In early 2019, I made a deliberate decision.

I’d been watching organisations build their entire strategic direction around technology, choosing the tools first, the brand position second, and the humans a distant third. And I’d come to a conclusion that felt obvious to me but genuinely controversial in most of the rooms I was walking into: the sequence was wrong. Not just suboptimal. Wrong in a way that was going to cost organisations dearly over the next decade.

So I wrote about it, committed to it publicly, and then took the argument on the road, to Canada, Mexico, the Middle East, Asia, New Zealand, and across Australia. The post was called “The future is Human 1st, Brand 2nd, Technology 3rd” and I said, in those specific words, that I was dedicating that year’s speaking, consulting, and media to “the noble cause of rehumanising technology.”

I was 100% comfortable with the position. Still am.

But seven years later, I want to revisit what I said. Not because the direction was wrong, but because something happened in the gap between what I argued and what organisations actually did. And what they did is now arriving as a leadership crisis that’s being explained away as something else.

What I meant by Human First

The argument wasn’t sentimental. It wasn’t “be nicer to your people” or “run more town halls.” It was structural.

Technology-first organisations build their capability and assume people will follow. Brand-first organisations wrap their messaging around what they can already do. Human-first means something different: you start with the actual need of the actual human on the other end (customer, employee, community member) and then reverse-engineer the technology and brand required to meet that need.

In boardrooms and strategy sessions across three continents that year, I kept asking the same question. “Why should your customers care?” And I kept being met by silence, followed, as I wrote at the time, by what could only be described as tumbleweed, and then a list of what I called “motherhood statements”: we’re the biggest, we have great distribution, we’re well-known, blah blah blah.

That silence was diagnostic. Human need had been treated as an afterthought in most strategic planning conversations. Technology came first because it was concrete: it had a price tag, a vendor, a delivery timeline. Human need was abstract, harder to quantify, and therefore easier to defer.

I developed a four-point framework that year to make the human-first question operational. I called it JUSI: Joyful, Unique, Seamless, Immersive. It was a checklist for assessing whether an organisation was genuinely placing human experience at the centre of its strategy, or just claiming to. I used it with one of North America’s largest universities in 60 minutes and they came out of it with a completely different model for retaining graduates as lifelong learners. The framework has since evolved into something more comprehensive, what I now call HUMAND, but the organising principle has remained constant: the question of what humans do best, what machines do best, and what AI does best is not settled by default. It has to be decided deliberately.

My argument in 2019 was that most organisations weren’t deciding deliberately. They were letting technology make the decision for them. I said the future belonged to the organisations that reversed the sequence.

I still say that. The problem is what organisations heard instead.

What they heard

Some organisations did genuinely try to put human need first. They redesigned products around user experience. They hired customer success teams. They built genuine feedback loops between what customers said and what strategy delivered. Some of it worked well.

But there was another interpretation that emerged as AI investment accelerated, particularly across 2023, 2024, and into 2025. It went something like this: if the future belongs to AI, then the highest-value humans in an organisation are the ones closest to the technology. Those furthest from the technology are, by implication, furthest from the future.

Who was furthest from the technology? The management layer.

Not the C-suite who commissioned the AI strategy. Not the technical teams building and deploying it. The layer in the middle: the people managing teams, coaching individual contributors, navigating the gap between what the board had decided and what was actually happening at the frontline. The operations managers. The team leaders. The general managers of function. The people most organisations file under “middle management.”

The case that got made against this layer, quietly, in budget reviews and efficiency presentations across those two years, was seductive: AI can do what these people do. It can produce reports faster, synthesise data more efficiently, surface performance patterns that a manager reviewing monthly results would take hours to find. The headcount cost of maintaining the management layer could be redeployed toward AI capability. The maths looked clean.

The problem is that the maths were wrong. Not the numbers. The model. Because what that case assumed was that the job of a middle manager was fundamentally information processing. And the job of a middle manager is not fundamentally information processing. It never was.

The data arriving now

I’m not writing from inference here. The evidence is arriving now from multiple directions, and it’s worth being precise about it.

Gartner, in its July 2026 analysis, found that 20% of organisations have already committed to eliminating 50% or more of their middle management roles by the end of this year. That’s not a forecast. It’s a programme already in execution.

The DDI Global Leadership Forecast, also published this month, found that 71% of leaders report high stress from the pace of organisational change. Stress is manageable. What makes this finding consequential is the number that sits alongside it: 40% of those stressed leaders are actively considering leaving. Not disengaged. Actively considering exit.

Forbes named the third-order consequence in June, directly: “Management Cuts Today Are Shaping A Leadership Shortage.” Not a risk. A shape already forming.

And in Australia specifically, Microsoft’s 2026 research shows that only 28% of Australian businesses have an aligned AI and human strategy. The other 72% are operating in a gap, AI capability moving faster than the human infrastructure built to work alongside it, to govern it, to check it against context.

These aren’t four separate stories. They’re the same story, told from four different vantage points. And the story is: we removed a human layer and called it modernisation, and the consequences of that removal are arriving now.

Who those managers actually were

This is where I want to be specific, because the abstraction of “middle management” does real damage to this conversation. It makes it too easy to assume that what was removed was bureaucracy, unnecessary hierarchy that stood between strategy and execution. Some of it was. But a lot of it wasn’t.

The people in this layer, across the organisations I’ve worked with over 30 years, were doing things that don’t appear in any AI capability roadmap.

They were coaching. Not performance review conversations. Actual coaching. The ones that happen informally, in the last five minutes of a one-on-one when someone finally surfaces what’s been sitting beneath every previous meeting. The manager who noticed that the high performer has gone quiet for three weeks and made the call to ask, directly. AI doesn’t have hallways. AI doesn’t notice quiet.

They were translating. Good strategy, even genuinely good strategy, arrives at the frontline as noise until someone with judgment and context turns it into meaning for this specific team, in this specific environment, with these specific people under these specific pressures. That translation work requires knowing both sides of the conversation. It can’t be done by a system that hasn’t met the team.

They were holding culture. Not the formal culture, not values statements and leadership frameworks and HR programmes. The informal culture. The unwritten understanding of what leadership actually means in this organisation, which is almost always different from what’s described in any document. This knowledge is embodied, relational, experiential. It doesn’t sit in a database. It exists in the people who have been present through multiple cycles of change and who carry the institutional memory of why certain things are done the way they are.

And they were absorbing. Change lands hardest on people in the middle. When a strategic shift is announced, the C-suite has context for why it’s happening. The frontline has a task to do. The management layer absorbs the organisational anxiety from both directions, holding the space between “we’ve decided” and “we’ve landed.” When that layer goes, that absorption work doesn’t disappear. It becomes unmanaged pressure that moves in other directions.

The organisations that cut this layer didn’t remove bureaucracy. They removed the human infrastructure of organisational function, and then replaced it with nothing, because what’s being replaced isn’t a system. It’s a capacity.

The ripple effects nobody modelled

Here’s what I keep coming back to when I look at the DDI data.

If 40% of the remaining leaders are considering leaving, what’s driving that? The surface answer is stress. But the deeper answer is that they’re being asked to do the work of leadership in an environment where the infrastructure that used to support that work has been removed.

The managers who weren’t cut are now doing more of the coaching, translation, and cultural holding, while simultaneously being expected to govern AI tools that have been introduced faster than anyone has thought clearly about how to integrate them. They’re doing two jobs. In a context with fewer colleagues who understand the history of the place. With less time to develop the people under them, because the development layer above them was removed and they absorbed that work too. And they’re doing it all at a pace that leaves no room to recover between cycles.

They’re making a calculation: is this inhabitable?

For many of them, the answer is no.

This is a Ripple Effect in the precise sense I use that term. The first-order decision was: cut the management layer, realise the cost saving, redeploy to AI capability. The second-order effect (the one now visible in the DDI numbers) is remaining leaders overwhelmed, coaching and translation work unmanaged, culture becoming harder to hold as institutional knowledge walks out the door. The third-order effect, still arriving, is a leadership succession void. The people who were supposed to develop into the next generation of senior leaders, who were supposed to learn by doing the work under more experienced managers, don’t have that developmental path anymore.

I wrote earlier this year about how AI has removed the entry-level learning pipeline, the years of junior work that used to build judgment and professional character before people took on real responsibility. That piece was about the bottom of the pipeline. This one is about the middle. And together they describe something that organisations haven’t been willing to name directly: the leadership pipeline has been compressed at both ends simultaneously.

The question that follows is one most organisations haven’t sat with long enough. Who will make the high-stakes judgment calls in five years? Who will have developed the contextual intelligence, the relational authority, and the cultural knowledge required to govern AI decisions at the level that matters? If you’ve removed both the entry-level learning path and the mid-level development layer, on what timeline do those capacities regenerate? And what happens in the gap?

I don’t ask that to be alarming. I ask it because the organisations with honest answers to that question have a preparation advantage that compounds year on year.

The 2019 argument, what I missed

I want to say this plainly: I wasn’t wrong about the direction. Human First was right then and it’s right now.

What I didn’t say clearly enough, or perhaps what didn’t need saying yet because the situation hadn’t arrived, is that Human First isn’t just about the external customer. It’s about the internal organisational structure required to serve that customer over time.

You can say Human First while simultaneously eliminating the humans who were supposed to make that commitment real inside the organisation. You can write it on a values board while cutting the management layer that was supposed to embody it in day-to-day decisions. You can build the customer-facing product experience around human need while treating the internal human infrastructure as overhead to be optimised.

What that produces isn’t Human First. It’s Human First at the margin and Human Last at the core.

The organisations I’m most concerned about right now are the ones that feel genuinely efficient. They’ve cut costs. They’ve deployed AI across multiple functions. Their dashboards look good. But underneath that efficiency, the human capacity to govern those AI decisions, to develop the next generation of leaders, to absorb organisational anxiety during change, that capacity has been removed without any model for what replaces it.

The confidence-preparedness gap is real and it’s widening. Organisations can feel more confident than they are. The management layer compression is one of the mechanisms widening that gap right now.

What an Inhabitable Future requires structurally

I spend a significant amount of time in keynotes and workshops asking this question: what makes a future Inhabitable?

Not just possible. Not just efficient. Actually liveable, trustworthy, and functional for the humans who have to work and lead inside it.

Why leadership feels heavier in 2026 isn’t mysterious. It’s the accumulated weight of structural decisions that have made the leadership environment less supported, less contextually rich, and more isolated. More AI capability, less human infrastructure. Higher velocity, fewer guides. Greater accountability, thinner development.

An Inhabitable Future for an organisation requires a human layer that can do what AI can’t. And what AI can’t do is heavily concentrated in exactly the capacities the management layer was providing.

The ability to notice a human being in distress and respond to it, rather than flag it in a workflow.

The ability to translate between the language of strategy and the reality of a team under pressure.

The ability to hold the history of why an organisation is the way it is, and to know when that history should inform a current decision and when it should be overridden.

The ability to develop people, not through a learning management system, but through the slow, informal, relational accumulation of challenge and feedback and trust over time.

None of this can be automated away without something being lost. And what’s lost doesn’t announce itself immediately. It shows up in the numbers 18 months later, when 71% of your remaining leaders are stressed and 40% are eyeing the door.

The organisations eliminating the management layer aren’t building Inhabitable Futures. They’re building efficient ones. Those are different things, and they have different durability.

What to do with this

I want to be careful here, because this isn’t an argument against AI adoption or against restructuring management hierarchies. Some of what was removed needed to go. Real inefficiencies existed in real organisations, and not every person with a management title was doing genuine leadership work.

The problem isn’t the removal of unnecessary hierarchy. It’s the removal of necessary human capacity under the guise of hierarchy removal, and the failure to notice the difference.

So here’s what I think this moment actually requires.

Organisations that have already made significant cuts to the management layer need to audit what was lost, not just what was saved. Not a headcount audit. A capacity audit. Which specific coaching, translation, and cultural holding functions were being performed by the people who left? Where are those functions now? Who’s doing them, or is anyone? That audit is uncomfortable, because it may reveal that the efficiency model had hidden costs that aren’t yet visible in the P&L. But the answer is still better known now than in 2028 when it arrives as a succession crisis.

The 40% of leaders considering leaving are a signal worth investigating properly. Not with another engagement survey. With direct, honest conversations about what’s become uninhabitable about their role, and what it would take to change that. The answers will point directly at what was removed without replacement.

Leadership development needs to become deliberate in a way it hasn’t been at most organisations. Leadership under uncertainty is a capacity that takes years to build, and it requires conditions that are now rarer: experience of being led well, time for reflection, exposure to senior judgment, gradual escalation of accountability. These conditions don’t emerge from AI tools. They emerge from organisational environments that treat human development as strategic infrastructure, not as a nice-to-have when the budget allows.

And for Australian organisations specifically, the 28% alignment figure is a starting point for an honest question: if 72% of businesses haven’t aligned their AI and human strategy, what is AI actually aligned with? The answer, in most cases, is efficiency. Which is a legitimate goal, but it’s not a sufficient one.

Efficiency without the human infrastructure to govern it, check it, contextualise it, and develop people within it: that’s not Human First. That’s Human Eventually, if there are any humans left who know how.

Choose Forward.