The market is tight, and the entry-level rungs are the ones under most pressure. Adzuna figures show a 30% drop in UK entry-level job postings since ChatGPT launched, with graduates facing the toughest job market since 2018 techUK – What's actually happening with entry-level and graduate jobs?. The tech sector has cut deeper still: graduate roles fell 46% in a single year, with a further 53% drop projected, according to Institute of Student Employers figures The Register.
The wider picture is what the Low Pay Commission calls a 'low hire, low fire' labour market: weaker recruitment, only slight increases in redundancies, and vacancies below pre-pandemic levels Low Pay Commission Report 2025, GOV.UK. Official vacancy counts keep drifting down, too, with early estimates for May to July 2026 pointing to 707,000 vacancies, a fall of 6,000 on the previous quarter ONS - Vacancies and jobs in the UK, August 2026.
So the question for employers has shifted. It's no longer 'can we find people?' It's 'are we hiring the ones who make our AI-assisted teams more reliable, not less?'
Demand hasn't disappeared. It has concentrated. Businesses took a cautious approach through 2025, but there were clear hotspots for specialist skills in AI, data, enterprise applications and cyber security Computer Weekly (tech recruitment outlook 2026). AI skills are the sharpest spike: demand rose nearly 200% in a year, with London accounting for 80% of AI-related job postings The Register (Accenture data).
The way in is changing as well. Apprenticeships have risen from 3% of AI hires in 2020 to 19% in 2025 GOV.UK / DSIT – AI Labour Market Survey 2025 report. That matters because apprenticeships build verification habits on the job: reading code, testing assumptions and asking whether output is correct before it ships.
Engineering feels the entry-level squeeze most acutely. Entry-level hiring in engineering has fallen to -19% while senior hiring sits at -10%, a nine-percentage-point gap GOV.UK - A snapshot of entry-level hiring in the UK. And most of these hires come from the resident labour force, so the domestic early-career pipeline is the pipeline GOV.UK - Professionals in IT and engineering. Neglect it now and the senior gap widens later.
Start by changing what you test for. AI can already generate plausible code, so producing code is no longer the differentiator. Reviewing it is. In your assessment, hand candidates a working-but-flawed AI-generated snippet and ask them to find what's wrong, explain why, and prove their fix with a test. You learn more from how someone interrogates output than from watching them write from a blank file.
Look for these verification habits in interviews and tasks:
Widen the door, too. With apprenticeships now a real route into AI teams GOV.UK / DSIT – AI Labour Market Survey 2025 report, structured earn-and-learn programmes give you people who build verification skills against your own codebase. And if you work in cyber security, treat this as urgent: 70% of cyber firms reported at least one hard-to-fill vacancy AI Labour Market Survey 2025 (Gardiner & Theobald), GOV.UK / publishing.service.gov.uk, and verification is the whole job there.
One more thing. In a low-hire market, speed still wins. When professional, scientific and technical vacancies do rise, they rise fast ONS – Vacancies and jobs in the UK: November 2025, so the employers who move first on strong early-career talent tend to secure it.
We help you find early-career engineers with the judgement to check what AI produces, then manage the process end to end. We search a database of 15 million people and surface a ranked shortlist in under 30 seconds, so you can focus your time on the verification tasks and questions that reveal who thinks, not just who types. Our recruitment agent manages recruitment end to end for 8% on a successful hire, with no monthly fee and no upfront cost. Ready to build a team that trusts, but verifies? Start your shortlist with Reed.ai today.
