Short version: “AI” is used to describe very different software, from simple rules to machine learning and language models. In labour hire, the useful parts are mostly practical: filtering and shortlisting workers, flagging exceptions, and chasing the routine follow-ups. Legal, employment and safety judgements still need a person. Australia has no standalone AI Act — the government relies on existing laws and voluntary guidance, and has announced plans to legislate Australian Standards for AI. From 10 December 2026, organisations covered by the Privacy Act must explain in their privacy policy certain uses of computer programs in decisions that significantly affect people. OnCrew’s own features are rule-based, and a person decides who is offered work.
“AI” covers very different things
Vendors use the word for almost anything automated. It’s worth asking which of these a tool actually is:
- Rules and filters — if a ticket has expired, flag it; if a worker is already booked, leave them off the list. Predictable and easy to explain.
- Scoring and ranking — points for set factors, such as distance or past shifts at a site, added up to order a list. Explainable if the tool shows its reasons.
- Machine learning — a model trained on past data to find patterns or make predictions. It can find patterns rules miss, but its outputs are harder to explain and depend heavily on the data it was trained on.
- Language models — software that generates text, such as chat assistants. Useful for drafting and answering questions, but it can produce confident, wrong answers.
- Configured automation — an action your business sets up to run on a trigger, such as a reminder text before a licence expires.
None of these is better in every case. What matters is what the tool does, what data it uses, whether it can show its reasons, and who is accountable for the outcome.
Where automation genuinely helps
- Repetitive admin — expiry reminders, chasing missing timesheets and confirmations, and sending offers once someone has chosen who to offer.
- Shortlisting — ruling out workers who aren’t qualified, cleared or available for a shift, and ordering the rest by factors you can see.
- Exceptions — surfacing what needs a person: a missed clock-in, an unapproved timesheet, a shift still short of workers, a credential about to lapse.
- Visibility — one view of shifts, attendance and approvals instead of a spreadsheet, a group chat and a whiteboard.
Judge any tool on what it takes off your team’s plate in an ordinary week — ideally by trialling it on your own shifts — rather than on percentages you can’t trace to a source.
Where people should stay in charge
- Legal and employment judgements — which award applies, a worker’s classification, whether someone is an employee or a contractor, and whether casual employment is genuine. Software can record what you decided; it shouldn’t make those calls. See sham contracting vs genuine casual employment.
- Safety decisions — risk assessment, consultation, supervision and notifying the regulator stay with the businesses that hold the WHS duties.
- Context a record can’t capture — a worker going through a hard time, a client relationship, a site with an unusual problem.
- Accountability — someone in your business should own each decision that affects workers, and be able to explain it.
Data quality decides what any tool can do
A tool can only block a worker on a ticket it knows about, with an expiry date it has been given. It can only show who was on site if clock-ins were recorded. Fragmented records — a roster in one place, tickets in another, attendance in a group chat — limit what any software can do, however it is labelled. Getting onboarding, credentials and attendance into one system is usually the first, unglamorous step.
The rules in Australia, as at October 2026
Privacy Act: transparency about automated decisions from 10 December 2026
Changes made by the Privacy and Other Legislation Amendment Act 2024 add a transparency obligation to Australian Privacy Principle 1, starting on 10 December 2026. It applies where an organisation has arranged for a computer program to make a decision, or do something substantially and directly related to making one, that could reasonably be expected to significantly affect an individual’s rights or interests, using their personal information. The organisation’s privacy policy must then describe the kinds of personal information used and the kinds of decisions involved. The OAIC published updated guidance, a fact sheet and a flowchart on 30 September 2026.
Whether it applies to your business depends on whether you are covered by the Privacy Act at all — most small businesses with an annual turnover of $3 million or less are not, although some are — and on how exemptions such as the employee records exemption apply to your situation. If you use software in decisions about who is offered work, it is worth getting advice before December.
Australia’s wider approach to AI
- National AI Plan (2 December 2025). The government’s approach builds on Australia’s existing, largely technology-neutral laws rather than a standalone AI Act, supported by a new AI Safety Institute.
- Guidance for AI Adoption. Voluntary government guidance setting out six essential practices: decide who is accountable; understand impacts and plan accordingly; measure and manage risks; share essential information; test and monitor; and maintain human control.
- Australian Standards for AI (announced 15 July 2026). The Prime Minister announced that the government will establish a set of Australian Standards for AI, described as mandatory, with the aim of bringing legislation to Parliament early in 2027. Until legislation passes, they are not law, and their detail is still being designed.
In the meantime, existing laws — including privacy law — already apply to how businesses use AI and automated tools.
What OnCrew actually does
Inside the OnCrew app, the features that sound like AI are rule-based. Here is what each one does:
- Smart Fill gives an admin a shortlist for a shift. It first rules out workers who aren’t compliant for the site, Ready to Work or available, then scores the rest on role fit, site and client history, distance and recency, and shows the reasons for each worker’s position. The admin decides who is offered the shift.
- Axis is OnCrew’s rule-based insights layer. It applies fixed rules to your current data to surface insights and suggested follow-ups for a person to act on — including a 30-day reliability score built from completed shifts, no-shows and lateness. That score is shown on the worker’s profile; Smart Fill doesn’t use it. Axis doesn’t train a model on your data or learn over time.
- Axis’s automated actions run in review-only mode. They record what they would have done and send nothing. Its command bar, where an admin can type a request, uses a rule-based parser rather than a language model, and only previews changes — a person has to confirm them.
- Configured automations are agency settings. Reminder texts before a credential expires are on unless your agency turns them off; follow-ups for missed clock-ins and unapproved timesheets run only where your agency has set them up.
- Site requirements apply the rules your office sets for each site. A missing or expired enforced requirement blocks an assignment unless an admin overrides it, and the override is logged. OnCrew doesn’t decide whether a worker is legally suitable for work.
OnCrew encrypts specific sensitive fields — medical, tax and bank details — and logs each time they are revealed. That supports good privacy practice, but it doesn’t make any business compliant with the Privacy Act on its own.
Questions to ask any vendor
- Which features use rules, which use machine learning or language models, and which are configured automations?
- Can the tool show why it ranked or flagged someone?
- What does it do without a person confirming it first?
- What data does it use, and where does that data come from?
- Who in our business is accountable for the decisions it supports, and how would we explain them to a worker?
Official sources checked for this article (October 2026): OAIC — new resources on transparency for use of AI and automated decision-making (30 September 2026) · OAIC — APP 1 guidelines · OAIC — small business · OAIC — employee records exemption · Minister for Industry and Innovation — National AI Plan (2 December 2025) · Department of Industry, Science and Resources — National AI Plan: keep Australians safe · Department of Industry, Science and Resources — Guidance for AI Adoption · Prime Minister — AI in Australia’s interests (15 July 2026)
