WindSnap — Florida wind mit photo cockpit

1. One-liner

Phone-first cockpit that validates every wind mit photo against Florida’s new OIR-B1-1802 form on-site.

2. Trend signal — why now?

Florida Office of Insurance Regulation (OIR) put a brand-new version of the Uniform Mitigation Verification Inspection Form OIR-B1-1802 (Rev. 04/26) into force on April 1, 2026 — the first major revision in over a decade. The new form rewires the rules of the game:

• “At least one photograph or document providing proof must accompany this form to validate each attribute marked in questions 2 through 9.” (Hope Home Inspections, FLOIR-adopted form). hopehomeinspections.com
• New required fields: Region (design wind speed), Roof Slope, retrofit categories, permit dates, product approval numbers, installation years. guardianangelinspections.com
• Carriers begin applying credits from the new form starting July 2026 — meaning the back-half of 2026 is when every Florida homeowner with an expiring policy needs a fresh packet that survives underwriting. cfbinspect.com
• Citizens Property Insurance shed 541,000 policies out of its book during 2025 (936K → 395K). Those policyholders are landing at private carriers and 17 new entrants to the FL market — all of whom underwrite fresh, by the new form. Florida Realtors · flgov.com
• Citizens has been kicking back reports already. From a high-volume FL wind mit inspector on the InterNACHI forum: “Citizens kicked back the report requiring roof slope images. Nowhere on form OIR-B1-1802 does it state that roof slope images are required.” (paraphrase from the active “Wind Mitigation Inspection Photo Requirements. Roof Slope Images?” thread). The new form does require roof slope. Backlash is already happening.
• Average FL homeowner premium: ~$7,100/yr. Wind mit credits commonly cut 30–45% off that. Every kicked-back report = thousands of dollars stuck for the homeowner and a re-do for the inspector.

Provenance:

• Signal 1 (demand): Citizens shed ~541K policies in 2025 + 17 new FL carriers entered; every shifted policy needs a fresh, new-form wind mit to lock in discounts. Inspectors and homeowners both feel the rush. — https://www.floridarealtors.org/news-media/news-articles/2026/01/citizens-policies-plummet-2025 — 2026-01
• Signal 2 (feasibility): Vision-language models (Gemini 2.5 Pro Vision, GPT-4o vision, Claude vision) can reliably classify roof shape (hip vs gable), identify roof-to-wall strap geometry, read product-approval labels, and confirm shutter ratings from a photo. Costs measured in cents per image. — generic capability, validated by InspectorData shipping AI photo categorization in 2026 — 2026-04
• Signal 3 (economic / regulatory): FLOIR Rule 69O-170.0155 adopting form OIR-B1-1802 (Rev. 04/26), effective 2026-04-01, requiring “at least one photograph or document providing proof…to validate each attribute marked in questions 2 through 9.” — https://flrules.elaws.us/far/notice/30104258 — 2026-04 Category: Regulatory arbitrage (with strong Tech-unlock overlay)

3. The opportunity

The new OIR-B1-1802 isn’t a redesign — it’s a photo-evidence regime change. Pre-April, an inspector ticked boxes and submitted a PDF. Post-April, every box needs a photo (or product approval doc) that actually proves the credit. Underwriters at the 17 new carriers and the depopulated-Citizens replacements are going to be picky, because every $2K–$3K discount they grant is real money out of their book.

Incumbents do not solve this:

• Spectora ($109/mo) has a FL wind mit template. It generates a PDF with whatever photos you drop in. It does not look at the photo and tell you “this isn’t a clear roof-to-wall strap shot — re-shoot.” Spectora pricing
• InspectorData ($69.99/mo) does AI photo categorization and comment generation — but it’s a general home-inspection platform. The AI labels “this is a roof” not “this is a single-wrap strap that meets attribute 6.B on OIR-B1-1802 (Rev. 04/26).” InspectorData product
• HomeGauge, Home Inspector Pro, ISN — same problem: general field inspection tools with FL wind mit as a bolt-on report type.

The 10× wedge: an AI cockpit that, on-site, while the inspector is still on the property, validates each photo against the specific form attribute, flags missing/insufficient evidence, extracts product approval numbers from label photos, and produces a packet that survives underwriting. The cost of being wrong is real — Citizens already kicks back reports. Inspectors who hand carriers clean packets keep getting referrals; the rest become a liability.

4. Target market

• Primary customer: Licensed Florida home inspectors (DBPR-licensed under Ch. 468 Pt. XV) who run solo or 2–3-person shops and bill wind mit + 4-point as a core service line. ~3,000–4,000 active licensees in FL doing this work. Many also include unlicensed-but-permitted wind mit specialists (PE / GC / architect under FS 627.711). Typical run rate: 5–15 wind mits/week, $75–$200 per inspection.
• Why they buy: The new form costs them more time per inspection and more re-do risk. The April 2026 rollout has been chaotic — NACHI forum threads, Facebook group posts, and inspectors openly complaining that carriers are kicking back reports for missing photos that aren’t explicitly listed as required. Every kickback = unpaid hours and pissed-off agents. A tool that prevents kickbacks pays for itself in one rescued inspection per month.
• Rough TAM reasoning: 7.5M FL single-family homes × wind mit renews every 5 years = 1.5M inspections/year. At $100 avg, that’s a ~$150M annual inspector-revenue layer. Even capturing 2,000 FL inspectors at $79/mo = ~$1.9M ARR. Add 4-point cockpit upsell ($30/mo) → $2.6M ARR. Plus per-inspection charges to homeowners shopping carriers post-Citizens-exit ($5/report direct-to-carrier delivery).
• Why now for them: Form went live 41 days ago. Underwriting credits start applying in 60 days (July 2026). 541K Citizens-shedded policies need fresh wind mits between now and end of 2026. Hurricane season opens June 1. Inspector calendars are about to explode and the form is brand new — perfect adoption window.

5. Product sketch (MVP)

• Phone-first OIR-B1-1802 capture mode — open inspection, tap each attribute (roof shape, deck attachment, SWR, opening protection, etc.), camera UI prompts the exact shot needed for that attribute with on-screen guides.
• On-device photo validation — vision model scores each photo: “Is this a clear roof slope shot? Is the strap visible? Is the product label legible?” Inspector gets a green check or a re-shoot prompt before leaving the roof.
• Product approval number + permit date OCR — point camera at a shutter label, panel sticker, or permit doc; auto-extracts the approval number and installation year into the form fields.
• Form-aware photo binder — final report PDF maps every claimed credit to the exact photos that prove it. Carrier underwriter clicks one link, sees the evidence, signs off.
• Carrier-specific delivery packets — pre-built export profiles for Citizens, Tower Hill, Slide, American Integrity, the major MGAs, plus the 17 new entrants as they publish requirements.
• Re-issue cockpit — when a Citizens-shedded homeowner shops a new carrier, the cockpit re-packages the existing wind mit into the new carrier’s format and routes it.
• Inspector earnings dashboard — per-inspection P&L, time-on-roof, kickback rate, top referring agents.

6. AI angle — what’s load-bearing

Vision is doing the work. Without AI:

• Inspector takes 50 photos, hopes 35 are usable, gets kicked back when 12 of them don’t clearly prove the credit, drives back to the property.
• Inspector hand-types product approval numbers from blurry label photos, transcription errors slip in, underwriter rejects.

With AI:

• Each shot is scored on-device against a known specification (“OIR-B1-1802 Q6.B requires visible strap geometry; this image is 0.3 confidence, re-shoot from 18 inches with flash”).
• Product approval numbers / install years are OCR’d from labels in <2 seconds with confidence ≥0.95 or flagged for manual entry.
• Roof shape (hip / gable / hip-and-gable / flat / other) is auto-classified from a roof-overview shot with the inspector confirming.

Remove the AI and the product is a worse Spectora. Keep the AI and it’s the only tool that prevents kickbacks before you leave the property.

7. Localization angle (if any)

This is the localization angle. Florida-specific regulatory form, Florida-specific carrier list, Florida-specific homeowner pain. Could expand to neighboring hurricane states (LA, TX, MS, AL, SC, NC) but their forms differ and credit structures differ. Stay Florida-first for at least 18 months. Once dominant in FL, the same playbook copies to TX wind/hail and LA fortified-roof discount inspections.

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

• Pricing:
  • Solo Inspector — $79/mo unlimited wind mits + 4-point. Annual $790. Undercuts Spectora ($109) and slots next to InspectorData ($70) but with specialization premium.
  • Team (2–3 inspectors) — $149/mo + $39 per additional seat.
  • Per-inspection pay-as-you-go — $7/report for inspectors who do <10/mo, paid from collected fees.
  • Homeowner re-routing — $9 add-on charged at delivery when an inspector re-packages an existing wind mit for a new carrier (Citizens-shedded shopper flow).
• ACV: ~$900–$1,200/customer/yr (mostly solo at $79–$99/mo).
• Rough math to $1M ARR: ~1,000 paying inspectors × $85/mo × 12 = $1.02M. Florida has 3K–4K active inspectors doing wind mit. 25% market penetration in 24 months is plausible given the regulatory forcing function.
• Rough math to $5M ARR: 2,500 FL inspectors × $90/mo + $1.2M in per-inspection homeowner re-routing fees + $400K from neighboring state expansion (TX, LA). All achievable inside 36 months if month-1-12 sets the FL category default.
• Expansion path: ⬆ seats per shop, ⬆ usage-based reroute fees, ⬆ adjacent products (4-point cockpit, mitigation grant paperwork for My Safe Florida Home program, roof age letters), ⬆ adjacent geographies (TX wind/hail, LA fortified-roof).

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

Florida wind mit inspectors all live in the same handful of digital rooms. The first 100 is a direct-outreach sprint.

• InterNACHI Florida + SUNachi forum — the dominant FL inspector community. Threads on the new form are already running. Drop a useful tool demo + free trial in the right thread, expect 30–50 trial signups in week 1.
• Florida wind-mit inspector Facebook groups — “New 2026 Wind Mitigation Form” group already exists (3,142-member group has active threads). Run a free pre-launch list, demo videos that show the photo-validation flow vs Spectora’s blank template. Expect 5–8% trial → paid conversion.
• Scrape the DBPR licensee list — public records request to FL DBPR returns every active Home Inspector with email/address. ~3K–4K names. Personalized Loom video showing their own market and a 30-day free trial. Realistic 1–2% close = 30–60 customers from email.
• Riding shotgun on FL insurance agent channels — independent agents lose deals when wind mits get kicked back. Partner with 2–3 agent associations (FAIA chapters) to offer their preferred-inspector list a discounted plan.
• My Safe Florida Home program piggyback — MSFH issues thousands of free wind-mit-eligibility inspections via approved inspectors. Get on the approved-tool list for grantee-funded inspections; that’s a built-in pipeline.

If those five channels can’t deliver the first 100 paying customers in 90 days, the idea is wrong.

10. Build complexity — justification

Medium. Off-the-shelf stack:

• React Native (or Flutter) app with offline-first photo capture
• Vision API (Gemini 2.5 Pro or GPT-4o vision) for photo validation; OCR via same model or AWS Textract for product labels
• PDF generation matching exact OIR-B1-1802 (Rev. 04/26) layout
• Stripe for billing, basic backend (Supabase / Postgres)
• Carrier-specific export modules — these are the only semi-bespoke pieces; 5–10 carriers in v1

A 2-person team (one mobile-leaning, one backend/AI) can ship a workable MVP in 10–14 weeks, with the next 6 weeks spent on real-inspector calibration of the vision validators. The vision-validation prompts and the form-attribute mapping are the moat — both are buildable but require sitting next to a real inspector for a week to calibrate.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Tool assists licensed inspectors; does not perform inspection or sign forms. Inspector remains the licensed signer.
Ethical — no harm / dark patterns	✅	Improves accuracy of regulatory submissions; in homeowners’ favor.
Market exists (evidence above)	✅	Active form transition, 3K–4K licensed inspectors, 1.5M inspections/year, $150M revenue layer.
1–5 person team can build this	✅	2 builders, ~14 weeks to MVP.
Launchable with <$50K / ₹40L	✅	<$20K covers vision API credits, dev tools, MVP infra, and first 90 days of customer-dev travel.

All five gates pass.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	17/20	Kickbacks = unpaid hours + reputational risk. Inspectors already complaining publicly. Cost-per-mistake is measured in hundreds of dollars and lost agent referrals. Hair-on-fire today, every day, for thousands of FL inspectors.
Demand evidence	15	13/15	Brand-new form, dated regulatory rule, public inspector complaints, 541K Citizens shifts, 17 new carriers, NACHI forum threads. Multiple converging signals; skeptic would nod.
Build feasibility	15	11/15	Mobile photo capture + vision API + PDF + carrier exports = ~14 weeks for a tight pair. Calibration of vision validators against real-world roof photos is the only true risk; mitigated by sitting with inspectors.
Distribution clarity	15	12/15	InterNACHI forum, FL FB groups, DBPR public list, MSFH program piggyback, agent associations. Named lists, named channels, plausible conversion math. Not a 2-week sprint to 100 — closer to 60 days.
Revenue mechanics	15	12/15	Pricing benchmarked against Spectora ($109) and InspectorData ($70). Inspectors already spend $80–$150/mo on report software. Per-inspection homeowner-reroute fee adds expansion. $1M ARR needs ~1K paying inspectors — credible at 25% market share.
Time to first revenue	10	8/10	4–6 weeks of inspector pilots → paid conversion. Annual upfront plans push some revenue earlier.
Defensibility	10	7/10	The vision-validation prompt library + form-attribute mapping + carrier-specific exports is real workflow lock-in. Not a patent moat but a 6–12 month head start that compounds with each new carrier integration. Spectora/InspectorData can copy but their generic platform shape works against a deep FL-only specialist.
Total	100	80/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · domain-expertise-required — Mobile + vision AI dev, plus a working relationship with one or two real FL inspectors for calibration and design partnership. A solo technical founder with a willing inspector advisor (perhaps revenue share or co-founder equity) is the ideal shape.

Key assumptions to validate (3–5)

1. Assumption: FL wind mit inspectors will pay $79/mo for a wind-mit-only specialty tool in addition to keeping their existing Spectora/HomeGauge subscription. How to test: Interview 30 active FL inspectors; show the photo-validation demo flow; ask price/willingness specifically. Kill if <40% would pay $79 net-new.
2. Assumption: Vision-language models can reliably classify roof-to-wall connection types (single wrap / double wrap / clips / toe-nail) from real on-site photos at ≥85% precision. How to test: Build a 500-image labeled set from public FL inspection portfolios and NACHI forums, benchmark Gemini 2.5 Pro vs GPT-4o vs Claude on the classification task. Kill if best model is below 80%.
3. Assumption: Citizens / major FL carriers will accept the cockpit’s exported PDF without protest. How to test: Submit 10 real packets via 3 different carrier intake channels; monitor approval-vs-kickback rate over 30 days. Kill if kickback rate is no better than current paper baseline.
4. Assumption: The “homeowner re-routing” SKU ($9/packet to repackage for a new carrier when Citizens-shed) generates real volume. How to test: Run a 30-day pilot with 3 large independent agencies; count re-route requests per agency. Kill if <50 re-routes/month/agency.

Risk flags

1. Regulatory drift: FLOIR could amend the form again, or specific carriers could publish kickback rules that make the validation prompts obsolete overnight. Mitigate by versioning the form templates and publishing prompt updates quickly (weekly cadence in year 1).
2. Incumbent response: InspectorData has AI photo categorization at $70/mo. They could ship a FL-wind-mit-specific package in 90 days. Mitigate by going deep on carrier-specific exports and the homeowner-reroute SKU they’re unlikely to bother with.
3. Vision-model error tail: Misclassifications could cause an inspector to claim a credit the photo doesn’t actually support, then get kicked back anyway. Mitigate with conservative thresholds and inspector-final-sign-off UX; never auto-claim a credit without the inspector tapping “I confirm.”
4. Hurricane-season volatility: If a major storm hits in 2026, the entire FL insurance market re-prices in real time and the carrier list churns. That’s actually accretive to the product (re-route SKU spikes) but it changes integrations week to week. Need to staff for support volume.

14. Structured verdict

Score:                  80/100
Verdict:                STRONG GO
Confidence:             High
Best-fit builder:       Technical solo founder (mobile + vision-AI experience) with a FL home inspector as design partner / advisor / revenue share.
Time to revenue:        6–10 weeks from MVP to first paid pilot.
Capital to launch:      $15K–$25K (vision API credits, dev tooling, travel to FL inspector ride-alongs).
Top 3 assumptions to validate first:
  1. WTP — 30 inspector interviews → ≥40% would pay $79 net-new for the photo-validation cockpit.
  2. Vision accuracy — labeled benchmark on 500 real roof / strap / shutter photos → ≥85% precision on attribute classification.
  3. Carrier acceptance — 10 real submissions through Citizens + 2 private carriers → kickback rate cut in half vs paper baseline.
Kill criteria:
  - Abandon if 30 inspector interviews yield <30% would-pay-$79 + a strong "I'd just use Spectora's template" response.
  - Abandon if vision model precision on roof-to-wall and opening-protection attributes is below 75% after 4 weeks of prompt + dataset work.
  - Abandon if a Spectora / InspectorData FL wind mit-specific package launches before the 90-day mark and lands above $1M ARR within 6 months — at that point the window has closed.

15. Next step — 1-week validation sprint

Goal: Falsifiable answer to “will FL wind mit inspectors pay $79/mo for a photo-validating wind-mit-only cockpit, and can a vision model actually do the validation?”

• Day 1: Pull the FL DBPR home inspector licensee list (public records). Filter for Florida-only active. Identify the 200 highest-volume inspectors (by Google Reviews, web presence, MSFH-approved status).
• Day 2: Personalized Loom to 60 of them showing a click-through Figma + a real photo-validation demo on 8 sample images. Ask for a 25-minute call.
• Day 3: Conduct 8–12 calls. Run the script — WTP at $49, $79, $99. Ask which specific kickback they got most recently. Ask what they’d love to never do again.
• Day 4: Stand up a 500-image labeled set from public inspection portfolios + NACHI forum photos. Run Gemini 2.5 Pro, GPT-4o, Claude vision on the roof-shape and roof-to-wall classification tasks.
• Day 5: Decision gate.
  • Go: ≥40% of interviewed inspectors said they’d pay $79+ for the cockpit AND best vision model hits ≥80% on the benchmark.
  • No-go: Below either threshold. Park the idea, swap to neighboring-state version, or pivot to the homeowner-reroute SKU as standalone.

The validation produces a yes/no, not a vibe.