DroneDispute — AB-75 dispute copilot for roofers

1. One-liner

Roofer-branded counter-report kit that disputes insurer drone-AI non-renewals under California’s AB-75.

2. Trend signal — why now?

Three things stacked in the last 90 days and made this a real category overnight:

• California AB-75 becomes operative July 1, 2026. It bans insurers from using aerial images older than 180 days to cancel/non-renew/reduce coverage unless they verify with an in-person inspection — and it forces them to hand the homeowner the image plus an appeal channel before the termination effective date. That dispute channel did not legally exist until this bill. It does on July 1.
• The “Linda Bennett $20K” wave. Late March 2026, ABC7 LA + Yahoo Finance + Gadget Review + dozens of derivative outlets covered Linda Bennett, a Santa Ana homeowner who got a State Farm letter ordering a $20K roof replacement after a secret drone scan. Her deadline to fix it: May 1, 2026. The story exploded — same week, dronexl, Carrier Management, NPR, and United Policyholders all ran follow-ups. Search interest in “insurance drone roof appeal” went from zero to a sustained spike.
• Insurer drone-AI is now industry default, not edge case. Carriers report cataloging ~99% of US residential properties using aerial imagery + AI (per Carrier Management, Mar 2026). California has had ~400K non-renewals since 2021. CA FAIR Plan policies quadrupled since 2015. Six other states (MA, NY, GA, WA, IN, OK) have AB-75 copycat bills in committee.

The category is “consumer rebuttal of insurer machine vision” and as of May 2026 it is empty. The roofer is the natural distributor — they already do free roof inspections to win replacement work, and they already drive to neighborhoods that just got non-renewed.

Provenance:

• Signal 1 (demand): “Santa Ana homeowner says insurance company used drone to inspect her roof without telling her” — https://uphelp.org/santa-ana-homeowner-says-insurance-company-used-drone-to-inspect-her-roof-without-telling-her/ — March 2026; “Homeowners angry: eyes in the sky and AI lead to lost insurance coverage” — https://uphelp.org/homeowners-angry-eyes-in-the-sky-and-ai-lead-to-lost-insurance-coverage/ — 2025–2026; CA FAIR Plan quadrupling — https://www.ppic.org/blog/a-deeper-look-at-californias-homeowner-insurance-challenges/ — 2025.
• Signal 2 (feasibility): GPT-4 / Claude 4 vision models can now reliably classify shingle granule loss, skylights vs. damage, moss vs. fungus, dimensional vs. 3-tab — at <$0.05/image. EagleView/Nearmap historical aerial APIs are commercially available. Cheap dispute generation is finally a Tuesday-afternoon build, not a research project.
• Signal 3 (economic): California AB-75 — https://leginfo.legislature.ca.gov/faces/billNavClient.xhtml?bill_id=202520260AB75 — operative July 1, 2026. Six-state copycat slate — https://www.dig-in.com/news/state-lawmakers-will-address-insurers-use-of-aerial-images. Marketing-side commentary: “Roofers Who Offer Documented Inspections Are Winning” — https://www.marketingcode.com/insurance-ai-canceling-roofs-roofers-fight-back/ — 2026. Category: Regulatory arbitrage

3. The opportunity

The drone-AI roof scan industry is incumbent-controlled and pointed in one direction. EagleView, Cotality, Nearmap, Loveland Innovations all sell to insurers — they hand carriers automated damage classifiers that say “non-renew this one.” Their conflict of interest is structural: they cannot turn around and sell the homeowner a tool that calls their own AI wrong.

That leaves a vacuum on the defense side. Today the homeowner gets a four-page letter, a grainy aerial photo, and 30–60 days to figure out what to do. The homeowner’s only practical move is: call a roofer. The roofer climbs up with a phone, takes a few iPhone photos, types up a free-form letter on their letterhead, and prays the insurer reads it. Win rate is bad and most roofers don’t even try because the legal channel barely existed before AB-75.

AB-75 changes the rules. From July 1, the insurer must hand over the image, must accept dispute evidence, and must dispatch in-person verification or drop the action. The roofer who walks into a non-renewed homeowner’s living room with an AI-annotated counter-report citing the actual statute wins the policy save and the replacement job that follows when the roof actually does need work. DroneDispute productizes that workflow into a roofer’s daily tool.

10× wedge over status quo: 48-hour turnaround vs. weeks, AI-annotated rebuttals citing the specific defect categories the AB-75 statute requires the carrier to address, statute-aware cover letter, and a homeowner-facing dispute packet that’s actually formatted to match how insurer ombuds and the CA Department of Insurance accept evidence.

4. Target market

• Primary customer: Independent and small-chain residential roofing contractors in California, 1–15 employees, $500K–$5M annual revenue, holding a CSLB C-39 license. CSLB licenses ~290K contractors total across all classifications; tens of thousands are C-39. Initial wedge: the ~3,000–5,000 small CA roofing shops that already pay for CRMs like JobNimbus, AccuLynx, Roofr, or QuoteIQ.
• Why they buy: A non-renewal letter in their territory is the highest-intent lead a roofer can get. The homeowner is reading the letter today, has 30–60 days, and will pay anyone who can credibly help. If the roofer can save the policy via dispute, they earn trust and almost always get the actual roof job — average ticket $8K–$25K. ROI on a $99–$299/mo subscription is one converted dispute per quarter.
• Rough TAM reasoning: ~50K residential roofing contractors in the US. CA alone has ~12K active C-39 holders. Wedge of 3K paying CA shops × $199/mo = ~$7M ARR before the law goes national. AB-75 copycats in MA/NY/GA/WA/IN/OK turn that into a $25M+ ceiling within 24 months if even half pass.
• Why now for them: AB-75 operative date is 60 days out. CA non-renewal letters are landing every week. Roofers are already being told by industry trade press (“Roofers Who Offer Documented Inspections Are Winning”) that this is their next lead source. They have no tool to do it. Whoever ships first owns the category.

5. Product sketch (MVP)

• Letter intake: Roofer or homeowner uploads the insurer’s non-renewal letter (PDF or photo) — system OCRs it, extracts cited deficiencies (e.g. “granule loss,” “moss accumulation,” “missing shingles,” “tarp visible”), the carrier, the deadline, and the aerial image used.
• Image rebuttal pack: Roofer adds 6–12 ground-level photos. AI annotates the insurer’s aerial image with circles, labels, and “what this actually is” callouts (skylight vs. damaged shingle, solar mount vs. structural defect, neighbor’s tree shadow vs. moss).
• Statute-aware cover letter: Auto-generated dispute letter citing AB-75 §X, demanding in-person verification, and pointing out where the insurer’s own evidence fails the freshness test or the misclassification test.
• Roof-condition counter-report: PDF on the roofer’s letterhead — bound photos, granule-loss measurement, estimated remaining useful life, signed contractor opinion. Formatted exactly the way CDI ombuds and major-carrier review desks accept.
• Submission packet: One-click ZIP with the letter, the photos, the contractor report, and a homeowner-signed authorization. Submission instructions per carrier.
• Lead pipeline view: Each dispute becomes a row — homeowner, deadline, status, follow-up date — and connects to the roofer’s existing job pipeline so a saved policy converts into a quoted repair or replacement.
• AB-75 alert mode (Phase 2): Roofer’s homeowner customers can forward suspicious carrier letters or be subscribed to a pre-emptive aerial-imagery review.

6. AI angle — what’s load-bearing

AI does the actual analysis work. A vision model classifies what’s in the insurer’s aerial image (granule loss percentage, shingle type, skylight presence, mounted equipment, tree-shadow vs. roof-defect), cross-references the roofer’s ground-level photos, and produces specific rebuttal points the human contractor would otherwise need an hour and roofing expertise to write. A second LLM pass turns the image findings into a statute-aware dispute letter that actually cites AB-75 language. Strip the AI and what’s left is a Word template the roofer already ignores. The leverage is concentrated in the 60-second turnaround — without it, the pricing collapses because the contractor’s time becomes the bottleneck.

7. Localization angle (if any)

This is a US play, not a localization play. The wedge IS the regulatory geography: California first because AB-75 is the operative statute, then expand state-by-state as MA/NY/GA/WA/IN/OK pass copycat bills. Each state’s version will have slightly different timelines and notice requirements — DroneDispute’s job is to encode that statute matrix and keep it updated. That state-by-state regulatory tracking is itself a moat against generic roofing CRMs that won’t bother.

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

• Pricing tier 1 — Roofer Lite: $99/month. Up to 5 disputes/month. Solo roofers.
• Pricing tier 2 — Roofer Pro: $249/month. Unlimited disputes, branded reports, lead pipeline view, white-label homeowner-facing landing page. Mainstream tier.
• Pricing tier 3 — Multi-shop: $599/month. 3+ users, custom letterhead, integrations to JobNimbus / AccuLynx / Roofr / QuickBooks. Probably 15–20% of customers.
• Per-dispute add-on: $19/dispute beyond plan limits. Margin booster for high-volume shops.
• ACV blended: ~$2,800/yr.
• Math to $1M ARR: ~360 paying roofer accounts. Reach by month 12 with concentrated CA distribution.
• Math to $5M ARR: ~1,800 paying accounts. Requires expansion into 2–3 copycat states by month 18 plus mid-market multi-shop segment.
• Expansion path: 1) Per-dispute volume growth as shops use it more, 2) Pro-tier upsells (branded letterhead, white-label landing page), 3) Direct-to-homeowner consumer SKU at $79/dispute as a separate flat-fee line, 4) Public-adjuster vertical at $349/mo when AB-75 disputes spill into actual claim litigation.

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

Three hard channels with named lists:

• Roofer Slack / Facebook groups. Roofing Insights Network, Roofing Mafia, RoofingProTalk, RCAW (Washington), several dozen state-level Facebook groups with 5K–40K members each. Drop a 90-second Loom: “AB-75 starts July 1. Here’s what a non-renewal dispute looks like for a State Farm letter we ran through this morning.” Expect 1–2% group-to-trial conversion = ~50–100 trials per group from a single post.
• Direct outreach to CA C-39 holders. CSLB license database is public. Filter to active C-39, residential, CA. Scrape ~12K records. Send personalized email + 60-second video referencing their city’s non-renewal density (use CDI public county data). 0.5% reply, 30% reply→demo, 30% demo→close = ~180 customers from one pass.
• United Policyholders + non-renewed-homeowner backchannel. UP collects homeowner complaints and runs free clinics. They’re already publishing on this exact problem. Partner: free DroneDispute access for their advocates, in exchange for a referral pipeline of homeowners who then get matched with subscribed roofers. Each homeowner clinic generates 10–30 inbound roofer requests.
• Insurance-AI-marketing PR cycle. Marketingcode and a half-dozen roofing marketing blogs are already writing “roofers who fight AI win” content with no product to recommend. Pitch one piece per month with named contractor case studies.
• Trade publications. Roofing Contractor Magazine, Roofers Coffee Shop, RoofersGuild — sponsored case-study posts at $500–$2K per placement, expected 20–40 trial signups each.

If first-100 doesn’t convert via the first two channels in 8 weeks, the idea is broken — kill it.

10. Build complexity — justification

Medium. Vision-model annotation, OCR letter intake, statute-aware letter generation, branded PDF report — all standard 2026 web stack on Anthropic/OpenAI APIs plus a doc-generation service. No custom ML, no on-device anything, no hardware. Two real risks: (a) image-rebuttal accuracy bar must be high enough that insurers don’t trivially dismiss the dispute — that’s prompt engineering and a 50-image golden set, not research; (b) statute-tracking matrix as copycat states pass laws — operational, not technical. v1 in 8–10 weeks for a builder + one part-time roofing-domain advisor.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	AB-75 explicitly creates the dispute channel; we’re building the consumer-side tool the statute presumes exists.
Ethical — no harm / dark patterns	✅	Increases consumer rights against insurer machine vision. Defensive, not offensive.
Market exists (evidence above)	✅	400K+ CA non-renewals since 2021; weekly press cycle on individual cases.
1–5 person team can build this	✅	Solo + roofing advisor + designer for v1.
Launchable with <$50K / ₹40L	✅	Vision API + standard SaaS stack + small CSLB scrape budget.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	18/20	Hair-on-fire. Homeowner has a literal deadline; roofer has a competitor closing tomorrow. Both feel pain weekly during scan-heavy quarters.
Demand evidence	15	13/15	Multiple named cases (Bennett, Arman, Van Kuren, Sveen), trade press explicitly framing this as roofer’s next lead source, statute being passed because demand exceeded political tolerance. Not yet quantified ad-spend or competitor revenue, hence not 14–15.
Build feasibility	15	12/15	Off-the-shelf vision + LLM + doc-gen, but the rebuttal accuracy bar is real and the statute matrix grows over time. 8–10 weeks for v1, not 4.
Distribution clarity	15	12/15	Three named channels, scraped CSLB list, partner with UP, trade-press hooks all align. Conversion math is “show, don’t tell” until first 50 customers prove it.
Revenue mechanics	15	12/15	$99–$599 tiers benchmark cleanly against existing roofer SaaS. ACV ~$2.8K. $1M needs 360 accounts in CA — within reach. $5M needs national rollout, which depends on copycat passage.
Time to first revenue	10	7/10	Pre-sale 5 CA shops in May/June, monetize in July when AB-75 turns on. ~8 weeks to first paid customer.
Defensibility	10	6/10	Soft moats: roofer brand lock-in, statute-tracking matrix as states multiply, accumulating dispute outcome data that improves win rate. Copyable in 6 months by a focused competitor; statute knowledge accumulates.
Total	100	80/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy (vision/LLM pipelines, doc-gen, integrations) and sales-heavy (CSLB cold outreach + trade-press cultivation + UP partnership). A solo technical founder with one part-time domain advisor (a CA C-39 holder or ex-public-adjuster) and a part-time SDR-style sales contractor can ship + sell.

Key assumptions to validate (3–5)

1. Assumption: CA roofers will pay $199–$249/mo for a tool that converts non-renewal letters into closed jobs. How to test: Pre-sell 10 CA C-39 shops in May–June via Loom + 30-min demo + $99 founding-member annual. ≥5 close = pass.
2. Assumption: AI-generated rebuttal reports successfully reverse insurer non-renewal at ≥20% rate vs. <5% baseline for free-form roofer letters. How to test: Run 25 real disputes between June–August with founding customers; track outcomes via CDI complaint data + carrier responses. ≥20% save rate = pass.
3. Assumption: AB-75 actually creates an enforced dispute channel rather than a paper-only right. How to test: Watch CDI bulletin + first 30 days of operative period; survey 20 CA homeowners who’ve received post-July non-renewals. Insurers responding to disputes = pass.
4. Assumption: Copycat laws pass in at least 2 of 6 states (MA/NY/GA/WA/IN/OK) within 12 months. How to test: Track legislative session calendars; if 0 of 6 advance past committee by Q1 2027, the geographic expansion thesis is delayed.
5. Assumption: Roofers will use a tool that lives outside their existing CRM (JobNimbus/AccuLynx/Roofr) rather than wait for those CRMs to add the feature. How to test: First-50-customer interviews — ask if they’d cancel if AccuLynx shipped equivalent. ≥70% say “use both” = pass.

Risk flags

1. Regulatory risk: AB-75 implementation could be delayed or watered down by carrier lobbying. Mitigation: track CDI rulemaking; have a non-CA Washington/Mass beachhead ready as backup.
2. Platform dependency: Insurer-controlled aerial imagery (EagleView etc.) could refuse to license historical images for dispute purposes. Mitigation: AB-75 forces the carrier to hand over the image to the homeowner, who supplies it to us — no need to license from EagleView directly.
3. Defensibility erosion: Roofer CRMs (JobNimbus, AccuLynx, Roofr) could ship equivalent features in 9–12 months. Mitigation: own the regulatory-tracking content moat, build dispute-outcome dataset early, and lock distribution via UP partnership before incumbents notice.
4. AI accuracy risk: A high-profile failed dispute where DroneDispute’s report was demonstrably wrong could damage the brand for the entire category. Mitigation: human-in-loop review for first 6 months of disputes, conservative confidence thresholds, explicit “contractor-signed” liability shift.

14. Structured verdict

Score:                  80/100
Verdict:                STRONG GO
Confidence:             Medium
Best-fit builder:       Technical solo founder + part-time CA C-39 contractor advisor + part-time SDR. Bonus: prior insurance or PropTech experience.
Time to revenue:        6–8 weeks (pre-sales May/June, monetized at AB-75 go-live July 1)
Capital to launch:      $15K–$30K (vision API credits, doc-gen tooling, CSLB list scraping, UP partnership marketing, founder cash to bridge first 90 days)
Top 3 assumptions to validate first:
  1. CA roofer willingness to pay — pre-sell 10 shops, target ≥5 closes.
  2. Dispute-success rate — 25 real disputes, target ≥20% policy-save rate.
  3. AB-75 enforcement reality — survey 20 post-July CA non-renewal homeowners.
Kill criteria:
  - Abandon if <5 of first 50 outreach roofers convert to paid by end of August 2026.
  - Abandon if AB-75 enforcement is paper-only (≤5% of disputes get carrier response in first 60 operative days).
  - Abandon if a roofer CRM (JobNimbus, AccuLynx, Roofr) ships an equivalent feature before our v1 launch.

15. Next step — 1-week validation sprint

• Day 1–2: Build a one-page landing — “Stop AB-75 non-renewals. Win the roof job that follows.” Mock a sample dispute report from one of the public Bennett-style cases. Pull the CSLB C-39 list, filter to ~500 active CA shops in Orange / LA / Ventura / San Diego counties.
• Day 3–4: Send 200 personalized cold emails to those C-39 holders citing their county’s non-renewal density. Goal: 10 demo bookings.
• Day 5: Run 5 demos. Ask each contractor: “Would you pay $199/month for this starting July 1? Would you put $99 down today as a founding member?” ≥3 founding-member checks cleared = GO. ≤1 = back to the cluster pile.

The validation is falsifiable: founders’ money on the table or it isn’t.