A decade of two-sided data, and a stubborn belief in potential.

Twelve years ago, the founding observation was simple. The polished CV is rarely the strongest signal of future performance. It’s a signal of who has been supported well — the parents who edited the personal statement, the family connection that opened the internship, the school that ran a careers programme worth attending. None of that is the candidate’s fault, and none of it tells you what they’ll actually do at work.

The average graduate vacancy gets 140 CVs. Most of them measure past advantage, not future potential.

The market has only got harder since. Candidates now use AI to write applications. Employers use AI to screen them. Volumes go up, signal quality goes down, and confidence in the decision at the end goes with it. Adding more screening is not the answer. The answer is to capture genuine behavioural evidence before a candidate writes a single word — and to make the eventual hiring decision on what they actually did, not what they (or their language model) claim.

That’s the system we’ve been building, one product at a time. It’s also, incidentally, how you find potential everywhere — first-generation candidates, apprentices, school-leavers, people whose CV doesn’t quite scan but whose work, when you look at it, clearly does.