QuoteShield — AI underquoting compliance copilot for Australian real-estate agents

1. One-liner

AI Statements of Information and weekly price reviews built to survive a Fair Trading audit.

2. Trend signal — why now?

Australia’s two largest property markets just made underquoting a six-figure problem.

• NSW (March 2026): New legislation lifts the maximum underquoting penalty from $22,000 to $110,000 (or 3× commission, whichever is greater), mandates a price or price guide on every advertisement, makes the Statement of Information (SOI) compulsory, and requires agents to review and revise the estimated selling price weekly with written notice to the seller plus evidence to support the revision. NSW Fair Trading also gains expanded powers to demand independent valuer verification.
• Victoria: The CAV underquoting taskforce has issued 203 fines totalling over $2.3M since September 2022, with 5,000+ complaints filed and 2,800+ campaigns monitored. CAV has publicly flagged that agents repeatedly choose comparable properties that “significantly differ in age, size, location, build type, renovation status or school zone.” From 2026 reserve prices must be published 7 days before auction.
• Industry voice: Tom Panos (top AU real-estate trainer) is openly framing 2026 as the year agents must use “AI as a first employee.” Buyer’s agent Paul Mulligan (NSW) publicly states “every single property I go through in Sydney is underquoted by 15–25%.”

The fines are no longer theoretical. 95 Sydney agents have already been fined, and one Melbourne agency was hit with a major fine for quoting “more than 30 per cent below” the actual reserve. Across NSW + VIC + QLD there are ~170,000 licensed agents and ~8,400 agencies — every one of them now needs documented, defensible price guides every week.

No purpose-built compliance tool exists for this. CRMs (Reapit, AgentBox, MRI Vault, Box+Dice) handle listings and trust accounting. CoreLogic/RP Data and PropTrack sell raw comparables. Realtair sells digital proposals at $189/agent/mo. None of them generate audit-ready SOIs with defensible comparable-selection reasoning, and none keep the weekly review trail that NSW Fair Trading will demand on inspection.

Provenance:

• Signal 1: NSW lifts underquoting penalty from $22K to $110K, makes SOI mandatory and requires weekly price reviews — https://www.nsw.gov.au/ministerial-releases/nsw-cracks-down-on-underquoting-tough-new-laws — March 2026
• Signal 2: Victoria CAV taskforce: 203 fines, $2.3M+, 5,000+ complaints, comparable-selection cited as recurring failure — https://www.consumer.vic.gov.au/latest-news/tackling-underquoting-through-new-comparable-property-guidelines — 2025–2026
• Signal 3: 95 Sydney agents fined for misleading price guides; Tom Panos / industry framing AI-first compliance for 2026 — https://eliteagent.com/underquoting-101-avoid-fines-build-trust-and-do-it-right/ — 2026 Category: Regulatory arbitrage

3. The opportunity

The reform effectively changes the unit of work for an Australian sales agent. A listing is no longer “one appraisal + a price guide on REA”; it’s now an evidence-backed SOI at sign-up + weekly written price reviews + a defensible audit trail kept for years. The economics: a single $110K fine wipes ~5 average commissions ($900K listing × 2.2% = ~$19,800). Principals personally carry the licensing risk. They will pay to insure against it.

Incumbents have the wrong shape:

• CoreLogic / PropTrack / Domain sell data, not compliance. Their reports list comparable sales but don’t justify which three to use, don’t generate the SOI document, don’t track weekly reviews, and don’t produce a CAV-acceptable audit trail.
• Reapit / AgentBox / MRI Vault are CRMs. Listing fields, contact records, signed agreements. The SOI sits as a PDF attachment, manually filled.
• Realtair does digital proposals and signing — not the underlying compliance brain.
• REI Forms Live (REINSW members) provides the form template only; the agent still does the work.

QuoteShield is the missing layer: feed it the property + listing brief, it pulls comparables from RP Data / Pricefinder / Domain, ranks them on similarity (build year, land size, beds/baths, school zone, renovation status, distance), generates the SOI with prose-justified comparable selection, runs a weekly auto-review against fresh data and triggers a written notice to the vendor, and stores everything so a Fair Trading inspector gets a clean audit pack on demand. AI does the comparable reasoning and the prose; the agent reviews and signs.

4. Target market

• Primary customer: Principals and licensees-in-charge at independent and small-network sales offices in NSW, VIC, QLD, WA. Sweet spot: 5–25 sales agents, 30–150 active listings at any time, $10–60M total transaction volume per year. Decision-maker is the principal who personally carries the licence and the fine.
• Why they buy: “We just got told we need a Statement of Information for every listing in NSW now, weekly price reviews in writing to vendors, and a $110K fine if we get caught short. Our ‘process’ today is two long-time agents pulling comparables in their head from RP Data. If Fair Trading walks in tomorrow we have nothing to show them.”
• Rough TAM reasoning: ~8,400 agencies in Australia (NSW 2,563, QLD 2,110, VIC 2,027). At $200/mo blended ARPM the addressable market is ~$20M ARR. Realistic 25–40% penetration of NSW + VIC over 3 years = $5–8M ARR. Adjacent expansion to NZ (similar legal framework, ~5,000 offices) takes the ceiling higher.
• Why now for them: NSW law lands in 2026 with weekly review obligations they cannot meet manually; Victoria CAV is actively writing tickets; agency principals just heard “$110K per breach” and went looking for a tool.

5. Product sketch (MVP)

• Listing intake. Paste a REA/Domain URL or fill a 10-field form. Pull comps automatically from RP Data / Pricefinder / Domain.
• AI comparable ranker. Pick the 3 best comparables, score each on build-year, land size, internal area, beds/baths, school zone, distance, sale recency. Surface the next 7 also-rans the agent can swap in.
• One-click SOI generator. State-correct PDF (NSW / VIC variants) with prose-justified comparable selection that reads like a human wrote it.
• Weekly auto-review. Every Monday, refresh comps and recompute the estimated price. If movement >X%, draft the written notice to the vendor and queue it for the agent’s signature.
• Audit pack export. One-click “Fair Trading inspection” pack: every SOI version, every weekly review, every vendor notice, timestamped, for the last 6 / 12 / 24 months.
• Outlier alerts. Flag listings where the published price guide is materially below the agent’s own estimate — the exact pattern CAV uses to write tickets.
• Vendor portal (lite). Vendor receives a branded weekly summary: estimate, market movement, comparable activity. Signed acknowledgements stored.
• Multi-office dashboard. Principal sees all listings, all reviews due / overdue, current compliance score per agent.

6. AI angle — what’s load-bearing

AI does three jobs no template can:

1. Comparable selection reasoning. This is the exact failure CAV calls out. Ranking + writing a defensible paragraph explaining why these three properties were chosen and why the rejected ones weren’t — is what an LLM with structured property attributes does well today and didn’t do reliably 18 months ago.
2. Weekly price-review narrative. Each week the model must read the new comparable transactions, decide whether to revise, and draft a written explanation to the vendor. Pure template engines can’t do this; valuers cost too much.
3. Audit-pack synthesis. When Fair Trading asks “show me how you arrived at this price guide on Aug 12, 2026,” the system reconstructs the comp set as it existed that day, the ranker’s output, and the prose used in the SOI.

Strip the AI out and you have a forms-and-storage tool — useful but a $20/mo product. With the AI it’s an evidence engine principals will pay $200–400/mo for.

7. Localization angle (if any)

This is a deliberately localized play. Three of the four moats are local:

• Form formats: NSW, VIC, QLD, WA each have different SOI / price-guide rules, comparable-radius rules (e.g. VIC metro = 2km / 6 months), and disclosure timing. Generic global proptech can’t ship these correctly.
• Data integrations: RP Data / Pricefinder / Domain / PropTrack are AU-only data sources. Connecting cleanly is gatekept.
• Distribution: REINSW, REIV, REIQ are the channels. AREC (Australasian Real Estate Conference) is the deal-flow event. Tom Panos and Elite Agent are the influencer surface.

Adjacent expansion: NZ (REINZ) shares the legal pattern. UK / Singapore differ enough to be separate products.

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

• Pricing: Three tiers, AUD/mo, billed per office (not per agent — agencies hate seat sprawl):
  • Solo / 1–2 agents — $149/mo
  • Boutique / 3–10 agents — $299/mo
  • Network / 11–25 agents — $549/mo
  • Enterprise (50+ agents, multi-office) — custom $1,500–4,000/mo
• ACV: ~$3,500 blended.
• Path to $1M ARR: ~285 paying offices × $3,500 = $1M. ~3% of the AU agency market. Doable in 12–15 months with a domain-expert founder.
• Path to $5M ARR: ~1,400 offices = ~17% of AU agencies. Doable in 24–30 months once REINSW / REIV partnerships are in place. NZ adds runway.
• Expansion path: Add (a) AML/KYC compliance module — agencies face a separate AML deadline July 2026, and the same principal-liability profile makes the same buyer pay; (b) Buyer’s-agent module (the 1,000+ buyer’s agents are also under price-guide scrutiny); (c) Vendor-disclosure portal upsell at $20/mo per active campaign.

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

• Distribute through REINSW + REIV CPD channels. Members must complete CPD hours; bundle a 30-min “How to survive a Fair Trading SOI inspection” webinar in exchange for member discount and listing in the REINSW/REIV partner directory. Email list = ~12,000 NSW members + ~4,000 VIC.
• Tom Panos / Elite Agent paid sponsorship. Single Tom Panos podcast feature + Elite Agent ad placement reaches the principal demographic at scale. Cost ~$15K AUD for placement, target 25–40 sign-ups.
• Cold outbound to the 95 Sydney agencies already publicly fined. Public list of fined offices via NSW Fair Trading. Personalised “you got fined for X — here’s how QuoteShield would have caught it” 2-min Loom. ~25% reply rate, 5–10% close = 5–10 logos.
• AREC26 conference booth + workshop. Australasian real-estate conference in May/June. ~3,000 attendees, mostly principals. Cost ~$25K AUD all-in. Realistic 30–80 trial signups.
• Partner with a Tier-2 valuer firm. Independent valuers want to be the “human escalation” for borderline cases. Refer-out partnership gives QuoteShield a credibility moat (“verified by a registered valuer when you escalate”) that competitors can’t slap on overnight.

That’s a clear path to ~150 paying offices in months 6–12 across these channels.

10. Build complexity — justification

Medium. Off-the-shelf: Next.js + Postgres + LLM APIs (Claude / GPT-4o) + PDF generation + Stripe. Custom: integrations to RP Data, Pricefinder, Domain (Domain has a free public tier; CoreLogic/Pricefinder are paid B2B partner deals — 4–8 weeks of contract work). The comparable-ranking model is rules + LLM, not custom training. Estimated 4–5 months for a 2-person team to ship a credible v1 covering NSW + VIC. Domain/CoreLogic deal terms and form-accuracy iteration are the real risk, not the engineering.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Helps agents comply with their own legal duties; doesn’t replace licensed valuers.
Ethical — no harm / dark patterns	✅	Reduces underquoting → benefits buyers and honest agents.
Market exists (evidence above)	✅	$2.3M in fines already issued in VIC; 95 fined in NSW; reform just enacted.
1–5 person team can build this	✅	2 engineers + 1 domain advisor for v1 in 4–5 months.
Launchable with <$50K / ₹40L	✅	~$30–40K AUD: data partnership setup + initial dev + AREC booth.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	18/20	$110K fine per breach, mandatory weekly review, principal personal liability. Hair-on-fire for any office that’s been fined or knows someone who has. Slight discount: not every agent feels imminent risk yet.
Demand evidence	15	14/15	$2.3M in fines paid, 5,000+ complaints, 95 Sydney agents already fined, fresh NSW law. Multiple independent signals.
Build feasibility	15	12/15	Std web stack + LLMs + 3rd-party data integrations. Real friction = CoreLogic/Pricefinder partner contracts. ~4–5 months.
Distribution clarity	15	12/15	REINSW/REIV directories, AREC26, Elite Agent / Tom Panos, public fined-agency list. Concrete channels with named lists. Slight discount: still a sales-cycle business, not viral.
Revenue mechanics	15	13/15	$149–549/mo per office is small relative to $110K fine and ~$20K commission per deal. ACV math holds. ~285 offices to $1M ARR is achievable.
Time to first revenue	10	8/10	First paying agency in 4–6 weeks of soft launch (pre-sale to 5–10 design-partner principals).
Defensibility	10	6/10	Soft moats: data-partner contracts, REINSW/REIV alignment, accumulating audit trail per office (workflow lock-in). Not a deep moat — a US/global proptech could clone it given 12 months. Speed wins.
Total	100	83/100

13. Qualitative modifiers

Founder-fit tags

domain-expertise-required · sales-heavy · technical-heavy

Best-fit founder team: one ex-real-estate-principal or proptech operator (must speak the buyer’s language and have a starter Rolodex of NSW/VIC principals), one full-stack engineer comfortable with LLMs and 3rd-party data integrations, one part-time compliance advisor (ideally an ex-CAV / Fair Trading lawyer for form accuracy).

Key assumptions to validate (3–5)

1. Assumption: AU agency principals will pay $200–400/mo per office for a compliance copilot that’s not in their existing CRM. How to test: Pre-sell 30-day trials to 25 NSW principals at AREC26 or via cold outbound; target 5+ paying conversions at full price.
2. Assumption: RP Data / Pricefinder / Domain will sign a B2B data partnership at unit costs ($5–25/lookup) that fit a $200/mo product. How to test: Ring all three in week 1 with a one-pager; get written terms or rate cards inside 30 days.
3. Assumption: The AI-generated SOIs and comparable justifications hold up under a CAV inspection. How to test: Pay an ex-CAV underquoting investigator $5K AUD to grade 20 generated SOIs against real fined cases. >85% pass rate or kill.
4. Assumption: The weekly price-review notice replaces a real workload, not an imaginary one — agents are doing this manually today (or not doing it and getting fined). How to test: 30 in-person interviews with NSW principals in May–June 2026; ask “show me what you do today, in your CRM” rather than “would you buy this.”

Risk flags

1. Platform / data dependency risk. CoreLogic + Domain control the underlying comparable-sales data. If either declines to license at workable rates, MVP stalls. Mitigation: start with Domain (cheaper, public-tier API) and pricefinder.com.au; CoreLogic as upsell.
2. Regulatory shift risk. Fair Trading guidelines could move (e.g. mandate human valuer signoff over a threshold). Mitigation: partner with a valuer network early so you have human-in-loop available the day a rule changes.
3. Incumbent retaliation risk. Reapit / MRI / AgentBox could ship a mediocre SOI module bundled into existing CRMs at zero marginal price. Mitigation: ship faster, win the principals first via REINSW/REIV channels, build the audit-trail moat that’s painful to migrate off.
4. Founder fit risk. A pure technical founder will not get past gatekeeper principals. This is a sales-heavy domain.

14. Structured verdict

Score:                  83/100
Verdict:                STRONG GO
Confidence:             High
Best-fit builder:       Ex-real-estate-principal or AU proptech operator + 1 senior engineer + part-time compliance advisor
Time to revenue:        4–6 weeks post-MVP (pre-sold design partners)
Capital to launch:      AUD $35–50K (data partnerships, dev, AREC booth, compliance advisor retainer)
Top 3 assumptions to validate first:
  1. Principals will pay $200–400/mo per office — pre-sell 25, convert 5+ at full price.
  2. RP Data / Pricefinder / Domain partner terms fit unit economics — written rate cards in 30 days.
  3. AI-generated SOIs hold up under CAV scrutiny — ex-CAV investigator scores 20 generated SOIs, >85% pass.
Kill criteria:
  - Abandon if <5 of 25 principal pre-sale conversations convert to paid trial in 30 days.
  - Abandon if no data partner gives workable rates within 60 days (cost per lookup > $30 AUD).
  - Abandon if Reapit / MRI ship a free SOI + weekly-review module bundled into the dominant CRM before MVP launch.

15. Next step — 1-week validation sprint

• Day 1–2: Pull the public NSW Fair Trading list of fined agents from 2024–2026. Build a contact list of 50 principals at fined or near-fined agencies. Email all three of CoreLogic, Pricefinder, Domain for partner-API rate cards.
• Day 3–4: Cold outreach (Loom + email) to the 50 principals: “you / a peer got fined for X. Would you pay $300/mo for a tool that auto-generates SOIs and weekly price reviews you can show Fair Trading?” Goal: 10 calls booked.
• Day 5: Run the 10 calls. Ask the falsifiable question: “If I shipped this in 8 weeks, would you pre-pay 3 months at $300/mo today?” Decide go / no-go based on: ≥3 of 10 say yes (12% pre-sale conversion at the principal level) → GO. ≤2 → re-shape (target buyer’s agents, or pivot to AML compliance for the same buyer).