When organisations talk about AI transformation, the conversation quickly becomes technical.

Which tools should we use? How do we integrate them into the existing infrastructure? What about compliance, data protection, governance and security? Who gets access? Which processes can be automated?

All of these questions matter.

But they are probably not the hardest part.

The hardest part is the same one organisations have struggled with for decades whenever they tried to introduce something new: people.

We have seen this before with digitalisation, organisational restructurings, new collaboration models and agile working. A decision is made somewhere at the top of the organisation. A programme is launched. New processes are designed. Employees receive presentations explaining what is going to change.

And then management is surprised when people do not enthusiastically embrace it.

The problem is often not that employees are per se resistant to change, or that everyone simply dislikes anything new. The problem is that nobody has given them a convincing reason why they should want the change in the first place.

The weekend golf problem

There is a pattern I have seen in organisations that could almost be described as the “weekend golf problem”.

Imagine a senior executive hearing about agile working during a conversation over the weekend.

“Have you heard about Agile? Everyone is doing it.”

“What exactly is it?”

“Something with sprints. Teams work faster.”

“Excellent. We need that.”

Monday morning, the management team receives the message: From now on, we are agile.

Teams are trained. New terminology appears. Sprints, stand-ups and retrospectives become part of organisational vocabulary.

Then, two weeks later, a senior manager walks into the room and says: “I know you planned something else for this sprint, but this is now the top priority.”

And suddenly the organisation discovers that everyone is expected to work differently — except the people at the top.

AI risks following exactly the same pattern.

Leadership teams decide that the organisation needs to become “AI-first”. Tools are introduced. Employees are encouraged — or instructed — to use them.

But the underlying ways of working remain largely unchanged.

That is not transformation.

It is simply adding another layer of technology to an existing system.

Faster is not necessarily better

One of the biggest risks with AI is that organisations will use it primarily to produce more of what they already produce.

More emails. More reports. More presentations. More analysis. More documentation. More communication.

We have seen this dynamic before.

When communication was slower and sending information required effort, people had to think more carefully about what was worth sending. A letter or fax took time. Communication therefore had a natural friction built into it.

Email removed much of that friction.

This was supposed to make work more efficient.

And technically, it did.

But instead of simply spending less time communicating, organisations dramatically increased the amount of communication.

Today, many employees spend large parts of their working day processing information that other employees were able to produce and distribute extremely easily.

AI could multiply this effect.

If producing a report takes ten minutes instead of three hours, organisations may not simply spend less time producing reports. They may start producing ten times as many reports.

If writing an email takes thirty seconds, employees may send even more emails.

If preparing a presentation becomes almost instantaneous, the number of presentations may explode.

The technology becomes more efficient.

The organisation does not.

The real opportunity is redesign

This is why the most important question should not be:

“How can AI help us do this task faster?”

The better question is:

“Why are we doing this task at all?”

AI creates an opportunity to rethink work rather than simply accelerate it.

Which meetings are actually necessary?

Which reports are genuinely used for decisions?

Which approval steps still add value?

Which information needs to be produced?

Where could employees gain more autonomy?

Which repetitive tasks could disappear entirely?

There is an old idea about perfection: it is achieved not when there is nothing left to add, but when there is nothing left to remove.

That principle may be particularly useful when organisations redesign work around AI.

The goal should not be more output.

The goal should be better work.

Give people a reason to participate

This leads to another uncomfortable question organisations need to answer:

What exactly do employees gain from AI transformation?

From a company perspective, the business case may be clear. Increased productivity. Lower costs. Faster decisions. Greater competitiveness.

But these arguments do not automatically create motivation among employees.

If the message employees hear is essentially:

“We will become much more productive, and potentially need fewer people,”

their lack of enthusiasm should not be surprising.

A credible AI transformation therefore needs an employee proposition as well.

Perhaps AI removes repetitive administrative work.

Perhaps teams gain greater autonomy.

Perhaps people can spend more time on judgment, creativity, relationships and problem-solving.

Perhaps roles become more interesting because technology takes over some of the least valuable parts of the work.

But these outcomes will not happen automatically.

They require deliberate work design.

And they require honesty.

Some tasks will disappear. Some roles will change significantly. In certain areas, fewer people may indeed be needed.

Pretending otherwise is unlikely to create trust.

People are often able to deal with uncertainty when there is trust in the organisation and in its leadership. If employees believe that they are valued, that management will communicate honestly and that difficult decisions will not simply be made over their heads, uncertainty becomes easier to navigate.

This is where trust and psychological safety matter.

Organisations that have built credibility over time — for example by showing that they do not immediately resort to layoffs whenever conditions become difficult — start from a very different position than organisations where employees already expect that efficiency gains will come at their expense.

People do not need guarantees that nothing will ever change.

But they do need reasons to believe that the organisation will deal with change fairly and transparently.

The people doing the work may have the best ideas

One of the simplest ways to improve AI transformation is also one of the most frequently neglected:

Talk to the people who actually do the work.

Employees usually know where processes are unnecessarily complicated. They know where information gets duplicated. They know which approvals slow everything down. They know which reports nobody reads.

They may therefore have a much better understanding of where AI could create meaningful value than someone looking at the organisation from several management levels above.

This does not mean every employee needs to design the AI strategy.

It means organisations should involve people early enough for their experience to influence how the technology is used.

Pilot projects can be particularly useful here.

Start with teams that are interested in experimenting. Let them test new ways of working. Learn from what succeeds and what fails. Allow other teams to observe what is happening.

Change has always spread unevenly through organisations. There will be early adopters, a large group that waits to see what happens, and some people who ultimately decide that the new way of working is not for them.

That is normal.

The goal should not be to eliminate every form of resistance.

The goal should be to create enough understanding, participation, trust and credibility that people can make sense of the change.

Three principles for AI transformation

If organisations want AI transformation to become more than another technology rollout, three principles may be a good starting point.

First, explain the why before the how. People need to understand what the organisation is trying to achieve and what the change means for their own work.

Second, redesign work instead of simply accelerating it. Using AI to produce more of the same may create additional workload rather than genuine productivity.

Third, involve the people who actually do the work. They often understand the processes, bottlenecks and opportunities better than anyone else.

There is also a fourth issue waiting behind all of this: leadership itself.

If AI fundamentally changes how organisations work, managers cannot expect everyone below them to change while their own behaviour remains untouched.

But that is a subject for another article.

For now, the central point is simpler.

AI transformation is not primarily a technology project.

It is an organisational change project enabled by technology.

And organisations that forget the human factor may discover that installing the technology was the easy part.