I work in a company in a rough economic patch. For a while now it’s been clear: you only keep up with AI. And it’s seen, above all, as a lever for cost and efficiency. People are being let go. Whether that’s down to the fuel crisis, or whether it’s almost a welcome excuse to move out the ones who aren’t all that AI-savvy — who knows.
In our marketing team it’s about fear too. Everyone clings to Claude Code and automates everything that isn’t nailed down. The interesting part: my colleagues don’t stay in their lane and quietly optimise their own patch first. They look into the neighbouring fields and poach.
And there I feel a resistance. When my young colleague produces not just the content for a blog but the content strategy to go with it — which really should come from me — and now does it with Claude Code. Well, he can, can’t he? I’d probably do the conceptual work with Claude Code too. For now, I think, the difference is still in what I put in at the front and what comes out the back.
The blind test
My strategy rests on experience and a feel for things. Something another person might not have. But run a blind test between his strategy and mine, hand it to the Head of Marketing or the CEO — and I reckon it makes no real difference anymore. His reads just as nicely. Only with less thought behind it.
The decision-makers usually don’t know all that much themselves and would rather go with the obvious. And the more generic version is easier and faster to grasp. So the content strategy is gone. Unless there’s one person with clear responsibility, who sets the frame, builds the skills and guidelines that then feed back into Claude Code so everyone has access. Otherwise that role is dead.
That holds for all the deep experts. Anyone packing their knowledge into documents that translate neatly into workflows will notice that the value they’ve banked isn’t valuable enough anymore. So they start wondering how much they should document at all. Or whether to keep a shadow system they can take with them — because that knowledge isn’t sitting in the LLMs yet.
Because it’s not about the knowledge. It’s about applying it, taking the right path. Even the AI experts will soon stop sharing their tricks, their skills, their hard-built agents quite so freely. Because the others just build on top, optimise further — and then you’re out.
More hated than the controllers
All the time the AI experts have poured into this over the last two or three years can be devoured within weeks by people who know far less but handle the tools well enough. So what does that mean for us? Whoever isn’t working with AI yet will catch up faster than they think. Eighty per cent is enough. But whoever’s been at it a long time, and sunk in money and hours, has to keep moving.
I can picture the good AI people drifting from company to company, automating everything that can be automated in six months flat. They’ll be even more hated than the controllers. Because together with the CEO and the controller they make the company competitive — and at the same time rip out its heart and soul. Because people get let go who matter to the culture, simply because they’re there. And now a cold, metric view just erases them.
Interesting times.
And the ones only climbing in now, going zero to a hundred like a rocket — which really is very easy — will realise they’re only executors. And once the boss realises it too, he’ll quickly think: why am I paying all these people in front of screens, when a fraction of that would do, to not have them there at all? It’s a narrow ridge to walk. Because when something actually happens and you need a creative spark, you won’t get one — you’d have to go and ask the models for it first.
So the problem with artificial intelligence isn’t unemployment. It’s de-skilling. Because the ones coming up now have to do everything fast and will never have the time or the focus to actually learn anything. They just feed it all into Claude or Gemini.
First published in German at reinergaertner.de, where I’ve been at it since 1997. AI did the heavy lifting on the translation. I did the heavy squinting at the result.