The Transformation Nobody Announced

The largest change most organisations are living through right now was never announced, never scoped, and has no owner. It’s happening anyway, from the bottom up, while the people at the top work out what to say about it.

A strapline that was really an intent

A while ago I interviewed Sarah McGovern, Director of Transformation at Manchester University NHS Foundation Trust, about the Hive programme: replacing more than 80 clinical systems and a great deal of paper with a single electronic patient record, across ten hospitals and 30,000 staff, most of it delivered during a pandemic. It’s still the most complex programme I’ve heard described first-hand.

What stayed with me wasn’t the scale. It was one sentence. Right from the beginning, Sarah told me, they were clear as an organisation that this wasn’t a technology programme. It was a transformation programme, clinically led, operationally delivered, digitally enabled. That was the strapline, and it went everywhere.

Read it again as an engineer rather than a communicator and you’ll see what it actually is. It names what matters and who owns it, clinical care and the clinicians, and it puts the technology firmly in the position of serving that. It’s an intent, stated so plainly that thirty thousand people could carry it around in their heads. Everything downstream, the hundred-plus decision groups, the peer-to-peer training, the go-live command structure, was execution in service of a purpose the organisation had already agreed.

Sarah gave a second reason for insisting on it, and it’s the one worth pinning up. If people think it’s a tech thing, she said, they see it as somebody else’s job.

Now look at what we’re doing with AI

Set that beside how AI is actually entering most organisations, and the contrast is uncomfortable.

There’s usually no programme. No strapline. Often no clear statement from anyone senior about what this is for, beyond a broad enthusiasm for not being left behind. What there is, in almost every organisation I see, is a large number of individuals quietly getting on with it: using whatever tool they’ve found, in whatever way seems to help, with no real guidance about what good looks like, what’s off-limits, or what any of it is meant to add up to.

It’s the largest change to knowledge work in decades, and it’s arriving without an announcement.

I don’t think that’s negligence, and I want to be fair about why it’s happening. Leaders are being asked to set direction on a technology few of them understand deeply, that their own experts disagree about, and that changes materially every few months. Committing publicly to a shape for it feels like promising the weather. So the safest-looking move is to wait until the picture clarifies, and encourage a bit of experimentation in the meantime.

The trouble is that waiting isn’t neutral. While the organisation waits, the adoption happens anyway, just without it.

I gave this a name once

I’ve watched a smaller version of this before, and I even wrote about it. Back in 2023 I described what I called an IT mutiny: an IT department adopting Agile on its own initiative, without the rest of the organisation coming with it.

The thing I kept insisting on then still holds. The mutineers weren’t malicious, and they weren’t rebelling. They were usually the most insightful people in the building: natural problem-solvers, plugged into every other department, hearing all the grumbles, watching how the best digital organisations worked and wanting some of that for their colleagues. They adopted because they could see a better way and got tired of waiting for permission that was never coming.

The damage came anyway. Their release cadence changed, so operational teams had to change too. Governance and change boards buckled. Reporting no longer fitted. Contracts didn’t match how work was now being done. Other departments ended up running two operating models at once, the old one and the imposed new one, and the friction landed on people who had never agreed to any of it. Eventually the best people, the ones with options, started to leave, and their tacit knowledge left with them.

Three years on, that reads like a rehearsal.

This time it’s everywhere, and you can’t see it

What’s happening with AI is the same pattern with two differences, and both make it harder.

The first is scale. An Agile mutiny was one department. This one isn’t confined to IT, or to any department. It’s every individual with a browser, in finance, in legal, in HR, in the clinical service, all adopting at once, at different speeds, to different standards, for different reasons. There’s no single mutinous crew to sit down with. There are thousands of small, sensible, private decisions accumulating into something nobody chose.

The second is visibility, and it’s the one that really worries me. The Agile mutiny announced itself. It changed the interfaces between departments, the release cadence, the governance gates, the contracts, so it showed up as friction someone could point at. Every warning sign I listed in 2023 depended on that. You could spot a mutiny because it made a noise.

This one makes no noise at all. When someone uses a model to draft the paper, sift the responses, or shape the analysis, nothing external changes. The document arrives on time and looks much as it always did. There’s no Change Advisory Board for “I wrote this with a chatbot”. The work is being quietly reshaped inside individual heads and individual afternoons, which is precisely where an organisation has no visibility and no controls.

Which is why the familiar change-management frame keeps missing. Almost everything most of us learned assumes the problem is people not wanting to change, and builds machinery for overcoming resistance. That machinery is aimed at the wrong target. People aren’t resisting; they’re adopting faster than their organisations can form a view. The enthusiasm isn’t the problem. The problem is that it has nothing to align to. That isn’t a resistance problem. It’s an intent vacuum, and pouring more change communications into it doesn’t touch the cause.

You can’t run it like Hive, and that’s the point

The obvious lesson from a programme like Hive is: do that. Plan it properly, brand it, resource it, appoint an executive owner, go live on a date.

I don’t think that lesson transfers, and it’s worth being honest about why. Hive had things AI adoption doesn’t. A known technology with a defined scope. A clear end state you could describe on day one. A date after which the old world was switched off. You can command-and-control a transformation like that because you can see its edges.

AI has none of those properties. There’s no go-live, no fixed scope, no end state, and the capability underneath keeps moving while you deliberate. Anyone waiting for it to hold still long enough to plan around will be waiting a long time.

But the thing that made Hive work wasn’t the plan. It was that the organisation knew what the whole exercise was for, in terms that had nothing to do with the technology, and said so relentlessly, in language a nurse could act on. That part transfers completely. And here’s what I find genuinely clarifying: when you can’t specify the execution, intent is the only stable thing you have. The less certain the technology, the more the direction has to come from what you’re trying to achieve, because that’s the one thing not being rewritten every quarter. Uncertainty doesn’t excuse you from governing intent. It’s the very condition that makes it the only available form of control.

So the useful question in most organisations isn’t “what’s our AI strategy”. It’s the plainer one Hive answered: what is this actually for, who owns that, and what’s the technology here to serve?

Intent has to survive the journey

Setting the intent is necessary and nowhere near sufficient, because intent degrades as it travels.

I’ve written about this on the machine side, where the gap between what you meant and what the model reconstructed is the central discipline. The same gap opens between people, and it widens with every layer. A board holds a clear purpose. It becomes a slide. The slide becomes a directive to use the tool. By the time it reaches the person doing the work, the why has fallen away and what’s left is a mandate with no meaning attached, which is exactly the “somebody else’s job” Sarah was guarding against.

Look again at what Hive actually spent its effort on and most of it was intent fidelity. Over a hundred decision groups, so the reasoning happened where the knowledge was. Roadshows, drop-ins, posters on noticeboards, because busy clinicians don’t read email. Peer-to-peer training, nurses teaching nurses, so the message arrived from someone who understood the work. None of that was communication for its own sake. It was machinery for getting a purpose to arrive intact at the far end of a very large organisation.

The first thing to go

There’s one more cost to the unannounced transformation, and it’s the one this series keeps circling.

When nobody states the intent, everything defaults to whatever is loudest, and right now that’s the technology. Attention goes to the tools, the pilots, the evaluations, the things that visibly need someone. The quieter work, growing the people who’ll have to run all this in five years, has no deadline attached and no vendor asking after it, so it slips. Hive did the opposite: it put its development effort into peers teaching peers, and treated readiness as something you build in people rather than install in a system.

And I’ve seen where the other road ends. In the mutinies I wrote about three years ago, the ending was always the same: the best people left, and the knowledge nobody had written down left with them.

The transformation is happening either way. The only real choice is whether anyone governs what it’s for, or whether we let thirty thousand private decisions decide it for us.

Next Article ->