AI interview software built for Australian hiring law.
Australian employers carry an obligation most global hiring platforms were not designed for: under the Fair Work Act you must prove a protected attribute played no part in a rejection, for six years afterwards. FirstPanel interviews every applicant and produces the record that answers it.
A global screening tool leaves you holding an Australian burden of proof.
Most AI hiring platforms are built for markets where the employer is presumed innocent. Australia is not one of them. Section 361 of the Fair Work Act presumes a rejection was made for the reason a candidate alleges unless you prove otherwise, s.360 requires only that a protected attribute formed part of the reasons, and a rejected applicant has six years to raise it. Meanwhile, from 10 December 2026, APP 1.7 requires your privacy policy to describe what your screening technology does with candidate data — which means knowing, in writing, what your vendor’s model actually sees. A ranked shortlist with no evidence behind it does not answer any of that.
Built for the way you hire
Evidence behind every score
Each of the eight MERIT-8™ agents cites verbatim, timestamped interview evidence for its rating — or abstains and routes a follow-up probe. A score without a citation never ships.
Content-only scoring
Scoring models receive transcript content and nothing else. No video frames, no audio features, no names, no postcodes, no inferred demographics — the proxies that create Australian discrimination exposure are absent by architecture, not disabled by a setting.
The Australian rule pack, in product
The Fair Work reverse onus, the six-year record floor and the APP 1.7 disclosure obligation are encoded as product rules. Criteria that would breach them are refused when a hiring manager tries to configure them, not caught in a quarterly audit.
Australian data residency
Candidate data held in-region, with retention configurable per artefact class — six years for the decision record, short cycles for raw media, so retention and minimisation stop fighting each other.
Built for the Australian calendar
A summer retail and hospitality peak, a February graduate intake, harvest and tourism seasons. Per-completed-interview pricing means you spin a peak round up and wind it down without carrying idle licence cost between them.
Every applicant, in their own language
Candidates interview in voice, video or text in whichever of 40+ languages they are strongest in — which in an Australian pipeline is the difference between assessing capability and assessing fluency.
The artefact your legal team is actually asking for
- Per-decision records retained to the six-year Fair Work floor
- APP 1.7 limb classification supplied for your privacy policy
- Per-requisition adverse-impact monitoring before shortlists ship
- A named human reviewer on every disposition — AI never rejects alone
The competencies that predict success
Eight AI agents each own one competency and score it against your rubric — every rating linked to the moment in the interview that earned it. These are the ones that matter most for this kind of role.
The predictive competency across most of what Australia hires at volume — retail, hospitality, care, contact centre. Probed through real scenarios, never self-rating.
Rosters, shifts and follow-through. The largest single driver of frontline churn, surfaced before you hire rather than in week three.
Cash handling, responsible service, safety-critical calls and knowing when to escalate — assessed on what a candidate actually did, with the evidence attached.
A rush, a delay, a system outage, a plan that fell apart. Composure and recovery, probed rather than asserted.
Questions teams ask
Is AI interviewing legal in Australia?+
Yes. No Australian law prohibits AI-assisted candidate assessment. The obligations are transparency, non-discrimination and explainability: disclose automated decision-making in your privacy policy from 10 December 2026 under APP 1.7, keep criteria job-relevant, and retain a per-decision record you could produce if a rejected applicant alleges a protected attribute played a part.
What does the Fair Work Act require of an AI screening tool?+
Nothing directly — it regulates the employer. But s.361 presumes a rejection was made for the alleged discriminatory reason unless you prove otherwise, which in practice means your tool has to be able to hand you the evidence. FirstPanel records the questions asked, the evidence cited for each score, the rules in force, the model versions and the named human who decided, contemporaneously and immutably.
Do you offer Australian data residency?+
Yes, an Australian residency option is available. Ask us in the pilot for the specifics that matter — primary region, backup region, sub-processor list and where model inference runs, which is the question most vendors cannot answer.
How long do you retain candidate records?+
Retention is configurable per artefact class. We recommend a six-year floor for the decision record specifically, because a rejected applicant is not caught by the 21-day dismissal clock and the general six-year limitation in s.544 applies — while raw video and audio can and generally should be deleted much sooner.
Does the AI reject candidates?+
Never. AI screens, humans decide — always. FirstPanel interviews, scores and ranks with evidence attached; a named human reviews and makes every hiring and rejection decision, and the override trail is part of the record.
Do you integrate with our ATS and with SEEK?+
Screening sits on top of your ATS rather than replacing it — the requisition and the candidate stay in your system of record, and results are written back. Talk to us about your specific stack; the API and SDK are documented on our developers page.
See it on one of your own roles
Pick your longest-open requisition. First interviewed shortlist in about two weeks — keep the reports either way.