Before I worked in recruiting, I was a developer. Since then I have spent more than a decade in talent acquisition across Latin America, EMEA and APAC. For example, I led tech recruitment teams at H&M Group, recruited at Klarna, and joined Lovable as one of its early recruiting hires.
Those years taught me that talent acquisition does not have a fixed job description. It has seasons. The title stays the same, but the work changes as a business moves from its first hires to hypergrowth, into new markets and into maturity. Teams struggle when they keep doing the job their company needed two stages ago.
The pressure to adapt is real. In Europe, annual applications per recruiter rose 385% between 2022 and 2025 (Greenhouse). AI is the obvious answer, and the obvious risk. Here is what each stage demands, how I would use AI there, and where I draw the line.
The Four Seasons of Talent Acquisition
Stage One: The First Hires, Where the Recruiter Is a Storyteller
In the earliest days there is no employer brand and little process. Candidates are betting on a vision and the people behind it.
When I joined Lovable, the job was largely translation. I had to turn an ambitious idea into an opportunity a senior candidate would leave a stable role for. Candor about what we did not know closed more candidates than hype ever could.
This is where putting AI between you and the candidate costs the most. Only 26% of candidates trust AI to evaluate them fairly (Gartner). So I use AI around the conversation, never instead of it: market research, intake forms, interview kits and job descriptions. None of it needs candidate data.
Stage Two: Hypergrowth, Where the Job Is Protecting the Bar
Klarna showed me rapid scaling from the inside, as the company grew to around 5,000 people. New teams, first-time hiring managers and constant urgency create one real risk: quietly lowering the bar.
Volume makes this harder. Applications per hire tripled between 2021 and 2024 (Ashby). Yet the evidence on what works is clear. A 2022 meta-analysis led by Paul Sackett found structured interviews to be the strongest predictor of job performance (SIOP).
Back then, calibration lived in spreadsheets and memory. Today I would make three changes:
· Generate structured interview kits in an afternoon.
· Train interviewers on synthetic candidate profiles.
· Use anonymized funnel data to spot where pass rates drift.
But fix the process first. Automating a broken one only repeats mistakes faster.
Stage Three: New Markets, Where You Listen Before You Hire
New markets are the most underestimated inflection point. Working across regions, including training individuals in China and South Korea, taught me that approach norms, compensation, notice periods and decision speed all change by market. An outreach message that feels efficient in Stockholm can land as cold in São Paulo.
The job here is talent intelligence. Toby Culshaw, author of Talent Intelligence, frames it as combining people data with analysis of skills, competitors and geographies to drive business decisions.
Market mapping used to take me weeks. AI now drafts it in a hour, if you map companies and skills, not people. That means public salary benchmarks, competitor job postings, and employment rules you then verify with local counsel. AI shows what the market looks like. People on the ground tell you what it means.
Stage Four: Maturity, Where TA Becomes a Workforce Partner
At H&M Group, I led tech recruitment in a business with more than 4,100 stores and roughly 95,000 full-time equivalents (Annual Report 2025). At that scale, the big questions are about which skills come next and where people can grow.
Employers expect 39% of core skills to change by 2030 (WEF). LinkedIn also found that promoted employees are far more likely to still be at their company after three years than those who stay in the same role (LinkedIn).
This is where I see the profession heading. Recruiters will shape, with leadership, how teams are built before a requisition exists, alongside HR business partners who focus on the people already inside. AI helps by drafting future role profiles and running scenarios, such as what a team looks like if AI absorbs a third of its routine work.
Where I Draw the Line
My rule is simple: AI on the work about the job, humans on judgements about the person.
The AI Risk Ladder for Talent Teams
Rule of thumb: AI on the work about the job, humans on judgements about the person
Regulation agrees. Recruitment AI is classed as high risk under the EU AI Act, with those obligations now applying from December 2027 (Kinstellar). GDPR Article 22 already restricts decisions made solely by automated means (GDPR). This is not legal advice. Bring your privacy and legal teams in as co-designers, not end reviewers.
What Stays the Same
Hiring is still a human decision, and no tool replaces the trust that makes it feel safe. The recruiters who adapt fastest no longer wait for someone else to build their tools. My developer past helped, but today curiosity and good judgement is enough to get started.
Questions Worth Asking Your TA Team
· Is our talent function built for the stage we are entering, or the last one?
· Are we measuring what matters for this stage? If so, how? And how soon do we recalibrate for a world that is changing so fast?
· Where is AI freeing up time, and are we reinvesting it in the next stage's work?
Talent acquisition is not one job. It is at least four. The companies that recognize the shift early keep hiring well when everything else changes.
