ClawClaim — Pet-insurance claim cockpit for independent US vet clinics

1. One-liner

AI cockpit that turns any vet visit into a claim-ready packet for 12 US pet insurers in under 60 seconds.

2. Trend signal — why now?

Pet insurance went from a curiosity to a clinic operational problem in 36 months. NAPHIA’s 2025 State of the Industry report: 7.03M insured pets in North America (91% US), $5.2B written premium, 20.8% premium growth YoY, dog penetration 5.46%, four straight years of >20% growth. That means an independent 3-DVM clinic seeing 6,000 visits/year is now processing 5–15 insurance claims a week instead of 1 a month. The front desk is the bottleneck.

The supply side is screaming about it. Lemonade publishes “A Vet’s Guide to Pet-Insurance Optimized Medical Records” literally begging vets to write claim-ready SOAP notes — quote: “the fifteen minutes spent creating comprehensive SOAP notes can save clients weeks of frustration and thousands of dollars in delayed reimbursements.” Insurer engineering blogs (Insurnest 2026) call out manual record review as the #1 margin sink for MGAs. Trupanion + ezyVet shipped VetDirect Pay integration in 2022 and it became Trupanion’s lead acquisition lever. That precedent — clinic-side automation as a wedge — is now public and proven.

The catch: Trupanion-ezyVet only covers one insurer and two PIMS (ezyVet, IDEXX). MetLife, Pets Best, ASPCA, Embrace, Lemonade, Healthy Paws, Spot, Fetch, Figo, AKC, Nationwide, Pumpkin — all 11 of them — still make the clinic assemble the claim by hand, on whichever PIMS they happen to run (AVImark, Cornerstone, Shepherd, Vetspire, Digitail, NaVetor, DaySmart, Provet…). That fragmentation is the wedge. Until LLMs got good at unstructured SOAP extraction + form mapping (cheap as of 2025), there was no economic way to build the cross-insurer cross-PIMS layer.

Provenance:

• Signal 1 (Demand — insurer-side): Lemonade and Bankrate publishing 2025/2026 “vet’s guide” articles instructing clinics how to write claim-ready notes — Lemonade vet guide — May 2026
• Signal 2 (Feasibility — tech unlock): Insurnest 2026 publishes claims-automation playbook noting AI invoice + SOAP extraction now reliable enough for insurance adjudication — Insurnest 2026 — Apr 2026
• Signal 3 (Economic — market growth): NAPHIA 2025 SOI — $5.2B premium, 20.8% YoY growth, 12.7% growth in insured US pets — NAPHIA SOI 2025 — Feb 2025 Category: Workflow automation

3. The opportunity

The pet insurance ecosystem has a clinic-side workflow gap that nobody owns. Three player types touch it badly:

• PIMS vendors (AVImark/Henry Schein, Cornerstone/IDEXX, Shepherd, Vetspire, Digitail) won’t build per-insurer claim packets because they want to stay neutral and because each insurer’s form is a special snowflake. They expose “export invoice as PDF” and stop there.
• Insurer-side tools (Trupanion VetDirect, Pets Best Direct Pay) only work for that one insurer’s policies — useless when the next client walks in with MetLife.
• Comparison/advisor tools (Pawlicy Advisor) play upstream at the buy/recommend stage. Once a policy is sold, they’re done.

Result: the clinic becomes the unpaid claims-processing arm of 12 different insurers. A practice manager I’d interview will describe it identically across clinics: “Carla spends Tuesdays printing invoices, finding the right SOAP page, faxing or uploading, and answering ‘where’s my reimbursement’ calls for the next two weeks.”

The 10× move: a single cockpit that ingests the PIMS visit record (PDF export or PIMS API where available), uses LLM extraction to pull diagnosis codes, line-item charges, signs/symptoms, and procedure dates, then auto-fills the correct insurer’s PDF form (or hits their portal API/email intake) — all in <60 seconds. Includes a status board for outstanding claims, a “needs more docs” pre-flight check, and a client-facing reimbursement tracker. Vs. status quo (8–20 min/claim by Carla, plus 5–10% kickback rate for missing docs), this is genuinely 10× on labor and dollars left on the table.

4. Target market

• Primary customer: Practice manager or hospital administrator at independent companion-animal general practice in US — 1–5 DVMs, 4–15 total staff, $1.5M–$8M annual revenue. They’re the ones who feel the pain because they pay the labor and field the client phone calls.
• Why they buy: Front-desk overtime is up, insured-client share is climbing every quarter, claim kickbacks generate angry calls. They already pay $300–$600/mo for PIMS and $200/mo for Weave-style comms — adding $149–$249/mo for a tool that gets back ~6 staff hours a week is an obvious yes when reframed as “$25/hour CSR labor saved.”
• Rough TAM reasoning: ~32K US veterinary practices total. Strip out specialty/ER (75% corporate) and corporate-owned GPs (~30% of GPs) → roughly 18K–22K independent companion-animal GPs. Even at 5% paid penetration with $199/mo ACV → $2.1M ARR. 15% penetration at $249/mo → $9.9M ARR. The ceiling exists but the bootstrap zone is wide open.
• Why now for them: Pet insurance hit the “5+ claims/week” threshold where it’s annoying enough to staff for. Trupanion-ezyVet integration showed clinics what’s possible — now non-ezyVet clinics with non-Trupanion clients are asking why they can’t have parity.

5. Product sketch (MVP)

• Universal intake — Drag-drop a visit PDF from any PIMS, paste an invoice export, or connect via PIMS API (start with Shepherd, Vetspire, Digitail — newer cloud PIMS with open APIs)
• Insurer-aware packet builder — Select the insurer; system auto-fills that insurer’s claim PDF with itemized charges, diagnosis, dates, signs/symptoms, and DVM signature placeholder. Day-one support for the top 6 insurers (Nationwide, MetLife, Trupanion, Pets Best, Embrace, ASPCA) covering ~85% of US policies
• Pre-flight validator — Flags missing items before submission: “ASPCA needs prior medical history for first claims,” “MetLife wants SOAP for any GI claim >$500,” “Embrace needs pre-cert for this surgery code”
• One-click submit — Where insurer has portal API or email intake, submit directly; otherwise generate a client-shareable upload link with the packet pre-bundled
• Status board — Tracks every open claim, days outstanding, predicted reimbursement timeline; auto-pings client when reimbursement hits their account
• Client-facing reimbursement tracker — Branded mini-page (no login) where pet owner can see “Submitted → Under review → Approved $X → Paid”
• Monthly insurance report — Auto-generated practice report: claims submitted, kickback rate, avg reimbursement time per insurer (clinic uses this internally and to advise clients on which insurers to recommend)

6. AI angle — what’s load-bearing

AI is the spine. Three load-bearing places:

1. SOAP/invoice extraction. Vet notes are wildly heterogeneous — handwritten-then-OCRed, free-text in PIMS templates, line items with non-standard abbreviations (“DUDE” = “drink/urinate/defecate/eat”). Regex won’t cut it. Vision + LLM extraction with a vet-vocabulary dictionary turns this into structured {date, problem_list, procedures, charges, notes}. Without AI, this is a $5M-of-engineering rules engine that never catches up.
2. Insurer form mapping. Each insurer’s form has different field names for the same concept (“Diagnosis” vs “Presenting complaint” vs “Reason for visit”). LLM-as-mapper between extracted record schema and insurer form is the only way to maintain 12 insurers × form updates without a team of analysts.
3. Pre-flight validator. “Will this claim get kicked back?” requires reasoning over insurer policy rules + extracted record + claim history. Classifier + LLM check against per-insurer denial patterns.

Strip the AI: you have a PDF generator nobody buys.

7. Localization angle (if any)

US-first, intentionally. This is a US-shaped market problem — 12 specific insurers, English-language SOAP notes, US-specific HIPAA-doesn’t-apply-but-data-privacy-norms-do environment. Canada (~9% of NA pet insurance market, much more concentrated insurer landscape — Trupanion, Petsecure, Pets Plus Us) is a natural v2 with a 90-day re-skin. UK is a different beast (separate insurer set, different claim norms, much higher penetration ~25%) and would deserve its own dedicated build, not a localization.

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

• Pricing: Three tiers — Solo $149/mo (1 DVM, up to 30 claims/mo), Practice $249/mo (up to 5 DVMs, unlimited claims), Multi-site $499/mo (up to 5 locations, shared status board). White-glove onboarding +$499 one-time.
• ACV: Blended ~$2,400/year. Practice tier is the meat.
• Rough math to $1M ARR: 350 practice-tier clinics @ $249/mo × 12 = $1.05M. ~1.7% of the 22K-clinic TAM.
• Rough math to $5M ARR: 1,650 practice-tier clinics → 7.5% TAM penetration. Realistic in 24 months if first 200 are happy.
• Expansion path: (a) Per-DVM seat creep as practices grow, (b) Insurer-funded distribution kickbacks — multiple insurers will pay $5–$15 per submitted claim because it dramatically lowers their adjudication cost (Trupanion already proves this with VetDirect — they fund the integration), (c) Client-facing premium tier for pet owners ($5/mo to track all their claims across insurers).

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

Concrete, in this order:

• The directory blitz. Buy/scrape the AAHA member list (~3,500 accredited companion-animal hospitals — already heavily independent, already quality-conscious). For each, ship a personalized 90-second Loom: “I pulled an estimated claim volume for [Clinic Name] based on Yelp review mentions of insurance + Google reviews mentioning reimbursement. You’re probably losing $X/month in CSR time. Here’s what we’d save you.” Expect 8–12% reply, 3% close → 100+ trials.
• Practice manager forums. VHMA (Veterinary Hospital Managers Association) has ~3,500 members, an active Facebook group, and a paid conference. Sponsor one VHMA conference ($8K), run a “claim packet showdown” booth where you race a practice manager’s manual process vs ClawClaim. Expect 40 leads, 15 close.
• Insurer-side referral piggyback. Pawlicy Advisor sits inside hundreds of clinics already and is friendly to vendors that improve the post-sale experience. Partner pitch: “We close the loop on what Pawlicy started — they sell the policy, we file the claims.” Co-marketing to Pawlicy’s clinic list gets you another ~50 warm intros.
• PIMS marketplace plays for the cloud PIMS. Shepherd, Vetspire, Digitail, Provet all have integrations pages and partner programs. Ship integrations for 2 of these in month 4 — they actively co-promote because it makes their PIMS more attractive vs. AVImark.
• dvm360 / VetGirl / VIN content distribution. Long-form “the real cost of pet insurance claims at your clinic” with the math worked out — vet practice managers actually read this stuff. One viral piece + insurer comparison chart = inbound funnel.

If after 100 cold Looms and a VHMA booth you don’t have 25 paying clinics, the pricing or the demo experience is wrong — not the market.

10. Build complexity — justification

Medium. v1 is: PDF upload + cloud LLM extraction (Claude/GPT class) + a per-insurer form schema (~6 forms manually mapped) + PDF generator + a status board. No on-prem PIMS integration required for v1 — clinics export visit summaries as PDF. Realistic for a 2-person team in 12–16 weeks. The non-trivial parts: (1) building and maintaining the per-insurer form schema as insurers update forms (manageable with a quarterly review cadence), (2) the eval harness to keep extraction accuracy >97% on real-world messy SOAP notes (this is the hard engineering), (3) PIMS API integrations for v1.1 — but those are nice-to-have, not blocking. Custom models or novel research not required.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	No HIPAA for animals; standard SaaS data-handling. State-level vet practice acts don’t restrict admin software.
Ethical — no harm / dark patterns	✅	Improves reimbursement for both clinic and pet owner. No conflicts; insurer-funded distribution is opt-in transparent.
Market exists (evidence above)	✅	$5.2B premium, 7M insured pets, 20% CAGR, existing 1-insurer integration commercial success.
1–5 person team can build this	✅	2-person team ships v1 in 12–16 weeks.
Launchable with <$50K / ₹40L	✅	$25–35K covers infra, 6 insurer form maps, AAHA list, VHMA booth, founder time.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	18/20	Felt weekly, growing weekly, costs real money in CSR labor + delayed reimbursements + client churn from frustration. Not hair-on-fire (clinic doesn’t shut down) but firmly “if you do this I’ll buy it tomorrow.”
Demand evidence	15	14/15	NAPHIA hard numbers, Trupanion-ezyVet commercial validation, insurer-side public begging for better records, practice manager forum volume. Multiple independent signals.
Build feasibility	15	12/15	LLM extraction is the right call, but eval rigor + per-insurer form maintenance is real ongoing work. Not a weekend project.
Distribution clarity	15	13/15	AAHA + VHMA + PIMS marketplaces + Pawlicy partner play is concrete and named. Conversion math holds up.
Revenue mechanics	15	13/15	$249 ACV proven by Weave, VetBilling, etc. for vet tools. Insurer kickback as expansion is a clear bonus path.
Time to first revenue	10	8/10	6–8 weeks after v1 ships, pilots → paid via founder-led sales. Not pre-sold, but short funnel.
Defensibility	10	7/10	Per-insurer form library + extraction eval set + denial-pattern data accumulate into a real moat by month 18. Plus PIMS integration partnerships are sticky. Not patent-level, but copying it in month 3 means redoing the form library from scratch.
Total	100	82/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy (LLM eval pipelines, document extraction, PDF rendering) · domain-expertise-required (a vet-side or insurer-side advisor on the cap table makes this dramatically faster; cold without one, you’ll burn 2 months learning what “presenting complaint” means).

Key assumptions to validate (3–5)

1. Assumption: Independent clinics will pay $249/mo for this. How to test: 20 cold interviews with practice managers — show a working demo of the Claude-extraction → packet flow on their actual exported PDF (ask them to bring one). Ask “would you pay $249/mo today?” Need 12+ yes.
2. Assumption: LLM extraction can hit >97% field accuracy on real-world messy SOAP notes across 5 different PIMS. How to test: Build extraction eval harness with 200 hand-labeled visit PDFs from 5 PIMS, measure precision/recall per field. Need >95% on dollars-and-dates fields, >90% on diagnosis/symptoms.
3. Assumption: Insurers won’t actively block clinic-side automation (e.g., terms-of-service bans). How to test: Read the top 6 insurers’ clinic ToS + portal ToS, ring a friendly attorney for a 1-page memo. Need: no explicit prohibition on submitting on behalf of policyholder.
4. Assumption: AAHA member list cold outreach hits 8%+ Loom reply rate. How to test: Send 100 Looms in week 1 of validation sprint. Need 8+ replies, 2+ booked demos.
5. Assumption: Insurer-funded distribution (kickback per claim) is achievable within 12 months. How to test: Once at 50 paying clinics, take meetings with 3 insurers’ BD/innovation teams using the volume data as leverage. Need 1 of 3 to seriously engage.

Risk flags

1. PIMS vendor competitive response: Cornerstone (IDEXX) or AVImark (Henry Schein) decides to build this in-house and gives it away to lock in their PIMS base. Mitigation: speed, cross-PIMS positioning, build relationships with the cloud PIMS that don’t have native claim modules.
2. Insurer direct-pay sprawl: If MetLife/Pets Best/ASPCA all build Trupanion-style direct-pay integrations, the urgency drops. Mitigation: most insurers won’t — direct-pay requires actuarial restructuring. Plus the cross-insurer cockpit is still useful even if 3/12 insurers have direct-pay.
3. PIMS API access: Older PIMS (AVImark, Cornerstone) have closed-ish APIs, so cloud-only PDF intake works for v1 but limits depth. Long-term solve is integration partnerships with the cloud PIMS where the market is migrating anyway.

14. Structured verdict

Score:                  82/100
Verdict:                STRONG GO
Confidence:             High
Best-fit builder:       Technical co-founder + a part-time vet practice manager advisor (cap-table equity, not employee)
Time to revenue:        10–14 weeks from build start
Capital to launch:      $25–35K ($30–42L)
Top 3 assumptions to validate first:
  1. 12+ of 20 cold-interviewed practice managers say they'd pay $249/mo after seeing a working demo on their own PDF
  2. LLM extraction hits >95% precision on dollars/dates fields, >90% on diagnosis fields across 200 PIMS-exported visit PDFs
  3. Top 6 insurers' ToS don't prohibit clinic-side claim submission on behalf of policyholder
Kill criteria:
  - Abandon if <6 of 20 practice manager interviews convert to "I'd pay today"
  - Abandon if LLM extraction can't sustain >90% accuracy on diagnosis/symptoms fields after 2 iteration cycles
  - Abandon if a PIMS vendor or major insurer announces a free cross-insurer clinic-side claim builder before your v1

15. Next step — 1-week validation sprint

• Day 1–2: Pull AAHA member list. Build a 60-second Loom on a fake-but-realistic clinic profile. Identify and reach out to 30 practice managers (LinkedIn, VHMA Facebook group, dvm360 commenters). Ask for 20-min call + bring one anonymized visit PDF.
• Day 3–4: Hold 12–15 interviews. Live-demo a Claude-powered extraction → MetLife/Embrace/Nationwide claim-form fill on their actual PDF. End each call with: “If this product existed today, would you pay $249/mo?”
• Day 5: Decide. Go criterion: 12+ unambiguous “yes I’d pay” answers AND at least 5 verbal pilot commitments. Anything less is a VALIDATE (run a 6-week pilot with 5 clinics for free, see if they convert to paid). Sub-6 yeses → PASS, write up findings, look at vet-side adjacent ideas (estimate copilot, client-comms automation).

The 1-week sprint produces a falsifiable result: a count of paid commitments and a measured extraction accuracy from real-world PDFs. Either the market is there or it isn’t.