Anyone who hears about the promise of Textmetrics may quickly think: ‘Ah, here we go again with another generative A.I. tool – a kind of ChatGPT for recruiters.’ Marcel Leeman, one of the initiators behind the product, understands where that misconception comes from. But his software is much more than an LLM – short for Large Language Model – he immediately adds. ‘Our product is to generative A.I. what an app is to a smartphone.’
Generative A.I. has limits

‘Naturally, you do not want to lie in your job advertisement.’
But unfortunately, he immediately qualifies this. ‘Things are not that simple. The quality of the job advertisement written by a tool such as ChatGPT depends heavily on your prompt – the writing instruction you give it. An incorrect prompt produces an incorrect result, and even if your prompt is right, you may still be disappointed. ChatGPT always produces a result, after all, but you can never be sure whether it has actually followed your prompt : asking for a B1-level text does not necessarily mean you will get one. And it can get worse: many A.I. tools hallucinate, meaning they make untrue claims. Annoying, because naturally you do not want to lie in your job advertisement.’
Textmetrics pushes the boundaries
Users of Textmetrics do not face those risks, Leeman knows – the software his company offers is much more than generative A.I. How does that work? Textmetrics not only writes job advertisements, it also checks them itself. For example, the software examines whether a newly generated (or existing) text is sufficiently inclusive. In other words: does everyone feel addressed, or is your current message only appealing to a specific group?
‘Our product ensures that an LLM generates a better text.’
‘The algorithms of Textmetrics generate job advertisements using an LLM, and then check them,’ Leeman clarifies once again. ‘In other words: our product ensures that an LLM generates a better text. To be fair: that LLM is not ours, as developing such a product is only possible for extraordinarily large organisations with extraordinarily large budgets – think Microsoft. But you can use the LLM in our interface, so if you think our software does not generate texts, you are doing us a disservice. That is why I compare it to apps on a smartphone. Thanks to our software, you can use an existing function in a way that genuinely adds something.’
A.I. governance as a safety net
Just as every app developer closely monitors the latest iOS and Android updates, Leeman watches what is happening in the market for generative A.I. with great interest. ‘What I find interesting at the moment is seeing how tools such as ChatGPT are entering companies,’ he says. ‘This does not happen through the usual process, where an organisation encounters a problem, convenes a committee to solve it, and eventually comes up with a solution. No, generative A.I. is entering organisations from the bottom up. ‘Ordinary’ office workers start having ChatGPT write their reports, without it ever having been discussed in a meeting.’
‘Ordinary office workers start having ChatGPT write their reports, without it ever having been discussed in a meeting.’
In itself, that is of course ‘fantastic’, continues the Textmetrics CEO. ‘I am convinced that generative A.I. can add a great deal of value. But as a company, you also need to be careful with it. After all, you do not have a contract with the supplier of an LLM, so if ChatGPT becomes a crucial part of your operations without you realising it, the consequences will be disastrous if that platform suddenly makes off with your intellectual property or receives negative publicity. It is therefore important to invest heavily in A.I. governance. That means incorporating the use of A.I. into your processes from the top down, and ensuring you have a safety net if a tool suddenly no longer does what you want.’
Read more?
Curious about how Textmetrics improves job advertisements? Kyrill Poelmans, co-founder of the initiative, explains it in this article.
Want to know more?
On Thursday 29 June at 11.00, Marcel Leeman will give a webinar entitled: ‘Never write job advertisements yourself again thanks to A.I.’. He will discuss, for example, automatic translations, content personalisation, smart chatbots and even guiding writing processes through ChatGPT. He will discuss the future of A.I. text generation and everything involved, such as: is generative A.I. safe? Does training it not take just as much effort as creating content yourself? How do we ensure that we do not lose control of these types of A.I.? Register here:
Read this article in full
Become a Werf& member for unlimited access to all articles, research and exclusive insights into the labour market.
No obligations Cancel anytime Secure payment