ClaimBay — warranty-claim builder for small truck fleets

1. One-liner

Turns a bay-floor photo, fault code and tech note into a filed OEM warranty claim before the window closes.

2. Trend signal — why now?

Warranty money is sitting on the floor of every small truck shop, and the industry press has finally put numbers on it. Fleets without structured tracking recover less than 30–35% of claimable repair costs, leaving 12–18% of eligible repairs unclaimed every year. A 50-vehicle fleet without systematic tracking routinely writes off $15,000–$40,000 annually — money already paid for in the truck’s purchase price and simply never collected. Even a 10-truck operation leaves ~$7,500 on the table at a 50% recovery rate.

The cause is not laziness — it’s a documentation gap at the exact wrong moment. A successful claim requires photographing the defective part, recording diagnostic fault codes, and capturing technician notes before the component is scrapped, then filing with VMRS-coded line items within a 30–60 day window. Miss any single step and the OEM auto-denies. Work orders without VMRS codes are rejected at the portal or bounce back for manual re-coding, “wasting administrator hours.”

What changed: (a) the pain is now quantified and being marketed against, and (b) vision + LLM models can now read a bay-floor photo of a cracked turbo, parse a freehand tech note, and map both to the right VMRS system/assembly/component codes — the precise manual step that gets skipped. That capability was research-grade two years ago; it’s a cheap API call today.

Provenance:

• Signal 1 (Demand): Small fleets recover <35% of claimable warranty; 50-truck fleet writes off $15K–$40K/yr from missing docs and missed windows — https://fleetrabbit.com/industry/transportation-and-logistics/truck-fleet-warranty-recovery-and-oem-claim-tracking-2026 — 2026-07-11
• Signal 2 (Feasibility): Claims require part photo + fault codes + tech notes captured before scrap, VMRS-coded, filed in 30–60 days; VMRS coding is mandatory on OEM portals and rejects on submission if missing — https://www.fleetmaintenance.com/shop-operations/article/55378267/how-digital-warranty-tracking-gets-fleets-their-money — 2026-07-11
• Signal 3 (Economic): Warranty-tracking market monetizing — Decisiv runs 7M assets for 74K owners; mid fleet spends $1.5K–2.5K/yr on tracking and recovers $40K–150K — https://www.decisiv.com/ — 2026-07-11 Category: Workflow automation

3. The opportunity

The incumbents — Decisiv, Dossier/AMCS, FleetRabbit, Oxmaint, Squarerigger — are all CMMS-first. Warranty recovery is a bundled feature you only get after you migrate your entire maintenance operation onto their platform: work orders, PM scheduling, parts inventory, the works. That’s a fine sell to a 100-truck fleet with a maintenance director. It’s a non-starter for the shop running on paper, QuickBooks, and a whiteboard — which is exactly the operation leaving the highest percentage unclaimed.

So there’s a wedge the CMMS players structurally can’t serve: a thin, capture-first tool that a tech opens on a phone at the bay, snaps three photos, dictates a sentence, and walks away — and it produces a portal-ready, VMRS-coded claim packet plus a countdown to the filing deadline. No platform migration. No ripping out how the shop already works. It sits beside the existing workflow, not on top of it.

The 10× isn’t a better dashboard. It’s collapsing a 30-minute documentation-and-coding chore that techs skip under pressure into a 90-second bay-floor capture — so the claim actually gets filed at all. A claim filed at 90% quality beats the perfect claim that never got written because the tech was already on the next truck.

4. Target market

• Primary customer: Owner-operators and small fleet owners (1–25 trucks) in the US, and independent heavy-truck repair shops (ASE T1–T8) that service them. The buyer is the owner or the shop foreman — someone who signs the checks and feels the write-off personally.
• Why they buy: “I know I’m leaving money on the table but I don’t have a person to chase warranty paperwork, and by the time I remember, the 60 days are gone.” The pain is concrete, dollar-denominated, and recurring on every major component failure (turbo, injector, aftertreatment, transmission).
• Rough TAM reasoning: ~2M+ trucks operate in fleets of 1–20 in the US; the vast majority run no CMMS. Even 30,000 small fleets/shops at $150–400/mo is a $50M–140M ARR ceiling — comfortably a sub-$5M target for a bootstrapper capturing a sliver.
• Why now for them: Amazon-style margin squeeze on freight rates in 2025–26 means every recoverable dollar matters more; and OEMs have grown stricter about documentation and maintenance-record proof, so the manual approach fails more often than it used to.

5. Product sketch (MVP)

• Bay capture (mobile): tech snaps the defective part, the odometer, and the diagnostic scan-tool screen; dictates a one-line note. Done in under 2 minutes, at the truck.
• AI claim assembly: vision + LLM reads the photos and note, extracts fault codes and failure description, and maps them to the correct VMRS system/assembly/component codes.
• Coverage check: given VIN + in-service date + mileage, flags whether the repair is likely in-warranty before the shop eats the cost — the “check coverage before you approve” step that manual tracking fails.
• Deadline countdown: every captured event gets a filing-window timer (30/60/90 days by OEM) with escalating reminders; nothing silently expires.
• Portal-ready packet: generates the VMRS-coded claim narrative, part numbers, tech ID, and evidence pack formatted for the target OEM’s portal (Eaton, PACCAR, Cummins, Detroit, etc.) — copy-paste or PDF export for v1.
• Recovery ledger: running dollar total of claims filed, approved, pending, and expired — the number the owner checks to feel the ROI.
• Late-claim mode: for events past the window, assembles the “complete explanation + full documentation” package OEMs will sometimes still accept.

6. AI angle — what’s load-bearing

Remove the AI and this is a form nobody fills out — which is precisely today’s failure state. The load-bearing work is the VMRS coding and claim-narrative generation from unstructured bay-floor inputs: a photo of a scorched injector, a scan-tool screen with fault codes, and a tech saying “number 4 was hammering, metal in the oil.” Mapping that to 013-002-014-style VMRS codes plus a defensible failure narrative is the exact skilled step that gets skipped and causes rejections. Vision extraction (read the fault-code screen and the part) + LLM reasoning (map to codes, draft the narrative) is what turns a 30-minute expert chore into a 90-second capture. Without it you’re just another tracking spreadsheet.

7. Localization angle (if any)

N/A — this is a US-first play. The moat is US OEM warranty-portal formats, VMRS (a North American standard maintained by ATA), and US fleet economics. A future EU/India adaptation would need entirely different OEM programs and standards; not a v1 wedge. Deliberately US to break the recent portfolio tilt toward compliance-defense plays and add an under-served TradeTech/automotive vertical.

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

• Pricing: $149/mo for owner-operators/tiny fleets (up to 5 trucks), $349/mo for 6–25 trucks and independent shops. Optional success fee: 8–10% of recovered dollars for shops that want zero subscription risk — the ROI is self-funding since a single recovered turbo claim ($2,000–8,000) covers a year of subscription.
• ACV: ~$3,000 blended.
• Rough math to $1M ARR: ~330 customers × $3,000 = $1M. Given a 50-truck fleet recovers $40K+, a $349/mo tool is a rounding error — the sell is easy once they see one claim land.
• Rough math to $5M ARR: ~1,650 customers, or a mix of subscription + success fees on a few large-volume shops. Requires OEM-portal coverage across the big five truck makers and word-of-mouth in the independent-shop community.
• Expansion path: per-truck pricing as fleets grow; add recall management, extended-warranty tracking, and a “warranty audit” one-time service that back-files the last 60 days of missed claims (instant cash for the customer, instant proof for you).

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

• Free warranty audit as the hook: offer any small fleet a free review — “send me your last 90 days of major repair orders, I’ll tell you exactly how much you left on the table.” The number is the pitch. Convert the ones staring at a $12,000 write-off.
• TruckersReport + heavy-duty repair forums: owner-operators actively discuss warranty confusion there (e.g., “Dumb question about truck warranty” threads). Show up with the audit offer and specific recovery numbers, not a landing page.
• Independent shop channel: cold-visit and call the ASE T1–T8 shops in three trucking corridors (I-40, I-80, I-10 metros). Shops that service small fleets are a force multiplier — one shop refers many owner-operators, and the shop itself files claims on its own parts.
• Parts-distributor co-sell: heavy-truck parts distributors (who eat some warranty friction) are a natural referral partner — they want the OEM to reimburse instead of the customer disputing.
• Recovery-share pilots: for the first 20 shops, run pure success-fee — zero risk to them, and the filed-claim dollar totals become the case studies that close the next 80 on subscription.

10. Build complexity — justification

Medium. Off-the-shelf: mobile capture, vision + LLM APIs, VMRS code reference data (licensable from ATA), standard web stack. Custom work: reliable VMRS mapping (needs a domain advisor and a validation loop so codes aren’t hallucinated), and per-OEM portal packet formatting — each of the big five (PACCAR, Daimler/Detroit, Cummins, Eaton, Volvo/Mack) has its own claim format, so coverage is incremental integration work. A pair could ship a credible v1 covering two OEMs in ~10–14 weeks; full five-OEM coverage is a 4–5 month arc.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Assembling and filing legitimate warranty claims on behalf of the owner; no regulatory approval needed.
Ethical — no harm / dark patterns	✅	Recovers money customers are genuinely owed; success-fee model aligns incentives.
Market exists (evidence above)	✅	Quantified write-offs, active incumbents, VMRS-mandated portals.
1–5 person team can build this	✅	Pair + domain advisor; incremental OEM coverage.
Launchable with <$50K / ₹40L	✅	API + VMRS licensing + one advisor. No hardware, no inventory.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	16/20	Dollar-denominated, recurring, felt on every major failure — but it’s a “known leak” not a “hair on fire, act today” pain, so it’s a 16 not a 19.
Demand evidence	15	12/15	Hard quantified write-offs and multiple monetizing incumbents; docked because most evidence is vendor-published, thin on raw verbatim small-fleet complaints.
Build feasibility	15	11/15	Off-the-shelf capture + AI, but VMRS-mapping accuracy and per-OEM portal formats are real integration work.
Distribution clarity	15	12/15	The free-audit hook and shop channel are concrete; owner-operators are dispersed and hard to reach at scale.
Revenue mechanics	15	12/15	ROI is self-evident (one claim pays a year); success-fee de-risks the sell. Churn risk if claims are infrequent for a given customer.
Time to first revenue	10	7/10	Free audit → paid conversion can happen in weeks, but the two-OEM MVP gates the first real filings.
Defensibility	10	4/10	Execution + accumulating VMRS-mapping and portal know-how; a CMMS incumbent could ship a lite tier. Head-start moat, not a structural one.
Total	100	74/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · domain-expertise-required — needs solid vision/LLM engineering plus a warranty/fleet-maintenance advisor who knows VMRS and how each OEM portal actually behaves.

Key assumptions to validate (3–5)

1. Assumption: AI can map bay-floor photos + freehand notes to correct VMRS codes at a quality OEMs accept. How to test: run 50 historical real repair orders through the pipeline and have a warranty-recovery expert grade the code + narrative accuracy; target ≥85% acceptance-grade.
2. Assumption: small fleets/shops will pay $149–349/mo (or a success fee) once they see a recovered dollar. How to test: run free audits for 20 shops, count how many convert to paid after the first filed claim lands.
3. Assumption: filing via copy-paste/PDF (not deep portal API integration) is good enough for v1. How to test: file 10 real claims manually through the packet output; measure approval rate vs. a CMMS-native submission.
4. Assumption: the free-audit hook produces a usable pipeline. How to test: offer it in two forums + 30 shop calls; measure audit requests and average discovered write-off per audit.

Risk flags

1. Platform dependency: OEM portals can change formats or restrict third-party submissions; mitigate by keeping the human copy-paste path and not over-investing in fragile deep integrations early.
2. Defensibility: a CMMS incumbent (FleetRabbit, Oxmaint) can bolt on a capture-first lite tier; the defense is speed, independent-shop trust, and VMRS-mapping accuracy that compounds with data.
3. Frequency risk: a tiny owner-operator has few major failures per year, so single-truck ARR can churn between claims — the shop and multi-truck segments carry the model.

14. Structured verdict

Score:                  74/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Technical founder + fleet-maintenance/warranty domain advisor
Time to revenue:        8–14 weeks (free audit → paid after first two-OEM MVP filings)
Capital to launch:      $15–30K (API + VMRS licensing + one advisor)
Top 3 assumptions to validate first:
  1. VMRS-mapping accuracy on 50 real ROs graded ≥85% by an expert
  2. Free-audit → paid conversion after first filed claim (20 shops)
  3. Copy-paste/PDF packet approval rate vs CMMS-native (10 real filings)
Kill criteria:
  - Abandon if AI VMRS-mapping accuracy stays <80% acceptance-grade after tuning on 50 real ROs
  - Abandon if <15% of audited shops convert to paid after seeing a real recovered claim
  - Abandon if a major OEM blocks non-CMMS third-party claim assembly before v1 ships

15. Next step — 1-week validation sprint

• Day 1–2: collect 30–50 real historical heavy-truck repair orders (from an advisor shop or forum volunteers) that were never filed as warranty claims. This is the ground-truth set.
• Day 3–4: run them through a prototype pipeline (photo/scan/note → VMRS code + claim narrative) and have a warranty-recovery expert grade each for acceptance quality and total recoverable dollars discovered.
• Day 5: decide go/no-go on a falsifiable threshold: ≥85% of the 50 ROs graded acceptance-quality and an average discovered write-off ≥$1,500 per shop. Below either bar, the AI isn’t load-bearing enough or the money isn’t big enough — stop.