Skip to content
Guide

AI recruitment automation: what works and what doesn't

Everyone is talking about AI in recruitment, but few agencies actually get it running. The tools are there, the promises are big, and yet little changes day to day. This guide explains where the time really leaks in your process, what you should and shouldn't automate, and what a project looks like when you take it seriously. Written from practice, not from a product demo.

By Martin Veltman

Where the time leaks in recruitment

Most agencies are sitting on a goldmine of ATS data that nobody touches anymore. Candidates from years of work, thousands of CVs, conversations and contact moments. But the moment someone doesn't fit one vacancy, you never hear about them again. A new assignment comes in, and a recruiter starts from scratch on LinkedIn, while the right person is often already in your own database.

On top of that, recruiters write the same outreach a hundred times over. Screening CVs and submitting candidates is the work that takes the most hours and gets noticed the least. And between an intake call and an actual submittal there is a pile of admin: writing up the profile, building a Boolean, typing out notes, updating the ATS. None of that is recruitment, yet it eats half the day. Those are the places where recruitment automation pays off.

What to automate, and what not to

The rule of thumb in recruitment automation is simple: automate the repetitive work, not the judgement and not the relationship. AI systems are good at preparing, sorting and summarising. They are not good at having the conversation, weighing the edge cases, or building trust with a candidate. That is human work, and it should stay human work.

Do automate: scanning your database on every new vacancy, drafting a first personal outreach message, a first CV screening against your own criteria, and the admin after a call. Do not automate: the decision whether to submit someone, the intro conversation, and the tone you use with a client. The AI delivers a reasoned draft; you approve it or adjust it. That way you keep the speed and your own judgement.

How a project works

AI tools promise a lot but often sit unused because nobody sets them up. We set them up and we keep maintaining them. A project starts with a 30-minute diagnostic call where we map out where the most time is lost in your process. Then a two-week pilot for a fixed price agreed up front, so you know where you stand. The systems run in your own environment, inside your existing ATS and tooling; recruiters do not have to learn a new platform. After that, a fixed monthly fee for ongoing maintenance and optimisation.

Where you start depends on your biggest bottleneck. If your goldmine is full of old candidates, Database activation is the logical start. If submitting candidates takes the most hours, we look at Submittals. If the admin between intake and submittal is what slows you down, Production fits better. One system first, expand as it proves itself.

What it delivers

The goal of recruitment automation is not AI because we can, it is time back for the work that matters. At Forta Solutions we trained a CV checker on 22 years of Linda's own criteria. The CV check went from 30 minutes per candidate to 0, and Linda saves around 8 hours a week. That time now goes to personal conversations and placements, not to screening. At Match Masters we built a dashboard for the step before the ATS, saving Steijn an average of two hours a day on admin.

The gain is not only in hours. Because the first screening and the submittals become more consistent, your agency looks sharper, regardless of who is at the keyboard or how busy the day is. The recruiters get time back for the human work: candidate contact and the client relationship.

Does this also work for staffing agencies?

Yes. The logic is the same; only the dial you turn is different. In permanent recruitment it's about the speed and quality of your match: surfacing the right candidate sooner, including the people who have sat in your ATS for years. In staffing it's about fill rate: connecting your available and dormant flex pool to incoming requests as fast as possible. Whoever puts a good candidate forward first wins the assignment.

So for a staffing agency, automating recruitment mostly means keeping your flex pool alive. The system sees who becomes available, who hasn't worked in a while, and who fits a new request, and lines up the first approach. It connects to your staffing software (Easyflex, Carerix, Bullhorn) just as well as to an ATS. The rest of the project is identical: start small, prove it in two weeks, expand as it works.

AI in recruitment, without the scaremongering

Your recruiters keep doing the work; the AI handles the prep. A system that prepares outreach or pre-sorts CVs takes the dull part off your hands, not the craft. The agencies stepping into this now are not doing it to automate people away, but to give their best recruiters more time for the part only a human can do. That is the whole point: less typing, more attention.

Where is the most time lost in your process?

Book a no-obligation 30-minute call. No pitch, we figure it out and send you a concrete proposal two days later.