A McKinsey report caught my eye this week.

The headline was simple: AI is everywhere, but the “agentic organisation” isn’t—yet.

And it landed because it captures exactly where we are.

AI has moved from hype to default.
Every boardroom conversation, every strategy deck, every founder pitch.

And yet… very little has actually changed where it matters.

Most companies still aren’t seeing a meaningful bottom-line impact from AI.  

That’s not a failure of the technology.

It’s a failure of how organisations are using it.

What’s happening in practice is fairly predictable.

AI is being layered on top of existing ways of working.

A bit faster here.
A bit more efficient there.
Slightly better outputs.

But the underlying system hasn’t changed.

And that’s the problem.

Because the real value doesn’t come from making the current model incrementally better.
It comes from redesigning the model entirely.

Companies that are starting to gain traction are thinking differently.

They’re not asking how AI improves a task.

They’re asking what the workflow would look like if you rebuilt it from scratch.

That’s a much harder question.

But it’s the only one that really matters.

Most workflows today are fragmented across teams, full of handoffs, and dependent on coordination and interpretation. AI—particularly agentic AI—starts to break that structure.

Not by optimising individual steps.

But by connecting the whole thing.

And when you do that, you move from efficiency gains to something more meaningful.

You start to change how the business actually works.

The part that’s being underestimated is what this means for people.

Most roles aren’t disappearing.

But they are changing—quite materially.

Roughly three-quarters of roles will need to be reshaped in some way.  

That creates a capability issue, as well as a mindset one.

Because the job is shifting.

Less doing.
More judging.

Less execution.
More oversight.

We’re moving from people being in the loop to being above it.

And that sounds simple, but it’s a very different way of operating.

Where this gets uncomfortable is in leadership.

Because this isn’t something you can delegate to a transformation team or an IT function.

It requires leaders to rethink how they spend their time, how decisions get made, and how teams are structured.

Most organisations have added layers over the last decade.

AI has the potential to remove them.

That’s faster and more efficient, but also disruptive.

And that’s why progress is slower than expected.

Not because the tech isn’t ready.

But because the organisation isn’t.

The companies making real progress consistently do a few things.

They’re focusing on a small number of end-to-end use cases.

They’re embedding capability in the business, not isolating it in tech teams.

They’re thinking about systems, not tools.

And they’re investing heavily in people—because without that, nothing else sticks.

The bigger point here is this:

This isn’t really an AI story.

It’s an organisational one.

AI exposes how inefficient, fragmented, and slow most businesses already are.

And it creates the opportunity to fix that.

But only if you’re willing to change more than just the tools.

The gap between expectation and reality will close.

But not because the technology improves.

It will close when organisations finally adapt to it.

Until then, most of the value will remain theoretical.

And a small number of companies will quietly start to pull away.

Have a great week!