Neuroscale at SHRM Annual 2026

SHRM Annual is the largest HR gathering in the country, and this year it felt like the first one where AI in recruiting stopped being a novelty and became a line item. Nobody asked us whether they should be using it. Everyone asked us how to tell whether it was working.

We ran the booth as a working demo rather than a pitch. Recruiters brought their own job descriptions, we wrote the criteria together, and Arbi scored a real stage against them while they watched. Over four days that happened roughly three hundred times.

What people wanted to see#

The single most common request was the evidence. Not the score, the reason for the score. A screening verdict that cites the line on the resume it rests on turned out to be the thing that made a sceptical talent lead lean in, because it is the same thing they would have to produce if a hiring manager challenged the shortlist.

The second was speed on a real stage. Four hundred profiles scored in the time it took to get a coffee is a different conversation from a slide that says it.

The roundtable#

On the Tuesday evening we hosted a roundtable for twenty heads of talent. The theme was measurement: what a recruiting team can honestly claim to know about its own funnel. Three things came up in nearly every conversation.

• Screening consistency is unmeasured almost everywhere. Teams track time to fill and offer acceptance, and almost none track whether two recruiters reading the same profile reach the same verdict.
• Reply rates are down and nobody agrees why. The volume story and the personalisation story were both in the room, and the honest answer is that most teams cannot separate them with the data they keep.
• Audit pressure is arriving before the tooling. Several teams in regulated industries are being asked to explain hiring decisions they made with software that keeps no record of its reasoning.

What we took home#

A list of forty-one feature requests, most of them variations on one theme: show the work. The evidence layer is what people trust, and every part of the product that hides a decision behind a number is a part they trust less.

Thank you to everyone who stopped by, argued with us, and brought us a real stage to screen. See you in Las Vegas in October.