As organisations move from AI experimentation to implementation, the questions they face are equally changing: decisions about which tools to test or where AI might create efficiencies quickly become questions about how work should be organised, who should make decisions, where accountability sits and how people experience the change. At the same time, the longer-term impact of AI on jobs, skills and organisational structures remains uncertain, leaving leaders with important choices about how that transformation should unfold.
Those choices place leadership and culture at the centre of AI adoption. Organisations need to define what they are trying to achieve, create clear ownership, give people space to experiment, preserve human judgement and decide how productivity gains should translate into changes to work and roles. They also need to maintain trust while making those decisions in an environment where both the technology and its implications are developing rapidly.
This was the central theme of the closing panel discussion at our recent AI Agenda event. Bringing together perspectives from technology, workforce strategy, legal leadership and large-scale organisational transformation, the discussion repeatedly returned to a common conclusion: successful AI adoption depends as much on the organisational conditions surrounding the technology as on the technology itself.
Start with purpose
AI projects can easily begin with the technology itself: leaders see competitors experimenting, new tools appear in rapid succession, and pressure builds to demonstrate that the organisation is keeping pace.
Reflecting on that pressure, Tim Flagg, CEO and co-founder of UKAI, described a widening gap between excitement about AI and confidence in knowing what to do with it. The risk, he underlined, is that organisations respond to this uncertainty by prioritising activity over clarity: launching pilots, acquiring tools or setting ambitious efficiency targets before deciding exactly which problems they are trying to solve.
The panel shared how an early AI initiative had initially been approached much like a conventional technology implementation, with the CIO and technology function taking the lead. After six to nine months, the programme had struggled to gain traction and was eventually stopped and restarted.
The reset changed the way the organisation thought about the project: responsibility was shared between the CIO and CPO, reflecting the recognition that the workforce implications were inseparable from the technology itself.
That experience points to a wider lesson: effective AI transformation begins with organisational intent. Leaders therefore need a clear account of the problem they are trying to solve, the value they hope to create and the kind of organisation they want AI to help them build.
For Samantha Schlimper, Managing Director at Randstad Enterprise, with responsibility for Randstad Talent Advisory, that question of intent extends beyond immediate commercial outcomes, as leaders also need to consider what they want work to become, what they want their organisation to stand for and how AI adoption fits within those wider commitments.
These questions become particularly important when short-term incentives encourage rapid action: the possibility of reducing cost or increasing productivity may be compelling, but sustainable transformation requires leaders to consider what those decisions mean for capability, trust and organisational performance over a longer horizon.
Culture is part of the infrastructure
Culture can sometimes appear peripheral in conversations about technological transformation. In practice, however, it determines how people respond when the technology arrives.
Employees decide whether to experiment, whether to share concerns, whether to admit that something has gone wrong and whether to contribute their knowledge to redesigning work, and those behaviours are shaped by what people believe their organisation values and rewards.
Schlimper argued that psychological safety therefore becomes a basic condition for AI transformation. Employees need confidence about why change is happening and what it may mean for them. If people believe that revealing how their work is performed could simply accelerate the removal of their role, they have little incentive to share the detailed knowledge leaders need to redesign that work effectively.
That knowledge is critical as organisations operate through much more than formal processes. Much of how work actually happens exists tacitly in people’s experience, relationships, informal practices, and judgement. A process map may describe the sequence of tasks, while the employees performing those tasks understand where exceptions arise, which relationships matter and where judgement is required.
Culture therefore determines whether that knowledge becomes available to the organisation, and shapes how readily people learn.
Flagg argued that organisations need cultures characterised by humility, curiosity and experimentation. AI is developing too quickly for leaders to rely on complete expertise or settled playbooks. Leaders therefore need confidence in learning, asking basic questions and allowing others to contribute knowledge they themselves may lack.
Experimentation similarly works best when accompanied by clear guardrails: employees need to understand where they can explore, what information they can use, what risks require escalation and where human review remains essential. Within those boundaries, organisations can create space for people to test ideas, learn from failures and identify uses that senior leaders may never have designed from the top.
Redesign work with the people who understand it
The practical challenge becomes greater as AI moves from isolated tools into the structure of work itself.
At that point, leaders need to move beyond adoption metrics and examine the relationship between outcomes, tasks and skills. Which activities should remain human? Where can AI provide support? Where can automation remove unnecessary work? And what happens to the capabilities people previously developed through the tasks that disappear?
Schlimper described work redesign as a process of understanding the outcomes an organisation wants to create, the tasks involved and the skills required, before deciding how human and technological capabilities should interact.
Employee involvement is central to that process: the people doing the work often possess the richest understanding of how it really functions, and involving them can improve the design itself while also creating greater legitimacy around change.
Transparency is equally important: employees will naturally ask what productivity improvements mean for their own future. Will greater efficiency create more space for strategic, creative or relational work? Will jobs be redesigned? Will people be retrained? Will headcount fall?
Leadership credibility depends partly on how clearly those questions are addressed.
Oliver Loach, People Director at Mitie, contrasted approaches in which AI-driven efficiency translates quickly into workforce reduction with examples where organisations have deliberately redesigned roles around new forms of value. Those choices send powerful cultural signals. Employees learn from them what the organisation ultimately means when it talks about AI transformation.
Protect the capabilities you will need next
The redesign of work creates another, longer-term challenge: organisations need to preserve the capabilities through which expertise and judgement develop.
AI can increasingly produce information, generate drafts, analyse material and support decisions. As those capabilities become more widely available, the distinctive contribution of people may shift.
Lexi Lutz, General Counsel at Summize, argued that context and discernment become increasingly valuable. In professional environments such as law, knowledge has historically been a major source of expertise. AI makes access to knowledge much easier: experience, institutional understanding, judgement and the ability to recognise when an apparently convincing answer is wrong consequently become more significant.
This has profound implications for learning. Many professional skills have traditionally been developed through relatively routine work: junior employees learn by drafting, researching, checking, revising and sometimes making mistakes. If AI increasingly performs those activities, organisations need to think deliberately about how future employees will develop the judgement required to supervise the technology later in their careers.
The same issue applies to management and leadership. As hybrid teams involving humans and AI systems become more common, leaders may spend less time coordinating routine activity and more time exercising judgement, supporting people, interpreting context and deciding when automated outputs should be challenged.
AI literacy therefore needs to extend beyond knowing how to operate a tool: employees and managers need to understand when to use AI, why it is appropriate, how to evaluate its output, how to discuss its use and where responsibility ultimately sits.
The organisations that benefit most from AI may consequently be those that become increasingly deliberate about developing distinctly human capabilities alongside technological ones.
Leadership at the crossroads
At the end of the panel, each speaker was asked to identify one leadership discipline that organisations should strengthen over the coming year.
Perhaps unsurprisingly, none chose a technical capability. Instead, all four focused on distinctly human-centred attributes.
Flagg chose humility: the confidence to recognise the limits of one’s own knowledge and empower others to learn.
Schlimper chose intent: clarity about what work should become and what the organisation is ultimately trying to achieve.
Lutz chose empathy: understanding how technological change is experienced by individuals, particularly those who may feel uncertain about their future.
Loach chose transparency: explaining why AI is being introduced and what its consequences are likely to be.
Combined, those disciplines offer a useful framework for the next stage of AI transformation.
The future of AI at work is unlikely to emerge through a single predetermined path: organisations are making choices now about how technology is deployed, how work is redesigned, how productivity gains are distributed and which human capabilities they continue to cultivate.
Leadership and culture are therefore some of the principal forces shaping which of those possible futures emerges.
At this crossroads, the organisations best placed to succeed will certainly need technological capability, but they will also need the confidence to experiment, the judgement to set boundaries, the patience to build trust and the clarity to decide what kind of workplace they are trying to create.
