THE OPEN

I keep hearing a version of the same sentence from leaders this year:

It might be a nice thing to tell themselves because it puts the problem in someone else's lap to solve but the uncomfortable possibility is the opposite. In a lot of organisations the people are ready to experiment, and it is the organisation that has not moved: the permissions, the workflows, the incentives, the way managers actually behave.

If that is true, then a slow adoption number is not a verdict on your staff. It is a verdict on the system they are working inside, which means the problem is back on your lap.

Here is the figure that made me sit up. Across the research I have been reading, only around 11% of organisations are anywhere near what you could honestly call the AI reinvention stage, the point where the business has genuinely redesigned how work happens, rather than bought everyone a licence and called it done. And in a lot of cases people did not even wait for that licence. They started using their own AI to get the work done, while the organisation carried on assuming staff would reach for the tool it rolled out, not the one already on their phone. The same research keeps finding the same thing: people are readier to use these tools than their organisations are to change around them.

And there is a harder number that puts a figure on exactly that gap. Microsoft's 2026 Work Trend Index found that organisational factors, the culture, the manager support, the talent practices, account for more than twice the impact on whether AI works that individual effort does: 67% against 32%. Read that slowly. The system around a person matters more than twice as much as the person. Only about one in five workers sits in the zone where their own skill and their organisation's readiness actually reinforce each other.

Read those facts together and the usual story flips. We have spent two years talking about employee resistance as the blocker. The more honest reading is that individual readiness has run ahead of institutional readiness, and the gap is the problem.

This is why I have started treating fear of AI as an operating-model issue rather than a personal failing. When someone hesitates, the instinct is to send them on more training, as if the hesitation were a skills gap to be coached away. Often it is not. It is a perfectly rational response to a system that has not told them whether they are allowed to use the tool on real work, whose job it changes, who owns the output, or what happens to them if the experiment goes well. You cannot coach your way out of a permissions problem. You have to redesign the permissions.

This is the part I think we keep getting the wrong way round. The organisation, not the individual, is usually the thing that has not changed. If you feel held back, you may not be behind at all. You may be ahead of a system that has not caught up with you.

There is a second cost to all this, and it is the one I would worry about most, because it does not show up until later. The same pressure that stalls adoption is doing real damage somewhere this year's numbers will not show. In the rush to make the AI business case look good this year, a lot of employers are thinning out junior work, the analysis, the first drafts, the grunt tasks, precisely because that is the work AI does cheapest. It reads as efficiency. But those junior roles were also the training ground where your future managers learned the craft. Cut them to hit a number this year and you have not removed a cost, you have borrowed against your own pipeline. Harvard Business Review captured this well with the example of a 200-person analyst programme eliminated under AI ROI pressure, and the UK government's own AI Adoption Plan lands on the same instinct, arguing early-career roles should be redesigned, not reduced, after employment in computer programming for 16 to 24-year-olds fell 44% in a single year. Talent debt is real debt. It just does not show up on this year's slide.

I wrote the longer version of this argument up as a full piece, on why the fear you see in an AI rollout is a signal about the operating model, not a weakness in the person.

Forget the training budget for a fortnight. Run this instead: the two-list audit.

Write two lists:

Here is the version I see in almost every organisation I walk into:

Your Operator’s Move: Take the biggest item on the right list (or your own list) and decide this fortnight whether to make it official or shut it down. Pretending it’s not there is the only option you can’t take.

Here is the question I would put on the table this fortnight, and it is pointed at anyone who sells change for a living:

We tell clients to redesign the roles, name where humans still matter, change the incentives, stop measuring activity and start measuring outcomes. So run your own advice back on yourself.

Take the last transformation recommendation you put in front of a client, and count, honestly, how many of them you have actually done inside your own four walls. Most of us have sold a standard of change we have never held ourselves to. The ones who close that gap first are the ones who will still be believed when they walk into the room.

Each issue I share one piece of advice from someone else worth your time. This fortnight it is Stefano Puntoni, a professor at Wharton and co-director of its Human-AI Research group, who studies why AI adoption stalls inside organisations.

His advice cuts straight to the readiness question. Most organisations, he argues, point their AI at the cheapest possible work, shaving a bit of time off tasks nobody cared about. His advice however is the opposite:

In his mind, readiness is not something you audit in your people. It is time and permission you deliberately create for them, aimed at real, ambitious work, not at trimming the busywork.

Try this: name one talented person on your team and give them protected time this fortnight to point AI at your hardest problem, not your dullest one.

Five that shaped this issue, all worth your time:

So let me put the debate to you directly. Would your own organisation pass the readiness test you set for everyone else?

Whether you sell transformation, lead it, or buy it, hit reply and tell me the one thing you advise other people to do that you have not done yourselves, and what is actually stopping you. I read every reply (and i’ll keep it confidential!)

Keep building,
Alex

P.S. This whole issue really comes down to one question: is your organisation actually ready, or just licensed up? I built a sharper version of that test on the site, the Escape the Deck readiness check. Twelve honest questions, five minutes, scored in your browser, with a straight verdict on whether your model still holds under AI and the one move to make next. Free here, no sign-up. Next issue in a fortnight.