaphor that stayed with me. As AI orchestrates many models and tools at once, the human role becomes that of the conductor, with the AI models as the musicians. Baroness Martha Lane Fox captured the optimism when she said the possibility to have agency has never been stronger.
There is a thread connecting those ideas. An orchestra of brilliant musicians still needs a conductor to make music rather than noise. Agency only becomes outcome when someone provides direction. And the person who decides what gets played, to what end and within what guardrails, is an organisation’s leadership, not its technology team alone.
That is why I came away thinking of this as a leadership story more than a technology one. AWS framed five areas leaders are now asked to hold together: ethics, cost, security and governance, workforce skills, and service reinvention. What seems to separate the organisations using AI well from those still finding their feet is rarely the size of the compute budget. More often it is whether the whole leadership team has decided who owns each of those threads.
Five questions worth taking to your next board meeting
I offer these less as a scorecard and more as a single agenda item for a board or leadership team. In a public service context, two of them carry a particular charge, because we are not only adopting AI, we are answering to the people we serve as we do it.
- Ethics and trust. Is our use of AI fair, explainable and defensible to the public, and who owns that question? If you cannot put your position on a single page a frontline colleague would understand, that is a more revealing gap than any technical one.
- The citizen’s experience. As Tom Loosemore has argued this year, citizens increasingly meet the state through AI without always knowing it, which raises real questions of trust, equity and legitimacy. Public expectations are rising fast, shaped by the best private services people use. Are we matching that, and are we being honest with people about where AI sits in their interaction with us?
- Cost and value. Are we making deliberate investment choices, or quietly accumulating pilots that never scale? If no one can produce a list of every AI pilot running right now within a week, the real problem is governance, not money.
- Workforce and skills. Are our people supported to move from basic use towards genuine intelligence work, and who is leading that shift? This is the rare strategic problem you can begin working on today, for nothing.
- Service reinvention. Following Martha Lane Fox’s founding GDS principle, the test is not whether AI lets us do familiar things faster, but whether it lets us redesign a service around the people it currently fails. Pick one service and ask: if we designed this today, knowing what these tools can do, would it look anything like it does now?
The value is in naming these clearly, deciding together who holds each one, and giving that person enough authority to actually hold it.
A note on building the skills
On workforce, I want to be useful rather than promotional, so a few genuinely good starting points across the range. For free, the Open University and Microsoft have built an AI Hub on OpenLearn with more than thirty open courses, from first-day literacy to applied workplace use, and the civil service offers growing AI literacy training built around confidence and judgement rather than coding. For something more structured, a number of business schools now run senior AI leadership programmes aimed squarely at executives and the governance questions above, alongside shorter online options. There is a whole ecosystem here, free and paid, online and in person. The specific course matters far less than the decision to start, and a leader who has spent a couple of hours actually using these tools asks better questions of everyone who reports to them.
The public sector’s particular context
These questions land differently depending on where you sit, and the public sector navigates a context others often do not even see. Procurement that a private company can move through quickly travels a more considered path into a public body, which is why initiatives like ProQure matter. Legacy estates carrying critical national functions cannot be replaced on a whim. Decisions are made under public scrutiny and a standard of accountability the private sector rarely faces. Pay frameworks mean public organisations compete for scarce DDaT talent against salaries they cannot match.
I see these not as failings but as the genuinely complex conditions public leaders work within, often with great skill. They are also why the people questions above carry more weight here, not less. Where you cannot simply outspend the problem, the quality of your leadership, the clarity of your governance and the capability of your existing workforce are very nearly the whole game. And as AI reshapes who gets access to opportunity, keeping a steady focus on gender balance and diversity matters more, not less.
What I took away, and where I fit
The Prime Minister’s framing, shape it or be shaped by it, is a good one. For any organisation it seems to begin with a deceptively simple question: do we have the people who can direct this, and have we given them the room to do so?
I should be honest about my own place in all this. I am not the AI expert, and I would be wary of anyone in my position who claimed to be. What I do understand deeply is the context public leaders operate in, and how to find and convene the right people around a problem. Most of the capability this moment asks for can be built from where you already stand: start the courses, ask the five questions, decide who owns each thread. And on the occasions when the honest answer is “we don’t yet have the leadership this needs,” that is the conversation I am genuinely glad to have.
If any of this was useful, or if you are thinking about DDaT leadership in your own organisation, do get in touch.
Sarah Luxford is a Partner at GatenbySanderson specialising in Digital, Data and Technology (DDaT) leadership appointments across the public sector. Contact the team at ddat@gatenbysanderson.com.