Training is a beginning. Not a destination

“You come past on LinkedIn a lot and you talk about almost nothing but AI.”

I hear that regularly. And I take it as a compliment, because I am genuinely enthusiastic about it.

By now I have given a fair number of Copilot training sessions. And still I notice: it does not always land the way I would like.

Everyone is enthusiastic. But a few weeks later everything looks like it did before.

And that is a waste.

It is just like driving

Because training is a beginning. Not a destination.

Your lessons teach you to operate the car. But you learn to actually drive on the road. By covering distance. By anticipating. By making mistakes and doing it differently next time.

And a few cars later you still sometimes use that one tip from your instructor. But the rest is all experience.

AI works the same way. One session does not get you there. You have to try things, cover some distance, and keep the goal in view. That last one matters most.

Why it slides back after three weeks

I have seen enough of this by now to venture something about it, and it is almost never the content of the training.

It is that nothing changes about the work itself afterwards.

Somebody learns in two hours what is possible, and then returns to a day arranged exactly as it was the week before. Same meetings, same deadlines, same pile. And the new way costs more time the first five times than the old one, because you are learning.

So you do it once, it takes longer, and the second time you take the old route. Not out of laziness. Out of time pressure.

What does work, and it is a boring answer: one task. One concrete, recurring task that the whole department knows is being done differently from now on. Not “go and experiment”, because that is not an instruction and therefore not a priority. But: the monthly report is done this way from now on, and in four weeks we look at whether it got better.

That is small. It is also the only thing I consistently see stick.

The goal is not Copilot

And then the part I say in every session, which I may as well write down here.

Copilot helps you on your way. The goal is what you want to achieve: working faster, deciding more sharply, more room for what actually matters.

That sounds like a marketing line, but it is a practical one. Because if the goal is the tool, then you are finished the moment everybody has a licence. And then nothing has changed except your invoice.

If the goal is that the monthly report no longer costs a Friday afternoon, then you know exactly when you are done. And you can explain it to somebody who has no appetite for AI.

That is, I think, the most honest answer to why I talk about almost nothing but AI.

Not because the technology is so beautiful, although it is.

But because it is the first time in my career that the tooling changes faster than the work. And that gap is exactly where it gets interesting, and where it goes wrong.