AI in graduate recruitment

Source Unilever (with HireVue and Pymetrics) — public case. This is an industry example, not our project
~50,000
interview hours saved over 18 months
4 months → 4 weeks
cut in time-to-hire (~90%)

The problem

Screening hundreds of thousands of early-career applicants manually is slow, expensive and inconsistent.

The AI approach

Candidates complete neuroscience-based games (Pymetrics) and recorded video interviews assessed by AI (HireVue), narrowing the pool before human interviews.

Evidence it works

Screening ~250,000 applications for ~800 early-career roles across 50+ countries, Unilever reported saving ~50,000 hours of interview time over 18 months and £1M+ a year in costs, cutting time-to-hire from roughly four months to four weeks (~90%), lifting candidate completion to ~96%, and increasing diversity of hires by ~16%. The approach also evolved: HireVue dropped facial-expression analysis in 2021 — after a US FTC complaint over bias and validity — moving to assess language and content instead.

What “good” looks like

Faster, more consistent early screening that widens access and reduces bias — with transparent, defensible assessment criteria.

Feasibility & cost shape

Vendor platforms make deployment feasible, but the governance, fairness testing and candidate-communications work is the substance.

Our independent view

Useful for both the efficiency gain and the lesson that explainability isn’t optional in high-stakes, regulated decisions about people. We’d lead with the governance design.

Source & attribution

Based on publicly reported information about the Unilever (with HireVue and Pymetrics) work.

This is an industry example included for illustration. It is not a Leia Intelligence project, and no client of ours is implied. Figures are as publicly reported by the original parties.

Sources: Unilever / HireVue / Pymetrics case reporting · coverage of HireVue's 2021 move away from facial analysis

This is the class of problem we help organisations tackle — book a call.
Book a call