“You must enjoy working with AI.” That’s appearing more and more in job adverts for content specialists. As if it’s just part of the package. Like “being proactive,” “a team player,” or “no 9-to-5 mentality.” Just something you should fancy doing.
The assumption underneath seems logical. Because if you’re good with language and structure, you can probably write good prompts too. I mean, you work with words, you know the audience, you understand what good content looks like, so boom, promptEen prompt is de instructie die je aan een AI-model geeft zoals bijvoorbeeld ChatGPT. Het is hoe je communiceert met het systeem: wat je vraagt, hoe je het vraagt en... Meer design is your new side dish.
That sounds logical, but it’s not quite right.
A half-truth that can become a massive stumbling block
Because domain knowledge is important, obviously. Of course it is. If you don’t know how content processes work (what steps do you take to write a blog), you can’t translate them into usable prompts either. And if you can’t judge what good output is, you won’t know whether AI is delivering something decent or just clever-sounding rubbish.
But just because you know everything about content doesn’t mean you know how to steer a system. Or how to provide context, create structure, mitigate risks, in relation to that system.
Knowing what’s good isn’t the same as knowing how to get there.
It’s a bit like saying: “Right, you’re dead customer-friendly, can you just automate the entire email flow? Cheers mate!” You might be in the ballpark, but you’re missing exactly the layer where things can go wrong.
Because a prompt is more than a writing assignment. It’s actually a design problem and it’s a system instruction, but it’s also a blueprint. You determine what AI should do, in what role, with what goal, and within what boundaries.
And you do it in such a way that someone else using that same prompt gets roughly the same quality text, even without your charming presence. At least, within a business context.
Being good at that doesn’t just require creativity. It requires design skills, functional thinking, solid structuring ability, smart testing, endless iteration and constantly working on improvement, and obviously the ability to evaluate output as well.
And yes, that’s at a painfully less sexy level than “just having a chinwag with ChatGPT,” but it is what organisations need if they want to deploy AI seriously.
Domain knowledge is your starting point, but you won’t win the race with it alone
Let me emphasise it firmly once more: you need your professional expertise. If you’re deploying AI for content production, you need to know what good content is, what a client expects, what accessible content is, and so forth. Because without that knowledge, you’re rudderless.
But with only that knowledge, you’re rudderless too. Because if you can’t translate that expertise into a clear, repeatable instruction to a language model, you’ll stay stuck in trial and error (expensive joke and wastes time, mate). Or worse still, you think it’s working when it’s only turned out well by chance, making it unrepeatable.
Prompt design requires an extra layer, and that’s the translation. Of course you can learn all that, but first you need to know that’s how it works.
AI chaos with sauce on top
As long as everyone’s just writing prompts on instinct, you get chaos. There’s no control, no insight, no check on quality or risk. And everything leans on the random skills of whoever “does something with AI.”
That’s not a strategy. That’s AI roulette, where the house always wins in the end: Rien ne va plus, the money’s no longer yours!
And when it goes wrong (say, with incorrect interpretations, incomprehensible output, or biases) it’s much harder to work out where exactly things went wrong in chaos than when you’re working in a structured way, with proper processes and knowledge.
Mature AI use is structured, repeatable, and scalable
Who determines the AI’s role? Who guards the boundaries? What are the quality criteria? Which output gets through and which has to go back in the machine?
If you’re not asking those questions, you’re not working with AI. You’re playing with AI. And that’s fine, obviously, as long as you don’t confuse it with professional deployment.
AI is absolutely changing our work, but prompt design is (also absolutely) its own expertise. One that requires more than being a language dancer. And right now, many companies still underestimate that skill. Not on the technical side, mind you, because the real engineers don’t just do AI a bit on the side, but definitely in terms of use by content specialists.
Prompting isn’t a trick, it’s a new layer in how we (sometimes and increasingly often) create content. And that layer requires expertise, specialist knowledge (both of prompting and of the content (another bracket: content here means content expertise)), and a specific way of thinking.
Prompt design is the foundation for how you deploy AI. And if you skip that foundation, don’t act surprised when everything starts wobbling.
Ik ben het maar hoor, niet stressen.
I’ve spent over 25 years working in content strategy and digital transformation, which means I’ve seen enough technology hype cycles to be skeptical and enough genuine innovation to stay curious.
Want to talk shop? Do get in touch!
