Denmark-based Relink are responsible for a groundbreaking AI that matches applicants to jobs and jobs to applicants. Ståle Husby, Relink CEO and co-founder, talks about how the recruitment business is changing – and how Relink wants to be at the forefront.
"If you think about the whole recruitment process, it’s quite astonishing how little has been done in terms of using technology to make it easier and better,” says Ståle Fredlund Husby. “As a technology company, that’s how we stumbled into this space.”
Our current offering, Marlow API – it optimizes what we call prequalification. Let’s say on average an employer receives 120 applications for a particular job. Around 15 of those will end up part of a ‘pipeline’ or process vying for the job. What we can do is dramatically reduce the work related to manually deciding which 15.
Ståle Husby, Relink CEO and co-founder
So our value proposition is: less time spent reading profiles and more time talking to candidates.
If you take a company of 3,000 people, which wants to grow by 10 per cent – you might receive 36,000 profiles. Going through prequalification manually, recruiters spend on average 60 seconds per profile. Adding this up, for this one company, they will spend four months’ work just deciding who they want to talk to.
So, we want to make these processes much more efficient.
Also read: If software eats the world Syncano feeds it
Relink’s algorithm learns more and in theory gives even better results as time goes on – as it takes into account employers’ past preferences along with other evolving data. The algorithm trains itself to determine what types of candidates are more suitable for the particular job and particular organization, taking into account factors on both the employer and employee side – including experience, education level, industry trends, and more.
On this point, Ståle and his team are very aware of the potential for conscious or unconscious human bias to find its way into the system – a tendency of a recruiter to select men over women, for example. Ståle stresses that the system is designed to mitigate against bias.
One of the reasons recruiters tend to follow existing patterns is a lack of contextual understanding of things they don’t know. Let’s say you’re hiring an engineer – you know MIT is a great school. So, you focus on people from MIT and maybe a couple of other schools that you know.
There are several schools in Eastern Europe and in Southeast Asia that are just as good, yet you would maybe disqualify or at least not favor those profiles because you have no contextual understanding of the quality of that educational background.
Our product helps to change those patterns. It takes a profile, it takes into account all the relevant available data, analyzes that data, and then throws back a score, in terms of the fit of the candidate profile to a job.
The system does not feed back pictures, not names, not nationalities, not gender, not political views – the things that might be identifiable from a cover letter, which we might refer to as incremental elements that inform biased behavior.
Relink’s first MVP for its recruitment tool was a web application. It relied on sourcing data through APIs from LinkedIn and other sites. The second MVP crawled and indexed data, allowing some basic analysis. This MVP didn't provide any new data, but a simple visualization of open data. The launched product is an API connecting with existing applicant tracking systems, allowing for proprietary data – and in-depth analysis and predictions – to be produced.
“We spent a lot of time getting to grips with the huge ecosystem that is HR,” says Ståle. I think what we did wrong in the beginning was that we tried to grasp everything; because we saw so many opportunities, we were trying to become a one-stop-shop.
A eureka moment came when we realized we had to do one thing only. Our roadmap now is to more easily enable companies to find the right people and people to find the right jobs.
The role of the recruiter or HR personnel is shifting and will shift dramatically over the next 5-10 years, I think. Regardless of Relink, this is happening.
Ståle talks about the future he envisions
Our long-term ambition is that if you’re looking for a job, you shouldn’t be looking. The whole process of finding a job – the data is out there, all of the jobs in the world are available.
If you have your profile, and you know something about yourself, you should just be able to have a three- or four-sentence conversation with a chat box, and the chat box will tell you which job you should apply for. And I think that’s going to happen.
It’s going to happen with somebody’s technology fueling that chat box – and it might as well be ours.
Find out more about Relink.