FRI 2 OCT · all markets · live specs · governed view · week ending 2 October 2026
Every figure carries four labels, or it does not publish.
Its population, its basis, its definition version, and its freeze. Verellax does not survey anyone. Every figure is read from what an organisation has already published and carries its date, and every projection names the measures it is built on.
A collected count is never coverage. The gap between collected and canonical is mostly reposts and roles outside the named markets, kept by design.
Verellax measures the distance between what organisations say about AI and how they have designed the work around it.
The thesis has four nouns: AI is redistributing tasks faster than organisations are redesigning roles, decisions, and learning systems. Each noun has an instrument that reads it, and the first moves faster than the other three.
A benchmark, not a standard: what good looks like is found in the distribution, as the minority that has designed the work furthest, and is never set in advance.
We read what organisations have already published, never a survey.
A record joins only when it is collected, classified, and floored across the sector.
A figure names its base in the sentence that carries it.
Source: Verellax workforce intelligence · 43,967 live specs · October 2026 · week ending 2 October 2026
Open roles at the freeze. Every weekly figure on the site reads this base.
Live and closed roles grouped by when they were posted. The Index edition reads this base, one quarter at a time.
Named markets clear 150 live roles, 10 employers, and 30 AI-central roles: Australia, the UK, the US, Ireland, Canada, and Singapore.
Each role is read on whether it involves AI, whether AI sits at the centre of the work, and whether the role is accountable for what the model produces.
Source: Verellax workforce intelligence · 43,967 live specs · October 2026 · week ending 2 October 2026 · each share on the measure before it
7,144 (67.7%) meet none of the four. The four are reported one by one and never added into a score.
Source: Verellax workforce intelligence · 43,967 live specs · October 2026 · week ending 2 October 2026 · AI-central roles · governed view
For every live role written to govern AI, employers posted more than thirteen written to use it.
Source: Verellax workforce intelligence · 17,241 live roles that involve AI · all markets · live specs · week ending 2 October 2026
Each role that involves AI is also read for its relationship to AI, whether it is a new seat or a backfill, where on the ladder AI lands, and how long it stays open. The findings come from crossing a measure with a lens, and every cross is named before it is read.
Every AI-central role is then read on four tests. Accountable is one of them. Roles that involve AI and AI-central roles are never interchanged, and the finding is always about the specification, never a claim that an organisation has no one accountable. US test figures are held while the live and frozen US counts disagree.
Every published figure is a row, computed once, and frozen before it reaches a page.
A sector publishes at 10 or more companies. A cell below its floor renders translucent with its count, never suppressed and never headlined.
Two freezes. The weekly freeze runs every Monday at 10:00 AEST and every page re-renders from it. The edition freeze is quarterly and pins one pack; the edition and its press pack render from that and nothing else.
Naming. Edition One names no employer. From Edition Two, a company is named only in the positive, above the floor, after fourteen days' notice. A company may always read its own figures; nobody else sees them.
Corrections. An error is corrected in public, as a dated note, with the query shown.
Documents, not practice. Absence from a specification is not absence from the organisation.
The method reads roles as written, not what happens after the hire. It makes no claim about workflows, workload, or automation from job specifications alone, and a stated skill is what an employer asks for, never what anyone holds. AI-central roles cluster in head offices, so geography partly measures where decisions are written. Trends wait for six months of continuous collection at stable coverage.
Source: Verellax workforce intelligence · 43,967 live specs · October 2026 · week ending 2 October 2026 · the method changes only by version: methodology v3.0