The money is in the teardown.

Supplement recovery for collision repair

Shops leave thousands per repair uncollected — not because the damage isn't real, but because writing and defending a supplement takes an hour nobody in the shop has. Teardown documents the work you actually performed, prices it, and writes the argument that gets it approved.

Pre-pilot · Bay, Desk and Ledger · $500–1,000/mo per shop

$150–300 recovered per repair
estimate
Findings
RO 4471
Identified this teardown
$1,840
7 items · 2019 Honda CR-V · worded for this carrier, every line cited
Missing from the estimate
Blend adjacent panel — LT front door
Photo 4 · OEM §B-14
$214
Corrosion protection, inner rail
Photo 7 · OEM §C-02
$180
Pre- and post-repair scan
ADAS · position statement
$265
R&I front bumper cover
Photo 2 · 0.8 hrs
$96
Feather, prime and block
Photo 4 · 2.4 hrs
$488
+ 2 more · $597
Capture
Findings
Package

The problem

The damage is found.
The money isn't.

Nothing here is exotic. Every shop owner recognises all four of these, and most have accepted them as the cost of doing business.

The tech finds it and moves on

Hidden damage shows up the second the panel comes off. The technician is paid to fix cars, not to photograph and annotate them, so the find stays in his head until it is out of his head. By the time the estimator asks, the car is reassembled.

The write-up is rushed, so it gets denied

A supplement typed between two phone calls carries no labelled photo, no OEM procedure citation and no wording that carrier is known to accept. The adjuster denies it in thirty seconds — not because it's wrong, but because there is nothing in front of them to approve.

The magic words live in one person's head

Every shop has one estimator who knows which carrier pays for feather, prime and block if you call it the right thing and attach the right document. That knowledge is unwritten, unbacked up, and walks out with him.

Legitimate denials get eaten

Contesting one denied line takes an hour of phone calls and paperwork. On a $180 item most shops write it off. The carrier learns, correctly, that denial is close to free.

Undocumented is unpaid. None of this is about billing for work that wasn't done. It is about getting paid for work that already was — with the photo, the procedure and the line item to prove it.

How it works

Four steps, and only one of them
happens in the bay.

The technician's part takes under thirty seconds. Everything after that is the estimator reviewing work that has already been drafted for them.

01
Photograph the teardown As panels come off, the technician walks the car on their phone — photos and a few spoken words per finding. No forms, no typing, no trip to the office. That is the entire ask of the person holding the tools.
02
Drop in the estimate export Export the estimate from CCC, Mitchell or Audatex and drop the file in. Teardown reads the photos against what the insurer has already agreed to pay for, and against the OEM procedures for that vehicle.
03
Review the drafted findings Back comes the list of what is missing — undocumented damage the estimate never saw, plus OEM-required operations it omitted. Each line is pre-worded the way that specific carrier has historically approved it, with the procedure cited and the photo attached. Nothing is drafted that a photo or an estimate line doesn't support. The estimator approves, edits or kills each one, then keys the accepted lines into their estimating platform.
04
Send the package, answer the denials Labelled photos, justifications and citations go to the adjuster as one document instead of a forwarded email chain. When a line comes back denied, Teardown drafts the rebuttal with the evidence already attached. Approved, denied or settled, every outcome is recorded — so the next submission to that carrier is sharper than the last.

What it is

Three surfaces, one for each person who touches the job.

Nobody is asked to change how they work. The technician keeps fixing cars, the estimator keeps estimating, the owner finally gets a number.

Bay · phone · technician

Photograph and dictate the teardown

Under thirty seconds per car. Point, shoot, say what you found. It works with gloves on and no signal in the back of the shop.

Desk · browser · estimator

Review, approve, send, rebut

Findings with citations attached, a documentation package the adjuster can act on, and a drafted rebuttal waiting the moment something is denied.

Ledger · browser · owner

Dollars recovered, by carrier

What came back approved, what was denied and by whom, what is still outstanding. The number the shop can check against its own bank account.

What it knows about carriers

Which operations each insurer approves, what evidence they require, what wording works — held as structured rows, not vague similarity. Seeded from the public Who Pays for What? surveys, the Database Enhancement Gateway, and paid interviews with veteran estimators, then grown by outcomes. Retrieval is an exact filter and every answer is auditable, because money is involved.

What it knows about the vehicle

OEM repair procedures are copyrighted, so Teardown cites and links into the shop's existing ALLDATA or RepairLogic subscription rather than republishing them. OEM position statements are published freely by the manufacturers for exactly these disputes, and those go into the package in full.

The screens

The bay, the desk,
and the page the adjuster opens.

Design mockups, not a shipped product — but every figure, carrier and citation on them follows the rules the real thing will run on: fictional carriers, and no line without a photo or a procedure behind it.

Teardown's Draft screen: seven drafted supplement lines with hours, amounts and the carrier's historical approval rate, six of them accepted, beside the selected line's justification, photo evidence and cited OEM procedure.
Desk · draft supplement Each drafted line carries its hours, its amount and how often this carrier has approved that operation before. The estimator accepts, edits or kills each one — the total sent is the total accepted, never the total identified.
Teardown's Denials screen: the adjuster's stated reason for the denial on the left with what has historically worked with that carrier, and a drafted rebuttal letter with four attachments on the right.
Desk · denial rebuttal The denial arrives with the letter already written — the adjuster's stated reason, the procedure that answers it, and the photographs attached.
Teardown's Ledger screen: dollars recovered this month and year to date, recovery by month, approval rate and reply time by carrier, and the open supplements still awaiting a response.
Ledger · the owner's view Recovered, approved, still outstanding — and which carriers say yes, how often, and how long they take to answer.
Teardown's Bay screen on a phone: a large capture button, the running photo count for the repair order, and the last findings dictated at the car.
Bay · the technician's phone Point, shoot, say what you found. Under thirty seconds, one-handed, with gloves on.
A page of the documentation package: shop letterhead, claim and vehicle details, then each operation with its justification, a labelled teardown photograph and the cited manufacturer procedure.
The documentation package The actual deliverable — what the adjuster opens. One operation, one photograph, one procedure citation, laid out the same way every time. Procedures are cited and linked, never reproduced.

The market

Small enough to be ignored.
Big enough to be a business.

Roughly thirty thousand collision facilities operate in the US. We are not chasing all of them — the large consolidators build tooling in-house, so the target is the independents and small chains who have nobody to build it for them.

~30,000 US collision repair facilities Industry standard figure
10–15k Shops we can actually serve Estimate — excludes large MSOs
$2.88B Collision estimating software, 2026 $2.61B in 2025, 10.4% CAGR
The Business Research Company
$9–18k Found money per shop, per month Estimate — 40–80 insurance repairs at $150–300 recovered
$6–12k Revenue per shop, per year $500–1,000/mo flat SaaS
$100–200M Addressable market, per year Derived — 13,000 shops at $8k average
The recovery figure is a judgement, not a measurement. $150–300 per repair is a deliberately conservative estimate drawn from what shops describe leaving on the table. It is the first number the pilot replaces with a real one — and if the real one is materially lower, the pricing changes with it.

Why the unit economics hold

Drafting a supplement costs roughly $0.50–1.50 in model calls — long inputs, mostly cached — against $500–1,000 a month in subscription. The cost that matters is not compute, it is the sales call.

Year one, stated plainly

15–40 shops and $100–300K ARR. That is a judgement based on a solo founder selling direct, one shop at a time, and it assumes no channel deal appears to accelerate it.

Competition

The field is crowded. Nobody has won it.

Anyone who knows this industry can name five companies in the first minute, so there is no point pretending the space is empty. The opening is not absence. It is a conflict of interest that the largest players cannot resolve, and a set of shop-aligned vendors who are not built to exploit it.

WhoWhat they areWhy we win
CCCNASDAQ: CCCS The dominant estimating platform. Now markets AI-driven blueprinting. Counter-positioned. CCC serves shops and insurers, and the insurers hold the leverage. A tool that helps shops extract more from adjusters attacks its own larger revenue line.
Mitchell, AudatexEnlyte · Solera The other two estimating platforms. Qapter does AI photo estimating. Same conflict, same constraint. Neither can be seen coaching shops to argue with the carriers who pay them.
OEC RepairLogic, ALLDATAMost dangerous OEM procedure data, shop-aligned, already flag required operations. No structural conflict — this is the real threat. But they are data-subscription businesses rather than AI-native ones, and they own no insurer-outcome loop.
Small AI entrantsEstimate-analysis tools Early, and mostly marketing copy claiming OEM references. Unproven, thin on negotiation knowledge, and holding no outcome data. A real race, just not a decided one.
Estimate review servicesHuman · offshore blueprinting Shops already pay people to do exactly this work. Not only a competitor — the proof of willingness to pay. The question is throughput and price, not whether the job is worth doing.
Tractable, Ravin, CCC AIWell funded Serious collision AI, sold to insurers for cost containment. They are pointed the other way. The shop side is left to small entrants and legacy data vendors, and that asymmetry is the opening.
Teardown$500–1,000/mo Shop-aligned, AI-native, with an insurer-outcome loop that compounds. Only useful to one side of the table — which is exactly why the incumbents can see it and still not build it.
Positions are strategic judgements, not measurements. The claim is about incentives, not capability.

“Why won't CCC just build this?” They can, easily. They won't, because shipping it means selling shops a tool to extract money from the insurers who are their larger customers.

That is counter-positioning: visible to the incumbent, cheap to build, and still irrational for them to adopt. It buys years, not forever — and the outcome data accumulated in those years is what makes it hard to catch afterwards.

The honest end-game. Acquisition by CCC, Mitchell/Enlyte, Solera, or an MSO tooling player. A $30–80M outcome is the realistic ceiling for this business, and it is not pitched as anything larger.

The Foundry

Eight weeks. One number that has to be real.

The plan is not to build more product. It is to put a working version in one paying shop and count the dollars it recovered that would not otherwise have arrived.

Weeks 1–2
Ten shop interviews Ten independent shops and small chains. What they supplement today, what comes back denied, who writes it, how long it takes, and what they already pay a human or an outside service to do. The shops already buying estimate review are the ones to listen to hardest — they have priced this problem themselves.
Weeks 3–4
One paid pilot signed One shop, on a flat monthly fee, paying. Paid, because a free pilot measures politeness rather than demand. In parallel, seed the carrier knowledge base for the two or three insurers that shop deals with most, from the public survey data and interviews with veteran estimators.
Weeks 5–6
Run it on live repairs Bay and Desk in the shop's hands on real teardowns. Every finding tracked from draft to outcome: approved, denied, settled or withdrawn — plus how many minutes the estimator actually spent on each package.
Weeks 7–8
Count the money Dollars recovered, per repair and in total, measured against that shop's own supplement history from the previous quarter. Their baseline, their records, their bank account — not ours.

One number goes on the last slide: dollars recovered for the pilot shop, checked against their own supplements from the quarter before.

Not signups, not demos, not a waitlist. If the owner cannot point at money that would not otherwise have arrived, nothing has been demonstrated and we should say so.

What would make us stop

If ten shops tell us the supplements they skip are worth under $100 each, the business isn't there. If our drafted lines get approved at a lower rate than what the shop's own estimator writes, the product isn't better than the person. If two carriers refuse the documentation package outright, the distribution assumption is broken. Any one of those inside eight weeks is a reason to stop, and finding it in week six is cheaper than finding it in year two.

Questions we expect, answered short

  • Isn't this claim inflation? No. Every line traces to a photo taken during teardown, an OEM procedure, or a line already on the estimate. If the work wasn't performed or isn't evidenced, no line is generated — a package that can't survive an adjuster reading it is worthless to the shop.
  • Why not just keep the veteran estimator? Keep him. This gives him several times the throughput and writes down what he knows, so the shop still has it after he leaves.
  • Who pays? The shop, $500–1,000 a month, flat. Contingency pricing exists as a pilot-only option and only where it is legal in that state.
  • What stops a carrier blacklisting the shop? Nothing in the package is adversarial — it is the documentation the carrier asks for, supplied properly. Better evidence usually shortens the argument rather than starting one.

Stage and fit

Pre-pilot. Positioning, competitive analysis and the knowledge-base design are written down; the build and the shop interviews start now. Industry: Software, SaaS & AI-enabled Products, sold direct to independent collision repairers.

Deliberately not positioned as an insurance product. Teardown works for the shop, documents work already performed, and never touches the consumer's claim payout.

Founder Leonard Eshun Software engineer with over seventeen years building national enterprise systems, and a recent data science graduate focused on AI. Claude Certified Architect — Foundations.
[YOUR ROUTE INTO COLLISION REPAIR: the shop, the estimator or the repair that showed you where the money was going]