471 Australian workplace decisions in your AI advisor — your HR question has probably already been decided

FairWork Mate's case-law corpus, refreshed daily, is now 471 published Australian workplace decisions:

• 230 Fair Work Commission decisions — unfair dismissal, general protections, casual conversion disputes, modern-award variation, anti-bullying applications, FWCFB appeals
• 29 Federal Court of Australia judgments — Fair Work civil-penalty proceedings, sham-contracting prosecutions, larger appeals from FWC
• 212 Fair Work Ombudsman enforcement outcomes — enforceable undertakings, court-ordered penalties, criminal wage-theft prosecutions (since 2025), media releases on systemic underpayment

Each is summarised plain-English (facts, outcome, employer implication, employee implication), tagged by industry and legal issue, and embedded as a 1,536-dimension vector for semantic retrieval. When you ask the AI advisor a workplace question, it finds the 2-3 most-relevant cases and writes its answer grounded in them, with named citations.

The corpus is growing at ~30 new cases a week from automated daily ingestion of FCA RSS, weekly FWC bulletins and FWO media releases. By end of 2026 we expect 1,000+ decisions — the largest live, AI-grounded Australian workplace case-law corpus available to non-lawyers.

Most HR questions that feel novel to the person asking are routine to the FWC. The FWC hears 13,000+ unfair dismissal applications a year. The Federal Court decides 200+ Fair Work matters. The FWO publishes 80+ enforcement outcomes annually. Across 5-10 years of decisions, almost every common scenario has been litigated by someone, somewhere, in some industry.

The hard part isn't whether there's a precedent. It's finding it. A standard legal-research search would take 30-60 minutes per question across AustLII, the FWC bulletins archive, and the Federal Court's database. An AHRI-certified HR consultant might do it in 20 minutes from memory. FairWork Mate AI does it in 5-10 seconds via vector search across the 471-case corpus.

The output is a direct answer to your question, grounded in the 2-3 specific cases the AI judged most relevant, with named citations and links to the FairWork Mate plain-English summary plus the original judgment. The AI doesn't just say "there's precedent" — it names which precedent and why it applies to your facts.

The corpus is tagged by respondent industry. Heaviest coverage:

• Hospitality (45+ cases) — wage underpayment in restaurants and cafés, casual conversion in fast food, weekend penalty rate disputes, public-holiday roster misallocation, manager-vs-supervisor classification under MA000009
• Retail (30+ cases) — General Retail Award MA000004 classification disputes, casual conversion in big-box retail, shift cancellation penalties, after-hours penalty-rate enforcement
• Healthcare and aged care (35+ cases) — SCHADS Award MA000100 underpayment, sleepover allowances, broken-shift compensation, casual nursing conversion, agency-staff classification
• Construction and trades (40+ cases) — sham contracting prosecutions, building-and-construction-on-site award (MA000020) classification, all-hours allowance disputes, trade-union right-of-entry matters
• Education (25+ cases) — university casual academic underpayment (the Wesfarmers, Melbourne Uni, Sydney Uni and ANU back-pay matters), school-support-staff award coverage, sessional teacher classification
• Finance and professional services (20+ cases) — banking award MA000019 classification, manager-vs-employee distinctions, post-employment restraint enforcement, professional-and-managerial overtime
• Manufacturing and warehousing (25+ cases) — manufacturing award MA000010 classification, safety-disciplinary procedure unfair dismissal, casual conversion in food production

If your business sits in one of these industries, your routine HR questions almost certainly have direct industry-specific precedent in the corpus. If you're in a less-covered industry, the legal principles still apply — the AI will find the closest analogous case and reason from there.

Imagine the following scenario in your business: an employee returns from 12 weeks of carer's leave to look after a sick parent. Within a month of returning, their performance has visibly dropped. You want to start a formal performance-management process. Risk question: is this safe?

Ask FairWork Mate AI: "My employee returned from 12 weeks carer's leave 4 weeks ago. Their performance has dropped significantly. I want to start formal performance management. What's my legal exposure under general protections, and what's the precedent on similar fact patterns?"

The AI returns the framework: Fair Work Act section 340 prohibits adverse action because of an exercised workplace right. Carer's leave is a NES right. If you take adverse action (and formal performance management can qualify) and the leave was a "substantial and operative" reason, you're exposed to a general-protections application with reverse onus of proof. The AI then cites 3 specific cases from the corpus where employers either survived or lost similar fact patterns:

• A 2024 FCA decision where the employer survived because they could show the performance issue pre-dated the leave and was independently documented
• A 2025 FWC decision where the employer lost because the performance management started within weeks of return with no prior documentation
• A 2023 FWCFB appeal decision setting the test for "substantial and operative" reason

The AI then gives the practical advice: document the performance issue completely independently of the leave context, separate the timing, give the employee a clear improvement plan with measurable targets, and — critically — get a 30-minute lawyer call before issuing the first written warning. Total AI time: 5 minutes. Without the case-grounded AI: 2-3 hours of legal research, or a $500-1,000 lawyer engagement to get to the same place.

The retrieval works best when your question contains:

• The industry — "café", "construction site", "aged-care facility", "retail store", "trucking depot"
• The legal issue type — "unfair dismissal", "casual conversion", "underpayment", "sham contracting", "general protections", "redundancy"
• The relevant facts — tenure, employment type (casual/permanent/contractor), the specific event you're asking about
• The decision you're trying to make — "should I do X?", "what's my exposure if Y?", "what process do I need for Z?"

Bad question: "Can I fire someone?" (too vague for the retrieval to find specific precedent)

Good question: "I want to terminate a 4-year casual barista in our café for repeated lateness after two written warnings. Small business, 8 employees. What's my unfair dismissal exposure and what process protects me?" (industry + facts + legal issue + decision)

The good-question version returns: NES + Small Business Fair Dismissal Code applies (under 15 employees), small-business code requires the procedural-fairness steps you've already done (warnings + opportunity to respond) + a valid reason, exposure capped at 26 weeks pay or $87,500 if the termination is found unfair, with citations to recent small-business unfair dismissal cases involving casuals.

The 471-case corpus is the floor, not the ceiling. The growth pipeline:

• FCA RSS ingests daily, filtered by 36 Fair Work-relevant keywords. Currently catching 3-5 workplace decisions per day. Adds ~100/quarter.
• FWC weekly bulletin ingests every Monday morning. Substantive decisions (not procedural appeals) flow through the editorial queue. Adds ~30/week.
• FWO media releases ingest as published. Currently steady ~6/month, expected to lift as criminal wage-theft prosecutions ramp under the 2025 reforms.
• FCFCOA (Federal Circuit and Family Court) — pipeline in build for early 2026 inclusion. Adds another 50-100/year of small-business workplace and contractor matters.
• Industry-specific deep dives — manual ingestion of historically-significant decisions per industry (the wage-theft Wesfarmers and 7-Eleven matters, the casual-academic university actions, the aged-care SCHADS systematic underpayments). Adds 200+ historic decisions over the next 6 months.

By end of 2026 the corpus targets 1,000+ live published decisions, with industry coverage at 60+ cases per major sector. Every new case improves the AI's retrieval quality on similar facts. The moat compounds.

For HR managers and small-business operators making routine workplace decisions every week, the corpus is already practical. Try the AI advisor free with 2 questions/day, or unlock unlimited research from $499/mo for HR teams. Browse the full case database directly for industry-specific reading.