Recruitment Automation: What Actually Automates

Recruitment automation software promises to give you back the hours hiring eats. Some of that promise holds. Some of it moves work sideways or produces confident output that has to be checked anyway. This page separates the two.

Three kinds of work in hiring

Automation succeeds or fails depending on which kind it touches.

• Mechanical. Scheduling, status updates, reminders, moving records between systems. Rules are clear, mistakes are visible. Automate without hesitation.
• Judgement-heavy. Deciding whether someone can do the job, whether a gap in a CV matters, whether to take a risk on an unusual background. Automating the decision here is where things go wrong.
• Reading. Going through profiles, cross-referencing sources, working out what a candidate has actually done. Not a decision, but it consumes most of the time. This is the interesting middle.

What automates cleanly

Interview scheduling. The back-and-forth over calendars is pure overhead. Automating it is uncontroversial and saves real hours.

Status and follow-ups. Candidates dropping out because nobody replied for two weeks is a solved problem. Automated updates fix it.

Moving data between systems. Re-typing a candidate into the ATS is error-prone and pointless.

Deduplication. The same person appearing under three entries wastes attention. Machines are better at this than people.

What breaks when you automate it

Rejection by keyword. Filters that reject CVs for missing an exact phrase discard good candidates silently, and you never learn which ones. The failure is invisible, which is what makes it expensive.

A score with no reasoning. A tool that returns "87% match" and nothing else has not saved you the work: you still have to open the profile to find out whether the 87% means anything. Worse, it invites you to trust a number you cannot check.

Mass outreach. Automating volume without personalisation reduces reply rates and burns the employer brand. The sourcing bottleneck is rarely how many messages you can send.

The hiring decision itself. Beyond being a bad idea, it is increasingly a legal one — the EU AI Act classes employment-related AI as high-risk, and several jurisdictions now require audits of automated hiring decisions.

The useful middle: automating the reading

Between mechanical tasks and judgement sits the work that actually eats the week — reading. Finding a candidate's public activity, checking whether their stated experience matches what they have built, noticing that two profiles are the same person, working out which three of forty are worth a call.

This is not a decision. It is preparation for one. Automating it gives back the hours without taking the judgement away — provided the output is something a human can check. A verdict you can read and disagree with is useful; a percentage you cannot interrogate is not.

How to evaluate a tool

1. Does it show its reasoning? If the output is a number with no explanation, you will end up redoing the work.
2. Can you overrule it, and does it learn? Tools that treat their own output as final are the ones that quietly reject good people.
3. What does it cost per role, not per seat? Per-contact pricing makes careful research expensive, which is the opposite of what you want.
4. What does it refuse to do? A tool honest about its limits is easier to trust than one claiming to automate hiring end to end.

Where leadl.recruit sits

It automates the reading, not the decision. Paste a job description and it searches public professional activity across GitHub, Stack Overflow, LinkedIn, Reddit and Discord, assembles one profile per person, and returns a shortlist where each candidate carries a written verdict: what matches the role, what does not, the risks it found, and what to ask at interview.

You can read every verdict and disagree with it. That is the point — see AI screening for how the reasoning is produced.

Pricing is per search run, not per contact revealed: free plan, then €49, €149 and €399 a month. Careful research on a candidate costs nothing extra. See pricing.

It does not schedule interviews, track applicants or replace an ATS. For those, keep the system you have.