How the screening works.
Three deliberately separate stages run on each inventory row. Keeping them separate is what lets the tool be honest about what it can and cannot tell you.
Experimental tool — not approved for decision-making use. A screening aid only; it does not determine compliance, safety or exposure, and its coverage is incomplete.
1. Chemical identity
Each row is resolved to a BecSpec chemical identifier from its substance name, product name and — if you provide one — its CAS registry number. A match is only treated as confident on an exact CAS number or an exact known name/synonym. String similarity alone never resolves an identity: a mistyped or unknown name returns unable to resolve, not a no-change. CAS is optional evidence, never the key BecSpec relies on, so a substance still resolves without it.
2. Transition resolution
A resolved substance is matched to the WES→WEL change published by Safe Work Australia — a numeric limit change, a notation change, a new or removed entry, a “no threshold guidance concentration” (NTGC) treatment, a merge/split of related entries, or a 24 June 2026 ministerial decision that a proposed change does not commence. Every transition carries its source. Where a change is complex and BecSpec cannot represent it deterministically, it is shown as review-required with its evidence rather than reconstructed by guesswork.
3. Outcome
The classifier reports exactly one of five states, and never a compliance or safety verdict:
- Affected — confirmed change: a listed WES→WEL change applies.
- Review required: a change may apply, but the applicable entry (for example the physical form or fraction) or the change itself needs manual review.
- No listed change identified: no change to a listed limit was found. This is a statement about the regulatory change only — not that any exposure is acceptable.
- Unable to resolve: the substance could not be identified confidently.
- Process review required: the contaminant is generated by a work process and an inventory scan alone cannot assess it.
Why it prefers review over a confident answer
Where the evidence does not cleanly support a confident “affected” result — an identifier shared across several form-specific entries, an inhalable-versus-respirable fraction the inventory does not state, a merged/split entry family — the tool returns review or unresolved. A wrong confident classification is the one outcome the tool is built to never produce.