Nexus Interactive walkthrough
Request a demo

Illustrative walkthrough. No real candidates, no real outcomes.

See how Nexus screens a role, from the first sentence to the audit trail.

You set the parameters. Nexus does the reading and the arithmetic, and shows it. A person makes every decision.

The Nexus dashboard in dark mode: active jobs, applications received, average match and the actions waiting for a decision.

Before the demo: the shape of the process.

Every job moves from what you provide, through what Nexus computes, to what you decide. Amber marks a decision that belongs to a person, here and in the product.

  1. Role
  2. Parameters
  3. Job ad
  4. Applications
  5. Score
  6. Review
  7. Trace

What you provide

A role, described in a sentence or as an existing job ad. Then the applications, received through a public page or imported by your team.

What Nexus computes

It structures each CV, computes five published criteria, and drafts the job ad, interview questions and rejection emails. Every number can be traced back.

What stays with you

Approving the brief, the job ad and any rejection list. Deciding who to interview and who to hire. Nexus sends no email to candidates.

The walkthrough.

Seven steps, in the order the product runs them. Each one says why it exists and what happens next. Change any value: the arithmetic is the product’s own.

Why this step existsScreening starts from what the role needs, not from the CVs. One sentence is enough: Nexus asks for the rest.

Describe the role

What Nexus asks next

  • The required skills
  • The minimum years of experience
  • The location, the contract type and the work arrangement

Already have a job ad? Paste it instead, and this scoping step is skipped. The language you choose applies to everything the candidate reads, and cannot be changed afterwards.

What happens nextYou confirm the answers. That confirmation is the first of up to three human checkpoints.

The same journey, filmed.

Under two minutes in the real product, on the demonstration instance, without voice-over.

Measured, with its limits stated.

Twin CVs that differ by one attribute are scored against the same job. The gap must stay within 3 points out of 100, a limit set before measuring. Measured on October 22, 2026, with real model calls on synthetic profiles.

1

point at most, when twins differ by gender, origin suggested by the name, age or place of residence.

2

points at most, when the same CV is written in English or in French.

0.43

points of standard deviation when a strong profile is scored eight times. 0.86 for an average one. Limit: 2.

What these numbers are not, and what we do not have yet

  • Not an adverse impact analysis on real applicants.
  • Not an independent audit. One model was tested.
  • No SOC 2 or ISO 27001 certification yet.
  • No conformity assessment under the EU AI Act yet.

See it on your own roles.

A pilot runs 8 to 12 weeks, on one or two of your open roles, on an instance dedicated to your organization. The measures are agreed before it starts.