LicenceWatch

1. One-liner

Watches every labour-hire supplier’s state licence daily and warns you before an expiry turns into a $669,120 offence.

2. Trend signal — why now?

Four Australian states now licence labour hire providers, and in every one of them the business that uses the workers commits its own offence if the provider isn’t licensed. Three things changed in the last twelve months:

South Australia swallowed every industry. SA’s labour hire licensing scheme used to cover prescribed work only — horticulture, meat and seafood processing, cleaning, trolley collection. From 29 January 2026 the scheme covers all industries, with a six-month transition ending 29 July 2026 for providers newly caught by it (SafeWork SA; Baker McKenzie client alert, July 2026). Baker McKenzie’s own words: the SA regime “now applies across all industries, substantially expanding the number of businesses and workforce arrangements captured by the scheme.” Every SA business that hires temps, agency cleaners, seasonal packers or contract drivers just walked into a licensing regime it had never heard of.

Victoria is tightening the screws and about to reach through intermediaries. The Labour Hire Legislation Amendment (Licensing) Bill 2025 passed in December 2025 and lands progressively through 2026: an expanded fit-and-proper-person test and new financial viability requirements from 1 June 2026, with further amendments expected on or before 1 October 2026 that clarify the definition of labour hire services “to better cover arrangements in supply chains” and define specific construction activities as labour hire (Labour Hire Authority). Those later changes will empower the LHA to prosecute hosts who engaged unlicensed labour hire indirectly, through an intermediary. Your subcontractor’s subcontractor becomes your problem.

The regulator is publishing a monthly kill list. The LHA refused or cancelled 52 licences in May 2026 and took action against 24 more in June 2026 (7 applications refused, 9 licences cancelled, 7 with conditions imposed, 1 renewal refused) — each named, with ABNs, in a public monthly post (LHA, June 2026). Since 2019 the LHA has removed more than 1,000 non-compliant providers from the industry. Every one of those was somebody’s active supplier the day before.

The penalty is not theoretical and it is not scaled to your size. In Victoria a host that engages an unlicensed provider faces the same maximum as the unlicensed provider itself: $669,120 for a corporation, $167,280 for an individual (3,200 and 800 penalty units at $209.10 for 2026–27) (LHA penalties page). And in October 2024 the Supreme Court of Victoria issued $759,674 in total penalties in the LHA’s first construction prosecution — including $101,774 against Rocktown Pty Ltd purely for engaging two unlicensed subcontractors (LHA media release). Rocktown didn’t supply anyone. It hired.

There is no knowledge defence. Not knowing your provider had been deregistered last Tuesday is not an answer.

Provenance:
  - Signal 1 (demand): Host businesses are being penalised in their own right — Rocktown Pty Ltd fined $101,774 for engaging two unlicensed subcontractors; host maximum is $669,120 for a corporation, identical to the provider's — https://www.labourhireauthority.vic.gov.au/latest-news/record-total-penalties-issued-by-the-supreme-court-of-victoria-to-construction-companies-and-directors-for-unlicensed-labour-hire/ and https://www.labourhireauthority.vic.gov.au/provider/penalties/ — 2026-09-06
  - Signal 2 (feasibility): All four regulators publish machine-readable public registers, and Victoria posts a named monthly cancellation/refusal list with ABNs (52 actioned in May 2026, 24 in June 2026); Queensland maintains standing Suspended and Cancelled licence pages — https://www.labourhireauthority.vic.gov.au/latest-news/providers-subject-to-licensing-action-in-june-2026/ and https://ols.oir.qld.gov.au/cancelled-licence/ — 2026-09-06
  - Signal 3 (economic): South Australia extended licensing to all industries from 29 January 2026 with the transition ending 29 July 2026, and Victoria's 2026 amendments extend host liability to indirect engagement through intermediaries — https://www.bakermckenzie.com/en/insight/publications/2026/07/australia-labour-hire-reform-lands-in-vic-sa and https://www.labourhireauthority.vic.gov.au/about-us/labour-hire-law-changes-2026/ — 2026-09-06
  Category: Regulatory arbitrage

3. The opportunity

Every software product in this market is sold to the wrong party.

Cm3, Avetta, Rapid Global and the rest are contractor prequalification platforms. Cm3’s published pricing is $399–$3,045 per year charged to the contractor (GetApp Australia) — the supplier pays to get onto a big client’s approved list. Access Group, HirePOS and OnCrew sell rostering, timesheets and ticket-tracking to the labour hire agency. All of them are built for the provider side, because that’s who has historically bought compliance software.

But the party the law fines is the host. And the host’s tooling is: a bookmark.

What exists for hosts is genuinely free and genuinely inadequate:

• Victoria’s “Follow my Providers” sends email/SMS when a licence is granted, suspended, cancelled or has conditions changed. Real, useful, and Victorian licences only (LHA).
• Queensland, South Australia and the ACT offer register search and nothing else. QLD publishes standing suspended/cancelled lists you’re expected to read yourself. SA’s register lives inside the general CBS “Find a licence holder” search.
• Free checker tools from Rules Mate and FairWork Mate tell you whether you need a licence, point-in-time, and disclaim themselves: “Reference tool — not professional advice… Confirm the current position with the relevant state regulator or an employment lawyer before relying on this result.”

So the compliance instruction that every Australian employment lawyer publishes — “checking a licence once is not enough; monitor status, scope and expiry on an ongoing basis,” “re-check the register at renewal points, on contract extensions, and before high-risk project mobilisation,” “check the register the day after expiry” — is a manual chore across four separate government websites with four different search interfaces, no shared identifier convention beyond ABN, and licences that are not portable between states. A provider licensed in Queensland cannot lawfully supply into South Australia without a separate SA licence.

The gap: nobody sells the host a single screen that watches all four registers, matches them to the supplier list the host actually has, and shouts before the engagement becomes an offence. The regulator publishes the kill list. Nobody delivers it to the person who gets fined.

4. Target market

Primary customer: Operations manager, HR manager or CFO at a 50–500-employee Australian business that engages labour hire — construction and fit-out contractors, food and meat processors, warehousing and 3PL, aged care and disability services, commercial cleaning buyers, agriculture and horticulture packers, and manufacturers. Multi-site or multi-state operators feel it hardest. Sweet spot: businesses using between 3 and 40 labour suppliers, in VIC, SA and QLD.

Why they buy: In their advisors’ words — “Checking a licence once is not enough.” “This is not delegable.” “No proof of knowledge is required.” The pain moment is specific and recurring: a site supervisor needs bodies on Monday, calls a supplier the business has used for two years, and nobody re-checks the register because nobody’s job it is. Meanwhile in Victoria alone, 52 licences were refused or cancelled in a single month.

Rough TAM reasoning: Victoria’s LHA granted or renewed 185 licences in June 2026 alone and has an active licence base in the low thousands; Queensland’s register is comparable in scale, and SA’s all-industry expansion is pulling in providers who never needed a licence before. Each licensed provider supplies multiple hosts, so the host population is a multiple of the provider population — realistically tens of thousands of Australian businesses are in scope, of which the addressable slice (enough suppliers and enough risk awareness to pay) is a few thousand. This is a small market. It is exactly the right size for a bootstrapper and exactly the wrong size for a VC-funded entrant, which is the point.

Why now for them: SA’s transition ended 29 July 2026 — businesses that were legitimately outside the scheme in December are inside it now. Victoria’s supply-chain and construction amendments land on or before 1 October 2026 and extend liability to indirect engagement. Two dated events, both in the current quarter, both widening who can be fined.

5. Product sketch (MVP)

• Supplier list import — paste or upload your labour suppliers by name/ABN; LicenceWatch matches them against the VIC, QLD, SA and ACT registers and tells you which ones it could and could not resolve.
• Daily licence watch — every supplier re-checked daily across all four registers. Status, licence number, expiry date, industry scope where published, and any conditions.
• Pre-expiry countdown — alerts at 60, 30 and 7 days before a licence expires, plus a same-day alert the moment a status flips to suspended, cancelled, or conditions-imposed.
• Cross-state gap flag — you engage Provider X in South Australia but their only licence is Queensland. Flagged as an exposure, with the specific state you’re missing.
• Engage/don’t-engage check — a fast lookup for a supplier not yet on your list, for the Monday-morning “can we use these guys” question, with a timestamped result you can keep.
• Evidence file — an immutable, dated record of every check performed against every supplier, exportable as a PDF pack. This is what you hand a regulator or an insurer to show you did verify.
• Monthly exposure summary — one page: suppliers watched, actions detected, gaps outstanding, sent to the ops manager and the CFO.
• Contract-clause prompts — flags suppliers whose agreements predate the licensing change, with the notify-us-of-status-change clause your lawyer told you to insert.

6. AI angle — what’s load-bearing

The AI does entity resolution and change interpretation — the two things that make this hard and the two reasons a scraper alone isn’t a product.

Entity resolution. Your accounts payable file says “ABC Staffing.” The Victorian register says “A.B.C. Labour Solutions Pty Ltd ATF the Chen Family Trust.” The Queensland register lists the trustee company. The SA register lists a trading name. ABNs are sometimes present, sometimes not, and providers restructure. Fuzzy-matching supplier names to register entries across four differently-shaped registries, with a calibrated confidence score and a human-review queue for the ambiguous ones, is genuinely a model problem — and getting it wrong in either direction is expensive (false negative = unflagged exposure, false positive = you fired a compliant supplier).

Change interpretation. Victoria publishes cancellation actions as prose posts with tables; Queensland publishes separate suspended and cancelled pages; SA folds labour hire into a general licence search. Reading a monthly LHA post, extracting the named entities and ABNs, classifying the action type, and reconciling it against the live register state is a text-extraction job that changes shape whenever a regulator redesigns a page. A model handles that drift; a brittle parser breaks and you don’t find out until a customer gets fined.

Remove the AI and you have a link-checker that emails you when a page changes — which is what the free tools already are, and why they don’t clear the bar.

7. Localization angle

This is the localization play. Australian labour hire licensing is a four-state patchwork with no federal equivalent — Victoria, Queensland, South Australia and the ACT each have their own Act, own regulator, own register, own definition of “labour hire services,” and non-portable licences. NSW, WA, Tasmania and NT have no general scheme, which itself matters: a national employer needs to know which of its sites are in scope.

The UK’s Gangmasters and Labour Abuse Authority runs a comparable licensing regime and offers a formal “Active Check” service to labour users, which is a useful proof that host-side verification is a real category. But no global product can serve Australia, because the entire value is in the four specific registries and the interaction between them. Pricing must be AUD, and the buyer’s mental benchmark is Cm3’s $399–$3,045/yr contractor fee — which makes a few hundred dollars a month feel proportionate against a $669,120 maximum.

8. Business model — path to $1M–$5M ARR

• Pricing: AUD $199/mo for up to 10 watched suppliers (single state), $449/mo for up to 40 suppliers across all states, $899/mo for multi-entity groups with per-site supplier lists and API/SSO. Annual prepay at 10 months.
• ACV: ~AUD $5,400 blended (weighted to the $449 tier).
• Rough math to $1M ARR: 185 customers × $5,400 = $999K. That is 185 Australian businesses out of a scope population in the thousands.
• Rough math to $5M ARR: 700 customers at a higher blended ACV ($7,100), which requires two things to be true: (a) expansion into adjacent host-side licence monitoring — security licensing, RTO/ASQA status, building practitioner registration, contractor insurance currency — so the per-customer supplier count and price both rise; and (b) a channel relationship with the labour hire industry bodies or a mid-market insurance broker network. Australia alone probably tops out nearer $3M; the $5M case assumes the adjacent-registry expansion.
• Expansion path: more suppliers watched → more registries watched per supplier → per-site/entity seats for multi-site groups → the evidence pack as a paid artefact when an inspector or insurer asks.

9. Go-to-market wedge — first 100 customers

1. Mine the regulator’s own kill list, backwards. Every month the LHA names providers whose licences were refused or cancelled, with ABNs. Cross-reference each named provider against ASIC/ABN records, LinkedIn, and the provider’s own site to identify the industries and regions they served, then approach the plausible hosts in those segments with a specific, dated message: “Provider X’s licence was cancelled in June. If you used them, here’s what your exposure looks like.” This is free, recurring, and arrives with news the prospect cannot get anywhere else.
2. The SA transition list. SA’s all-industry expansion ended its grace period on 29 July 2026. Pull SA business directories and industry association member lists for the newly-caught sectors — labour-using manufacturers, food processors, transport, hospitality groups, aged care — and run a free “are your suppliers licensed in SA?” audit on their publicly-identifiable suppliers. Lead with the finding, not the pitch. Target 500 audits, 15% meeting rate.
3. Employment-law firms and workplace-relations consultants as referrers. Every one of them is currently publishing the manual checklist — “re-check the register at renewal points,” “check the register the day after expiry” — as free content. That’s a firm telling you it has a client need it can’t productise. Offer a co-branded version and a referral fee. Piper Alderman, Sprintlaw, and the mid-tier WR consultancies are all publishing on this right now.
4. Industry associations in the newly-caught verticals. Ai Group, state master builders associations, food and grocery councils, and the cleaning and security industry bodies all ran member alerts on the 2026 changes. Sponsor the member webinar, give the association a free group licence, get the member list in front of you.
5. Construction, timed to the Victorian amendment. When the VIC supply-chain/construction definition changes commence (on or before 1 October 2026), head contractors become liable for labour hire engaged indirectly through subbies. Cold outreach to VIC head contractors in the weeks around commencement, with the Rocktown penalty as the subject line.

10. Build complexity — justification

Low. Four public registers, no authentication required, no partner approval to obtain, no regulated data. The real work is monitoring infrastructure that degrades gracefully (registers change layout, rate-limit, or go down), the entity-resolution model with a human-review queue, and alerting that doesn’t cry wolf. No custom models — off-the-shelf LLM APIs handle extraction and matching. A solo technical founder ships a credible v1 in 6–8 weeks; a pair does it in 5 with a better evidence-pack export. The ongoing cost is a scraping/monitoring fleet and modest inference, both trivially covered at $199/mo per customer.

The one thing to build carefully from day one is the audit trail. The product’s job at the worst moment is to prove the customer checked. That record has to be tamper-evident and exportable, or the whole value proposition collapses under questioning.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Reads public government registers; no licence or approval needed to operate. Position as a monitoring tool, not legal advice — carry the same disclaimer the free checkers do.
Ethical — no harm / dark patterns	✅	Increases the odds workers are supplied by licensed providers. Directly aligned with the scheme’s worker-protection purpose.
Market exists (evidence above)	✅	Hosts actually fined ($101,774, Rocktown). 1,000+ providers removed since 2019. 52 licence actions in one month.
1–5 person team can build this	✅	Solo founder, 6–8 weeks to v1.
Launchable with <$50K / ₹40L	✅	Well under. Infrastructure and inference are the only real costs.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	17/20	$669,120 maximum with no knowledge defence, and a named host already fined $101,774. Not daily pain, but the exposure is continuous and the trigger is invisible. Held below 18 because most hosts have never been audited and dormant risk converts worse than a monthly invoice.
Demand evidence	15	12/15	Strong regulatory and enforcement evidence; lawyers publishing the manual workaround is a good proxy. Weak on direct customer voice — I found advisor commentary, not host complaints in their own words. That gap is the reason confidence is Medium.
Build feasibility	15	13/15	Public registers, no auth, no approvals. Docked for four-way registry drift and entity-resolution accuracy being genuinely load-bearing.
Distribution clarity	15	12/15	The monthly cancellation list is a free, recurring, named lead source with a real reason to call. Docked because the buyer is a mid-market ops/HR manager who is not searching for this.
Revenue mechanics	15	11/15	$199–$899/mo is defensible against a $669K maximum and Cm3’s $399–$3,045/yr contractor fee. $1M needs 185 customers — achievable. $5M needs the adjacent-registry expansion to work, which is unproven.
Time to first revenue	10	8/10	Pre-sellable off a free SA supplier audit; 6–8 week build; realistic first paid customer inside 10–12 weeks.
Defensibility	10	4/10	The honest weak axis. Public data, copyable product. The moat is accumulated resolved-entity mappings, alert-quality tuning, and the evidence archive customers won’t want to restart. Victoria could extend “Follow my Providers” nationally in concert with other states — unlikely, given they’re separate jurisdictions, but it’s the structural risk.
Total	100	77/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · sales-heavy

Technical because entity resolution and resilient multi-registry monitoring are the product. Sales-heavy because this buyer will not find you — outbound off the regulator’s own published lists is the entire first year.

Key assumptions to validate (3–5)

1. Assumption: Mid-market hosts perceive licence lapse as a live risk rather than a theoretical one, and will pay $199–$449/mo to remove it. How to test: 40 discovery calls with ops/HR managers in VIC and SA labour-using businesses. Lead with the Rocktown penalty and the June cancellation count; measure how many ask “how would I even know?” versus “our supplier handles that.”
2. Assumption: Entity resolution across the four registers can hit ≥95% precision on real supplier lists without heavy manual curation. How to test: get 10 real AP supplier lists, run the matcher, hand-verify every result, measure precision and recall and the size of the review queue.
3. Assumption: Victoria’s free “Follow my Providers” doesn’t neutralise the offer for VIC-only businesses. How to test: in discovery, ask VIC hosts whether they use it and whether they know it exists. If VIC-only prospects consistently say “we already have this covered,” pivot the wedge to SA and QLD and to multi-state operators, and re-check whether the remaining market is big enough.
4. Assumption: The Victorian supply-chain/indirect-engagement amendments actually commence on or before 1 October 2026 as the LHA states. How to test: track the commencement notice directly at the LHA and in Victorian legislation, not via vendor blogs. If it slips, one of the three timing signals weakens.

Risk flags

1. Regulator-kills-the-category: Any of the four regulators could ship Victoria-style free alerts, and a national harmonisation push would hollow out the multi-state value. Mitigate by moving fast into adjacent registries (security licensing, RTO status, building practitioner registration) so the product is a supplier-credential screen, not a labour-hire-only one.
2. Platform dependency on scraping: Four government sites with no API and no stability guarantee. Layout changes, rate limits or a robots policy shift break ingestion. Mitigate with per-registry redundancy, change detection, and a loud internal alarm when a registry goes stale — silently reporting “no change” when ingestion is broken is the failure mode that gets a customer fined.
3. Dormant-risk conversion: The single biggest commercial risk. Compliance products sold against a penalty nobody has personally experienced convert badly. The counter is the monthly cancellation list — turning an abstract fine into “your supplier was cancelled in June” — but if that doesn’t land in discovery, the pricing has to come down or the wedge has to change.
4. Market ceiling: Australia-only, four states, mid-market. $3M ARR is a realistic ceiling without the adjacent-registry expansion. Fine for a bootstrapper; wrong for anyone wanting more.

14. Structured verdict

Score:                  77/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Technical solo founder or pair, Australia-based or with AU
                        employment-relations network; comfortable with outbound sales
                        into mid-market ops and HR
Time to revenue:        10–12 weeks
Capital to launch:      AUD $8–15K ($5–10K USD)
Top 3 assumptions to validate first:
  1. Hosts will pay $199–$449/mo to remove licence-lapse risk — 40 discovery calls
     with VIC and SA labour-using ops/HR managers, lead with the Rocktown penalty
  2. Entity resolution hits ≥95% precision across four registers — run 10 real AP
     supplier lists through the matcher and hand-verify every match
  3. Victoria's free "Follow my Providers" doesn't already satisfy VIC-only buyers —
     ask every VIC prospect whether they use it and whether they knew it existed
Kill criteria:
  - Abandon if fewer than 6 of 40 discovery calls say they have no current process
    for re-checking supplier licences after the first engagement
  - Abandon if entity-resolution precision stays below 90% after two iterations on
    real supplier lists — an unreliable alert is worse than no alert here
  - Abandon if two or more of the four regulators ship free multi-provider alerting
    before v1 ships

15. Next step — 1-week validation sprint

• Day 1–2: Build the supplier-list matcher against all four registers as a throwaway script. Pull the last six months of Victorian monthly licensing-action posts and the standing Queensland suspended/cancelled lists into one table. Count how many distinct providers were actioned. That number is the pitch.
• Day 3–4: Pick 25 SA and VIC labour-using businesses in the newly-caught sectors. Identify their labour suppliers from public sources — site signage, LinkedIn, job ads, tender documents, ASIC records. Run each supplier through the matcher. Send every business a free one-page audit with the actual findings, including the specific unlicensed or cross-state-gap exposures found.
• Day 5: Count replies and book calls. Ask each respondent one question directly: “What’s your current process for re-checking a supplier’s licence after you’ve engaged them?”

Go/no-go: Go if ≥5 of 25 audited businesses reply, and at least one audit surfaces a real exposure (a supplier with a lapsed, cancelled, or wrong-state licence), and ≥6 of the discovery answers amount to “we don’t have one.” No-go if hosts consistently answer that the supplier’s contract or their prequalification platform already covers it — that means the risk is perceived as transferred, and a monitoring tool sold to the host is fighting a belief rather than a gap.