71% of candidates in Ireland now use AI when they apply for a job, according to Morgan McKinley's 2026 research. Applications look more polished than ever and often say less about the person behind them. In the same research, 31% of employers said AI-written CVs have made skills harder to verify.
So the question for an agency is no longer whether to use AI. It is where AI genuinely helps, and where a consultant's judgement matters more than ever. The clearest way to answer that is to follow one role from start to finish.
The example below is illustrative: a Financial Controller role for a Dublin client, worked by one consultant. The research and shortlist steps draw on a system we run for an executive search firm.
The brief
The client rings with the role. The consultant runs the intake call as normal, with the client told at the start that it is being recorded.
- Must have
- Qualified accountant (ACA, ACCA or CIMA) with at least three years running month-end for a group
- Must have
- Consolidation across more than one entity
- Nice to have
- Has moved a finance team from Sage to a cloud system
- Deal-breaker
- Cannot be in the office three days a week
- Salary
- €90,000 to €100,000, to confirm with the client
Every later step is measured against this scorecard, so ten minutes spent getting it right saves hours further on. It also gives the client something concrete to agree to before a single CV is sent.
Your own database first
Before the role is advertised, AI searches the agency's own CRM against the scorecard. It looks for people placed a few years ago who may be ready to step up, strong runners-up from past roles, and candidates whose call notes mention the right experience, even when their job title does not.
Applications and the shortlist
The advert goes out and the applications arrive, many of them written with help from AI. AI reads every one in full against the scorecard and writes a short note on each: the evidence for, the evidence against, and what to check on the call.
- For
- ACA. Four years as Group Financial Accountant, consolidating five entities.
- Against
- No team management mentioned. The role leads two people.
- Check on the call
- The CV lists Power BI and Python. Ask what they actually built with them.
- Suggested
- Call. Strong on both must-haves.
The reasons are the point. Because the note shows its working, the consultant can disagree with it in seconds, and the client can see exactly why a candidate was put forward. With so many CVs now polished by AI, the "check on the call" line is often the most useful part.
Interviews and feedback
This is where most of the chasing lives, and where candidates are most often left waiting.
Offer, references and the first weeks
- Start date
- Monday 9 November
- Notice
- One month. Resigning tomorrow morning.
- Counter-offer risk
- Medium. Her current employer offered a pay rise in the spring.
- Next
- References by Wednesday. Check-in call booked for Friday.
What the consultant does that AI cannot
Look back over the five steps and the pattern is clear. AI does the reading, the searching, the chasing and the record keeping. The consultant does the judging, the persuading and the relationships: reading a hesitant candidate, pushing back on a client's brief, sensing when a counter-offer is coming. That is the work clients pay a fee for, and it is exactly the work that gets squeezed when the week fills up with admin.
Is this allowed?
Yes, with a person in charge. This is where the rules stand in October 2026.
- The EU AI Act treats AI that filters applications or evaluates candidates as high-risk. In July the EU moved the start date for those rules from August 2026 to 2 December 2027. Many guides online still give the old date.
- From then, an agency using such a tool must put a trained person in charge of it, keep the logs it produces for at least six months, and tell candidates that AI is being used in decisions about them.
- One rule already applies: AI that infers the emotions of workers or job applicants, such as a video tool that scores facial expressions, is banned.
- In Ireland, the Workplace Relations Commission supervises AI used to screen applications and rank candidates. The Regulation of Artificial Intelligence Act 2026, signed in July, sets out how the rules are enforced here.
- GDPR already gives people the right not to be subject to a decision with a significant effect on them that is made by a computer alone. The EU's highest court has found that a score can count as that decision when someone else relies heavily on it, so a shortlist score your client acts on needs reasons a person has checked.
A system built like the example above already sits well with all of this. It shows its reasons, keeps a record of what it did, and leaves every decision about a candidate with a person.
Where to start
Pick the step that costs your desk the most hours. For most agencies that is the shortlist or the chasing. Start there, run it alongside your consultants for a month, and add the next step once the first one is trusted.
This article is general information, not legal advice.