Recently, a new tool was launched on the engineering firm Arcadis careers site, allowing users to upload their resumes and receive recommendations for suitable vacancies. This week, another similar innovation was introduced.
Now, energy company Zonneplan is launching an A.I. tool on their new job site. And here, you don’t have to enter a resume; you need to paste a LinkedIn profile into a text field, after which the algorithm (named KIKI AI) promises to do the rest and present you with suitable vacancies.
‘Scrolling through vacancies for hours, looking for a job that suits you? At Zonneplan, that’s no longer necessary,’ says recruitment manager Eldert Wijkstra enthusiastically. ‘The chances of applying for a job that truly fits you are much higher thanks to this A.I.’ Kiki, by the way, doesn’t just work based on a LinkedIn profile; you can also chat with her. According to Wijkstra, the language model can understand connections within texts. ‘This means Kiki can also analyze which positions are suitable based on personal characteristics.’
‘Looks beyond simple keywords’
Kiki looks far beyond simple keywords, adds Dennis Stolmeijer. ‘Instead of drawing from a traditional text database, the content from our CMS is translated into a vector database,’ says the Head of Development at Zonneplan. ‘This means that the data is converted into a format that can be directly understood by a language model, allowing the A.I. to deliver much more accurate and relevant results.’ In other words, he promises Kiki to look beyond just the literal job description.
‘Our method ensures that search results are not only based on simple keywords, but also on context and nuance.’
The A.I. assistant uses ‘ prompt stuffing ‘ to extract only the relevant vacancies from the database. Stolmeijer: ‘This method ensures that search results are not only based on simple keywords but also on context and nuance, which leads to much more accurate and personalized job matching. This way, job seekers can find the right job at Zonneplan more quickly and effectively.’ The intention is that it will also be possible to ask the chatbot questions beyond vacancies later ‘so the applicant can better discover if Zonneplan is a good fit for them.’
Putting it to the test
Promising, indeed, but let’s also put it to the test. Does this add something to the candidate’s experience? The first personal experience may be bad luck, but unfortunately, it did not start very promising. Heading to LinkedIn to retrieve your profile URL feels a bit cumbersome, but when the tool spins for minutes to develop suitable vacancies, it quickly raises questions. Could it be due to my profile? Am I not a good fit for Zonneplan? I’m left in the dark about it. In the meantime, I’ve manually scrolled through the vacancies that are less than 40 on the site.
The waiting does allow me to read the quirky text snippets by the tool. ‘The job train is on its way,’ ‘Donut worry, jobs are coming,’ and ‘Sherlock Holmes is looking for’ ‘watermelon-fresh jobs’ and ‘vacancies as refreshing as ice cream’; there’s no lack of wit. It does make the waiting more pleasant. Only ‘millions of options are being analyzed’ seems a bit exaggerated. However, on the second attempt, I didn’t experience that anymore: the matching now goes very quickly, and miraculously, the algorithm comes up with two more or less suitable vacancies. Great!
A Trend Born?
The big question, however, is: wouldn’t I have found these vacancies myself if I had scrolled through the list? It seems unlikely. Is the problem of manual scrolling here so significant that this matching tool offers a solution? To be completely honest, of course not. That these two vacancies have a specific link to my profile (and many other vacancies on the site do not) is something I could have figured out myself. It would be more interesting if unexpected vacancies appeared for me and, for example, matched more with other parts of my profile.
It is, of course, notable that in such a short time two major job sites have come up with a tool of roughly the same scope.
Just like with the tool from Arcadis, the conclusion (for now) can unfortunately only be an excellent idea, but it will probably not make the lives of most job seekers much more accessible (yet). Perhaps that will follow if the chatbot learns to have conversations beyond vacancies, but the input of a LinkedIn profile does not seem necessary. However, it is notable that in such a short time, two major job sites have developed a tool of roughly the same scope, trying to make matching easier early. Could this mean a trend is born?
Want to Know More?
Do you want to know which organizations are top regarding online candidate experience? On October 15, Digitaal-Werven will present the award for the best corporate career website for the 18th time. You can register here:
`
