Industry guide - recruitment and hiring
AI for recruitment: automate the chasing, never the choosing
Recruitment runs on ladders: chase the candidate, chase the client, chase the references. AI runs the ladders. Screening decisions - every advance, every rejection - stay human, recorded and defensible. That boundary is the whole design.
By Tylar Edwards - the engineer who scopes and builds these workflows. Updated August 2026.
Why the boundary matters more here than anywhere
Recruitment is the one industry where the tempting automation - letting AI reject candidates - is the one that creates legal and reputational risk. Australian guidance on AI-assisted recruitment consistently lands on the same principle: assistive is fine, decisive is not. The good news is that the hours were never in the deciding. They are in the chasing, scheduling, drafting and updating around it, and all of that automates cleanly.
Six recruitment workflows AI automates today
| Workflow | What AI does | What stays human |
|---|---|---|
| Application handling | Acknowledges every application, extracts CV details into the ATS, and drafts summary notes against the brief | Every advance and every rejection - AI never declines a candidate |
| Job ad drafting | Drafts the ad from the client brief in your house tone, formatted per job board, with inclusive-language checks | Approval, salary positioning and where it runs |
| Candidate communication ladders | Interview scheduling, confirmations, reminders and keep-warm sequences that stop candidates going cold | Offers, negotiations and delivering bad news |
| Interview notes and submissions | Transcribes interviews and drafts the candidate submission profile in your house format | The recommendation and everything in it you would defend to the client |
| Client status updates | Assembles the weekly shortlist and pipeline update from ATS activity | The relationship call that update sets up |
| Compliance chasing | Runs the ladder for right-to-work documents, licences and references, and checks submissions for completeness | Judgment on responses, gaps and red flags |
What it is worth
Using the capacity model behind our free assessment at professional services rates: eight hours a week of recovered ladder-running models to an indicative $13,140-$20,610 a year per consultant at $90/hr. The second-order effect is usually bigger: candidates who stop going cold between interview and offer are placements that stop falling over.
Common questions
- Can AI screen candidates?
- AI can rank and summarise applications against a brief, but the decision to advance or reject a candidate should stay human, recorded and reviewable. Automated rejection is where discrimination risk concentrates, and Australian guidance on AI-assisted recruitment points the same way. Our design rule: AI drafts and chases, a person chooses.
- What should a recruitment agency automate first?
- Candidate communication ladders. Candidates going cold between interview and offer is lost revenue, the ladder is pure process, and the failure mode of a missed nudge is small. It is also where agencies feel the difference within the first week.
- Does this work with our ATS?
- Workflows are built around the systems you already run - your ATS or CRM, Microsoft 365 or Google Workspace, and your billing handoff - rather than replacing them.
- What is it worth?
- At professional services capacity rates, eight recovered hours a week models to an indicative $13,000-21,000 a year per consultant - before counting the placements saved by candidates who no longer go cold in the gaps.