Session Speaker: James Holden, Holden Thinking
Theme: Future Facing: Technology and Operations
Date: Summer Trimester, July 2026
At our recent retreat, business psychologist James Holden brought a deliberately human perspective to AI. Rather than focusing on tools or technical capability, James explored what happens to people when technology changes faster than organisations can absorb it. Members challenged his thinking against the realities of running businesses at very different stages of AI adoption. This article looks at what that means for SME leaders.
It’s easy to frame AI as a technology decision. Which platform should we use? Where should we automate? What can we make faster? What will it cost? How quickly should we move?
But those questions tend to miss much of what leaders are actually dealing with.
AI is arriving in businesses already coping with continuous change. It’s developing so fast that something learned a few months ago can already feel out of date. Combine that with the fact that some employees are enthusiastic, others are sceptical or anxious, and some are just getting on with experimenting while others don’t even know where to begin.
This means that leaders are dealing with uncertainty.
They’re being asked to make decisions about technology they may understand less well than people elsewhere in the business. They’re having to judge investments where the economics and capabilities are still moving. At the same time, employees want to know what AI means for their jobs, their expertise and what the business will expect from them.
James’s argument was that the technology has arrived, but the transition hasn’t, and that AI adoption isn’t primarily a technology challenge anymore; it’s increasingly a people and behaviour challenge.
One of James Holden’s most useful ideas was beginner mode.
Many experienced people have spent years becoming good at what they do. Expertise brings confidence, professional identity and often status within an organisation.
AI unsettles that.
Suddenly, someone with twenty years’ experience can find themselves alongside a colleague who has been experimenting with AI for six months and can complete parts of the work dramatically differently.
James shared research indicating that a third of professionals are embarrassed by how little they know about AI. His wider point was more interesting than the statistic. AI is forcing people who are accustomed to being competent back onto a learning curve.
That creates a psychological challenge as much as a skills gap.
It’s relatively easy to tell someone they need training. It’s harder to acknowledge what it feels like to go from expert to novice. Or to admit, particularly as a leader, that you don’t know what the right answer is.
And because the technology keeps changing, there may not be a comfortable point at which everyone finally catches up. Learning, experimenting, discarding what no longer works and learning again could simply become part of working life.
That requires more than technical competence. It requires people to become comfortable with not knowing.
The discussion uncovered just how uneven AI adoption can be inside a single organisation.
A business can be experimenting, using and embedding AI at the same time. One team may have redesigned a workflow around it while another is using it to tidy emails. Some individuals may be considerably further ahead than their managers.
James described an AI adoption ladder running from watching, through experimenting and everyday use, to embedding AI into workflows and eventually reimagining products, services or the organisation itself. The conversation suggested that it’s less a neat ladder than a moving picture. New capabilities can send even experienced users back into experimentation.
That matters because leaders can easily misread where their organisation really is.
James highlighted further research showing a substantial perception gap between senior leaders and employees over how widespread AI adoption actually is. Giving everyone access to a tool can look like adoption from the boardroom. It can feel very different to the person still wondering what they’re supposed to do with it.
There’s another challenge as well. AI is supposed to remove work, but during transition it can create more of it.
There are tools to evaluate, prompts to learn, outputs to check, policies to consider and new possibilities appearing continuously. Experimentation itself consumes time and attention.
For an already busy organisation, AI can therefore feel less like productivity and more like another demand competing for limited cognitive capacity.
James’s phrase for this was particularly useful: AI can be the final straw, rather than the root cause.
This was perhaps the strongest leadership implication from the session because AI challenges a familiar source of leadership authority: knowing more than everyone else.
For many SME leaders, expertise is part of how they got there. They’ve solved problems, made judgement calls and accumulated enough experience to know what good looks like, but if technology is changing faster than any individual can follow, that model becomes increasingly difficult to sustain.
James compared an older model of leadership, based around having the answers, owning what good looks like and controlling the work, with one based around asking better questions, building collective intelligence and creating the conditions in which other people can succeed.
The interesting challenge from the room was that this isn’t really a new definition of leadership and perhaps AI is just simply exposing the difference between leadership and expertise more clearly.
An SME owner doesn’t need to be the best AI user in the business. They do need to know whether people can experiment safely. They need enough understanding to ask sensible questions. They need to recognise where good thinking is happening and connect people who can learn from each other.
Most importantly, they still have to exercise judgement.
James framed the increasingly valuable human contribution around three areas: sense-making, alignment and judgement.
Sense-making is understanding context, asking better questions and seeing how the pieces fit together. Alignment is understanding human needs, building trust and reconciling competing priorities. Judgement is evaluating what AI produces, making trade-offs and remaining accountable for the consequences.
That last point became particularly interesting in the discussion because the risk isn’t simply that AI gets something wrong. Humans get things wrong too.
The bigger risk is that because AI can produce something quickly, confidently and repeatedly, people gradually stop exercising the judgement required to question it, much like they are often encouraged to do with processes today.
AI can remove cognitive load. But judgement is one form of cognitive load leaders shouldn’t be too eager to remove.
This also changes how leaders should think about slow adoption. When someone isn’t using AI, it’s tempting to conclude that they need more training or need to be told to use it.
James offered a more useful framework drawn from behavioural science: capability, opportunity and motivation.
Capability asks whether somebody actually knows how to use AI effectively and has enough confidence and judgement to evaluate the result.
Opportunity asks whether the organisation makes that possible. Have people got time to experiment? Can they access the right systems and data? Are the guardrails clear enough that they know what they can and can’t do?
Motivation goes deeper. Does the individual actually want to engage?
That’s where questions of trust, identity and reward become important.
Imagine an experienced employee who learns to use AI and can now complete their work considerably faster. What happens next?
Do they benefit from the time saved? Does the business simply give them more work? Does their performance expectation increase permanently? What happens to the colleague who chooses not to use AI? And what happens to the status of the experienced expert when a less experienced colleague becomes dramatically more productive?
These aren’t technology questions. They’re questions about the unwritten deal between a business and its people.
James argued that this deal is already starting to shift. Employees will increasingly want to understand how AI-enabled productivity is rewarded, what level of AI use is expected, how it will be monitored and what remains valuable about their contribution.
For SME leaders, leaving those questions unanswered is itself an answer. People will fill the gaps with their own assumptions.
The most useful question of the session wasn’t about adoption at all.
We all knew the familiar question that employees ask:
“Will AI replace me?”
But James replaced it with:
“As AI reshapes how work gets done… where can I make the biggest difference?”
This is an important shift in thinking. The first question encourages people to defend the work they currently do. The second asks them to reconsider where their value lies both today and in the future. And that applies just as much to leaders.
AI will keep changing. Tools will improve, costs will move and some of today’s experiments will turn out to be dead ends.
Waiting until all of that settles isn’t much of a strategy, but neither is pushing people to adopt AI simply because the technology exists.
The more useful leadership question is what’s preventing sensible change in your business.
Sometimes that will be capability. Sometimes the organisation itself is creating friction. Sometimes people understand perfectly well what AI can do and simply don’t trust what its arrival means for them.
Those require very different leadership responses. Perhaps that was the session's most useful implication.
AI isn’t removing the human element from leadership. It’s making it harder to ignore.
As intelligence and answers become easier to access, the scarce qualities may increasingly be context, curiosity, trust and judgement.
And those remain leadership responsibilities.