What changed
A named carrier (USAA) reportedly closed 51% of 2025 home-insurance claims with no payment (source article), while cheap open multimodal models (Tess-4-27B) now draft document-grounded letters at near-zero marginal cost. The pain spike and the cheap capability arrive together.
Why now
The 51% no-pay figure is fresh, named, and searchable — denial anger is acute right now and homeowners actively search 'USAA denied my claim, how to appeal' before a short appeal window closes. Search intent is the distribution.
Converging signals
FACT: a news report claims USAA closed 51% of home claims without payment in 2025 (expressnews.com). FACT: an open 27B multimodal model exists (huggingface Tess-4-27B). HYPOTHESIS: denied homeowners will pay for a personalized appeal — not established by the input, which carries an empty demand_evidence array.
Customer pain
A homeowner or small landlord holding a denial letter, angry, on a deadline, not knowing the policy language to cite or that a state Department of Insurance complaint is an option. Real and acute — but the pain is episodic, not recurring.
Who pays
The denied policyholder pays by card once. This is a one-time distressed-consumer purchase, not a subscription; there is no repeat buyer and no forced buyer.
Solved today
Free generic appeal templates (blogs, nonprofit consumer sites, state DOI sites), public adjusters (who take a % and are the real incumbent for large claims), attorneys (contingency, only worthwhile on big denials), or the homeowner writes it themselves.
Why current solutions are bad
Free templates aren't grounded in the specific denial reason or policy language and don't tell the user about DOI complaint rights; public adjusters/attorneys are overkill and unaffordable for small denials. Gap is real for mid-size claims too small for a pro but too painful to abandon.
Proposed product
Web form: paste denial reason + policy type + state + upload denial letter/damage photos → LLM drafts a citation-backed appeal letter referencing policy language and state DOI complaint rights, plus a checklist and the state commissioner complaint link. Stripe checkout, no login. Position explicitly as a document-drafting/information tool, NOT legal advice.
MVP version
Single page, one LLM call, Stripe payment link, per-state DOI complaint-link lookup table (all 50 pre-built), PDF export. Buildable in days. Add a prominent 'not legal advice / does not guarantee outcome' disclaimer and route large/complex denials to 'consider a public adjuster or attorney'.
30-day build
Ship MVP, pre-load 50-state DOI links and a few carrier-specific denial-reason patterns, seed SEO/content around named-carrier denial searches, buy a handful of high-intent search ads to test conversion on '[carrier] claim denied appeal'.
60-day build
Measure conversion and refund rate; add rush/expert-cited tier ($19+$20); add adjacent verticals (auto denial, health/EOB denial) which share the exact shape and are far larger markets; test a public-adjuster/attorney referral affiliate for claims too big for the tool.
90-day revenue plan
If conversion holds, scale content + paid search across carriers and states; revenue is volume-of-traffic-driven, not account-driven. Realistic path is a few hundred to low-thousands of one-time sales/month if SEO/ads convert — must be proven, not assumed.
Distribution path
High-intent SEO and paid search on denial keywords; the buyer is already searching in crisis. No platform gatekeeper. Weakness: every sale requires fresh acquisition — no retention, no referral loop, ad-dependent if SEO is slow.
Pricing hypothesis
$39 one-time (or $19 base + $20 rush/expert-cited). Reasonable for acute pain; low absolute price means acquisition cost must stay well under it.
Technical difficulty
Low. Trivial single-call LLM wrapper + Stripe + a static state-link table. That low bar cuts both ways — it is also the moat problem.
Legal / regulatory risk
Real gray area: drafting appeal letters and citing policy/DOI rights borders on unauthorized practice of law (UPL) if framed as legal advice. Mitigate with strict information-tool framing, no outcome guarantees, and no case-specific legal conclusions. Also handling denial letters + damage photos is sensitive PII.
Platform dependency
None on the delivery side (self-hosted + Stripe). Dependency is on search-acquisition channels and on Stripe not classifying it as high-risk.
Founder fit
Moderate, not high. This is a discretionary consumer AI-wrapper, NOT the founder's proven government-portal/forced-filer shape (no mandate, no per-filing recurring buyer, no forced-buyer class). It fits his complaint-mining and fast-prototyping strengths but not his strongest edge, and consumer distress marketing isn't his stated preference.
Breakout potential
Moderate. The letter-from-denial pattern generalizes hugely to auto, health/EOB, disability, and warranty denials — the same engine, far bigger pools. But each remains a one-time, acquisition-hungry, easily-cloned consumer product.
Final recommendation
WEAK PROCEED as a cheap, fast validation experiment only — build the MVP in days and spend a small ad budget to test whether distressed homeowners actually pay, because the core assumption (willingness to pay over free templates) is unproven. Do NOT treat it as a durable business until conversion and refund data exist. It ranks BELOW any live government-portal/forced-filer or claimable-money mandate for this founder.
Next action
Stand up the single-page MVP + Stripe + 50-state DOI link table, then run a 2-week high-intent search-ad + SEO test on named-carrier denial keywords; keep only if paid conversion clears CAC with an acceptable refund rate.