AI won’t fix timid leadership

In a recent interview, Matt Domo, founder of Amazon Web Services’ Database Division, renowned Future Architect and AI Strategist, and first on MSN’s 2025 ‘Top Ten AI Leaders to Follow’, argues that most AI projects go off the rails because organisations change the technology but do not change how leadership behaves, how work gets done, or how decisions are made. He is clear that “the feature war is over” and that buyers now expect real ROI, not theoretical promise.

From AI experiments to real outcomes

Organisations are not wasting money on AI because the technology is bad. They are wasting money because leadership teams are ducking the behavioural change required to make better, faster decisions.

Everyone is talking about models, vendors and features. Very few are talking about the only question that really matters: are your people actually working differently as a result?

The real constraint: behaviour, not readiness

We often hear that the constraint on AI is “organisational readiness.” That is only half true. The real constraint is leadership teams avoiding the discomfort of changing how decisions get made.

Most organisations are not stuck in pilot purgatory by accident - they are subtly choosing it. It is what happens when leaders:

• Invest in tools but avoid changing who is allowed to decide what, and on what basis.

• Celebrate demos and proofs‑of‑concept instead of confronting long‑running decision bottlenecks.

• Delegate “the AI thing” to innovation teams, keeping it safely away from core workflows and performance conversations.

  
  

You can see the pattern in local government, in FTSE‑listed companies and in multi‑million‑pound charities. The sectors are different; the avoidance looks the same.

If ROI isn’t obvious, it isn’t there

There is a simple test for whether your AI is working. If you cannot clearly point to what has tangibly improved - time saved, demand reduced, decisions made faster - then your AI is not working. It is just being used.

Dashboards, copilots and assistants can all look impressive in isolation. But unless they have changed:

• How long it takes someone to make a safe, confident decision

• How much avoidable demand shows up at your front door

• How much rework and failure demand is quietly draining capacity

…then they are theatre, not transformation.

AI is only human‑centred if it changes how people actually work. Otherwise, it is just another “nice” tool in an already complex process system.

Most leaders know what to do - they avoid doing it

In our work across sectors - from councils under financial pressure to multi‑million‑pound organisations in competitive markets - we rarely meet leaders who do not understand what needs to change.

Most leaders do not struggle with understanding what to do - they struggle with tolerating the discomfort of doing it consistently.

That discomfort looks like:

• Redesigning decision rights so some choices move closer to the frontline

• Explicitly asking teams to act on AI‑generated signals - and then backing them when they do

• Retiring legacy reports, meetings and processes that no longer make sense in an AI‑enabled environment

• Being transparent about new expectations: “this is now how we triage, prioritise and escalate”

This is the leadership work that turns AI from an experiment into infrastructure. And it cannot be delegated to a project team.

From tools to tangible behaviour change

So what does “doing it” actually look like? At Datnexa, we start from a simple premise: technology can accelerate or amplify impact, but it cannot drive it. It is behaviour that drives impact. And behaviour only changes when it is considered, demanded and practised consistently.

In practice, that means:

• Start with one critical decision

Identify a specific decision that, if made better and faster, would materially change your cost base, citizen or customer experience, or risk profile. Not a process. A decision.

• Design the new behaviour, not just the tool

Specify who will do what differently when AI is available: what they will look at, what they will stop doing, how they will know they are “using it right”.

• Make acting on signals the default, not the exception

  
  

If AI surfaces risk, opportunity or prioritisation signals, define the expected response and time frame. Better decisions should become the norm, not heroic exceptions.

• Measure value in human‑centred terms

Track things your people and communities actually feel: fewer hand-offs, shorter backlogs, less after‑hours firefighting, clearer accountability. Make it obvious that the new behaviour is paying off.

At Datnexa, our Human‑Centred AI, Leadership & OD Partner, Alexandra Thornton, has been developing our training programmes so they are grounded in the realities of UK local authorities and mission‑driven organisations. Drawing on her expertise in communication, behavioural change and performance, Alex works with leaders, managers and practitioners to build the mindsets, behaviours and governance needed to make confident, responsible decisions about AI and data.

Through high‑impact workshops, coaching and co‑created learning journeys, we bridge the often‑missed link between technology, strategy and organisational culture - ensuring AI initiatives are not just approved, but introduced, adopted and sustained in a genuinely human‑centred way. Our programmes move organisations beyond basic awareness into practical, breakthrough ways of working, enabling councils and other mission‑driven teams to use AI creatively and responsibly while protecting the people at the heart of their services.

This approach applies whether you are running an adult social care service, a national operation or a growth‑stage business. The domain changes; the mechanics of behaviour change do not.

As Alex herself says:

AI will remain an ad hoc - and increasingly expensive - tool in organisations unless their senior leaders have changed how their people think, decide and act.

Technology doesn’t drive impact - behaviour does. And behaviour only changes when it’s considered, demanded and practised consistently.

A challenge to senior leaders

So here is the mic‑drop question every leadership team should be wrestling with:

Where in your organisation are better decisions already obvious - but still not happening?

That is where your AI programme should start - not with another pilot, but with a deliberate shift in how people are expected and supported to decide.