Score C · 6.26/10Free to read$100K Hackathon · November 16, 2026 → December 16, 2026Nov 16, 2026

Chargemaster X-Ray

Turns a scanned hospital itemized statement into a line-by-line audit workbook plus a first-draft appeal packet for independent patient advocates, in under 20 minutes.

One product, one founder, thirty days. Everything below is built to be executed inside the contest window — the official rules still decide what counts.

Decide in 20 seconds

The facts that decide whether this is worth thirty days of yours.

Speculative6.26/10
MVP build (solo)
5 days
Launch price
$149
Sales to $1,000 revenue
7 customers
First paying customer
Cold email to named solo patient advocates with a 90-second Loom, offering a free audit returned in 24 hours.
Who pays

Owner-operator of a 1-5 person independent patient-advocacy or medical-billing-advocacy practice in the US, often a certified patient advocate or ex-hospital biller working hourly or on contingency.

Biggest execution risk

Operational simplicity 3.0/10

Remove regulated data, human review steps and custom infrastructure; keep the product self-serve and stateless on managed services.

Why this could work in 30 days: one founder can ship the MVP in 5 days; the plan targets a first sale on day 12; 7 sales at $149 clears $1,000 of collected revenue; strongest signal is speed to first revenue at 8.0/10.

Score and projections come from the generated report and are not guarantees. The $100K Hackathon is run by Ship or Die — read the official rules before you build.

Your first 24 hours

13 concrete actions. Tick them off as you go — nothing is saved or sent.

0/13
Entry price
$149
Solo build
5 days
30d base revenue
$2,688
30d hackathon profit
$2,567
Section 1 of 8

Product opportunity and positioning

Target customer

Owner-operator of a 1-5 person independent patient-advocacy or medical-billing-advocacy practice in the US, often a certified patient advocate or ex-hospital biller working hourly or on contingency.

Why now

The buyer already sells bill auditing as a service and already has the statement in hand, so there is no behaviour change to sell and no procurement cycle between the pitch and a $149 purchase.

Problem

Every new case opens with hours of retyping revenue codes, CPT/HCPCS codes, descriptions, units and charges off a scanned statement, then guessing which lines are duplicates, unbundled components, quantity mismatches or outliers, then retyping findings into appeal letters with no audit trail.

Solution

Advocate uploads a statement PDF or photo; vision-model OCR produces an editable line-item table with page anchors and confidence flags; a deterministic rules pass labels each line with a finding and dollar amount at issue; the product exports an Excel/CSV audit workbook and a print-ready appeal packet PDF, tracked per client case.

Differentiation · Spreadsheets and generic practice-management tools store the case but do not read the statement; this product does the transcription, the pattern flagging and the first-draft letter, and hands back a document the advocate must produce anyway.

Product opportunity and positioning

Chargemaster X-Ray sells to one named professional: the owner-operator of a 1-5 person independent patient-advocacy or medical-billing-advocacy practice in the US, usually a certified patient advocate or an ex-hospital biller working hourly or on contingency. Her trigger is concrete. A client hands over a 20-80 page hospital itemized statement and a payer or provider response deadline is days to weeks away.

Her first two hours on every case are identical and unbillable: retype revenue codes, CPT/HCPCS codes, descriptions, units and charges into a spreadsheet, then eyeball the sheet for exact duplicates, unbundled component charges, quantity mismatches and wild outliers. Then she retypes the surviving findings into appeal letters. Nothing about that work is judgement; it is transcription plus pattern matching plus typing, and it caps how many cases she can carry.

The JEV score for this concept is 62.6/100 (band C, confidence 0.66). The two strongest criteria are exactly the ones that decide a 30-day hackathon: speed to first revenue (8.0) and reachable customers (7.06). The buyer is a named professional with a published email address, a live client file and a document deadline, so a single cold email with a Loom can produce a paid case inside a week. The weakest criterion is operational simplicity (3.03), and the build plan answers it directly: no payer submission, no fax, no portal integration, no clinical coding advice, no human review queue, no 24/7 support promise. Upload in, workbook and letter out, documents auto-deleted after the retention window.

The positioning in one line: the fastest way to turn a hospital itemized statement into an audited spreadsheet and a signed-ready appeal packet. The competitor is not software, it is a blank Excel template and three hours of the advocate's own evening.

The wedge is deliberately narrow. It does not audit superbills, does not touch consumer patients, does not submit claims and does not bill on contingency. That narrowness is what makes a solo founder able to ship in five days and demo convincingly on the prospect's own case file.

Section 2 of 8

Smallest sellable MVP

Upload one statement PDF, get back an Excel audit workbook with per-line findings and dollar amounts plus a print-ready appeal packet PDF, behind a Stripe paywall.

Must-have features

  • PDF/JPG/PNG upload with vision-model OCR into an editable line-item table keeping page numbers
  • Confidence flags on low-confidence cells and a reconciliation check that summed line charges match the printed statement total
  • Deterministic rules pass: exact and near-duplicate lines, quantity/unit mismatches, same-day repeated charges, user-editable component-plus-comprehensive code pairs, outlier flag against a pasted benchmark column
  • Excel/CSV export with finding label, dollar amount at issue, confidence and suggested argument text
  • Appeal packet PDF merging each finding into an editable letter template with code, amount, deadline and advocate letterhead
  • Per-client case workspace with finding status (open, accepted, appealed, won) and running dollar total at issue
  • Stripe checkout for a $149 single case pass and a $249 monthly plan with per-statement metering and a hard cap
  • Automatic document deletion after the plan's retention window, with a deletion confirmation email

Explicitly out of scope

  • No submission to payers or providers, no fax, no portal or insurer email integration
  • No payer-specific policy database and no clinical coding interpretation or advice
  • No patient-facing app and no contingency or savings-share billing
  • No EHR, practice-management or clearinghouse integration and no public API
  • No team seats, roles, SSO or admin console
  • No mobile app in the first 30 days

Demo flow (what the buyer sees)

  1. Founder opens the intake link on a call or in a Loom with a real de-identified statement already uploaded
  2. Product OCRs 30 pages into a line-item table with page anchors in under two minutes
  3. Advocate sees confidence flags on three uncertain cells and corrects them in the grid
  4. Rules pass labels duplicate, unbundled and outlier lines with dollar amounts at issue
  5. Advocate clicks Export and receives the Excel workbook in email
  6. Advocate clicks Generate packet and receives a print-ready appeal PDF with their letterhead
Build plan

Technical implementation plan

Stack

Next.js on Vercel (serverless functions)Supabase Postgres with row-level security and private file storageVision-capable LLM API with JSON-schema structured output for line-item extractionpdf-lib for server-side appeal packet generationSheetJS for Excel/CSV workbook exportStripe Checkout plus Billing for the case pass and monthly planResend for transactional email and deliverable deliveryVercel Cron for retention-based document deletion

Architecture

A single Next.js app on Vercel talks to Supabase for auth, Postgres and private storage. Uploads go straight to a private bucket; a serverless job sends each page image to the vision model with a strict JSON schema and stores the returned line items keyed by page. A deterministic rules module runs over the stored rows (no model involved) and writes findings with dollar amounts. Exports are generated server-side from those rows: SheetJS for the workbook, pdf-lib for the letter packet. Stripe webhooks flip the case from locked to delivered and increment usage counters. A cron job deletes source documents past the retention window.

Build steps

  1. Day 1: repo, Vercel project, Supabase schema (cases, documents, line_items, findings, subscriptions) and Stripe test products
  2. Day 2: upload path plus page-by-page vision extraction with JSON schema, page anchoring and a reconciliation total check
  3. Day 3: editable line-item grid with confidence flags and manual corrections persisted
  4. Day 4: deterministic rules pass plus dollar-at-issue math and suggested-argument text
  5. Day 5: SheetJS workbook export and pdf-lib appeal packet, producing the sellable artifact
  6. Day 6: Stripe Checkout, webhook gating, usage metering, retention cron and Resend delivery
  7. Day 7: case workspace statuses, retention deletion confirmation and landing page
  8. Days 8-10: harden extraction on rotated scans, multi-column layouts and low-DPI photos

AI components

  • Vision LLM for OCR and structured line-item extraction with JSON schema output
  • LLM-drafted suggested-argument text per finding, generated from the finding type and the line's own data
  • LLM-drafted appeal letter paragraphs merged into the advocate's template
Hardest technical risk: Line-item extraction drift on skewed, multi-column or rotated scans, where a charge lands on the wrong code or page, silently corrupting the workbook. Mitigation: extract one page per call, anchor every row to its page, require the summed line charges to equal the printed page total, and surface a mismatch as a blocking warning rather than a silent pass.
Section 3 of 8

Pricing and monetization

Single Case Pass

Entry offer
$149 one-time

For A solo advocate testing the output on one live client file

  • One statement audit up to 150 line items
  • Excel/CSV audit workbook
  • One appeal packet PDF
  • 7-day document retention

Practice Plan

$249 / month

For A solo or 2-3 person advocacy practice running several files at once

  • Up to 10 statement audits per month
  • Unlimited appeal packet PDFs
  • Case workspace with finding status tracking
  • 30-day document retention
  • Email support

Model: A $149 one-time Single Case Pass for a test file, converting into a $249 per month Practice Plan with 10 statement audits and unlimited appeal packets.

First revenue offer: Send one de-identified or sample statement today; I will return the audit workbook and one appeal letter within 24 hours free, and if the findings are real you unlock the full workbook and PDF packet for $149.

Expected ACV: $224

Upsell path: Case pass buyers convert to the Practice Plan when they send a second case; plan buyers upgrade usage when they hit the 10-statement cap, and each converted advocate is asked for one peer introduction in the same email thread.

Target customers
30
Revenue model
$7,470
Costs
$260
Profit model
$7,210
Section 4 of 8

First-customer acquisition

Primary channel

Cold email to named solo patient advocates with a 90-second Loom, offering a free audit returned in 24 hours.

Where to find them

  • Public patient-advocate association member directories
  • State and regional advocacy group member listings
  • Independent medical-billing-advocacy practice websites with a named owner
  • LinkedIn searches for 'patient advocate' and 'medical billing advocate' filtered to owner or founder titles
  • Advocate-heavy subreddits and Facebook groups where practitioners discuss case workflow
  • Small-practice billing consultants listed on local business search results

Exact outreach motions

  • Build a 150-row list with owner name, practice name, direct email and one specific detail from their site
  • Send email one with the Loom link and the free-audit ask, subject referencing their actual job
  • Reply to free-audit acceptances within 24 hours with only the top two findings and their dollar amounts
  • Day-3 follow-up quoting finding count and total dollars found, not a reminder
  • Day-7 follow-up offering the $149 case pass to anyone still hesitating on $249
  • Ask every converted advocate for one peer introduction in the same thread

Message template

Subject: your itemized-statement review in 20 minutes

Hi {first_name} — you do itemized hospital statement reviews for clients. I built Chargemaster X-Ray: drop in the statement PDF, get back an Excel audit workbook with duplicates, unbundled components and outliers flagged with dollar amounts, plus a first-draft appeal packet with your letterhead.

90-second demo: {loom_link}

Want me to run one of your statements free? Send a de-identified or sample file and I will return the workbook within 24 hours, no charge and no card. If the findings are real, the full workbook plus packet is $149.

— {founder_name}
Week-one goal: 150 named cold emails sent, 10+ statements accepted for free audit, 3 paid conversions by day 12.
Section 5 of 8

30-day execution roadmap

MVP deadline: day 5First sale target: day 1230 days planned
Week 1 · Build & validate · 7 days
Day 1Foundations

Accounts and skeleton

  • Create Supabase, Vercel and Stripe products; connect Stripe to TrustMRR before any sale.
  • Commit Next.js skeleton and deploy a one-page placeholder with the intake form.

Deliverable: Live domain, Stripe test products, empty database schema.

Day 2Extraction

Vision OCR on one real statement

  • Build upload-to-private-storage and a page-by-page vision extraction job.
  • Enforce JSON schema output with page anchors and a page-total reconciliation check.

Deliverable: Line items extracted from one 30-page sample with page numbers.

Day 3Review UI

Editable line-item grid

  • Build the editable table with confidence flags and persisted manual corrections.
  • Show a blocking warning when summed charges do not match the printed total.

Deliverable: An advocate can correct three bad cells and see totals update.

Day 4Rules

Deterministic findings

  • Implement duplicate, quantity mismatch, same-day repeat and component-pair rules.
  • Add pasted benchmark column and outlier flag plus dollar-at-issue math.

Deliverable: Findings list with dollar amounts on the sample statement.

Day 5ArtifactsMVP shipped

Excel workbook and appeal PDF

  • Ship SheetJS workbook export with finding, amount, confidence and argument columns.
  • Ship pdf-lib appeal packet merging findings into an editable letter template.

Deliverable: MVP: one upload produces workbook plus print-ready packet.

Day 6Monetization

Paywall and metering

  • Wire Stripe Checkout for the $149 case pass and $249 plan with webhook gating.
  • Add usage metering, plan caps and the retention deletion cron.

Deliverable: End-to-end paid unlock of a delivered case.

Day 7List building

Case workspace plus 150-row list

  • Add finding status tracking and running dollar total per case.
  • Build the 150-row advocate list with names, emails and one site detail each.

Deliverable: Workspace live and a 150-row outreach list.

Week 2 · Launch & first sale · 7 days
Day 8Delivery

Email delivery and landing copy

  • Send workbook and packet by Resend with download links.
  • Write landing page copy around the 20-minute statement review.

Deliverable: Deliverables arrive by email; landing page published.

Day 9Demo asset

Loom and intake

  • Record the 90-second Loom on a synthetic statement showing three findings.
  • Publish the free-audit intake form with de-identification instructions.

Deliverable: Loom link and live intake form.

Day 10Outreach

First 40 emails

  • Send 40 cold emails with the Loom and the free-audit ask.

Deliverable: 40 sent, tracking sheet updated.

Day 11Free audits

Manual delivery

  • Run every accepted free audit by hand and log every extraction failure.
  • Fix the two worst OCR issues same day.

Deliverable: First free audit returned inside 24 hours.

Day 12RevenueFirst sale target

First paid conversion

  • Send the next 50 emails and follow up day-3 on the first batch.
  • Push the $149 case pass to anyone hesitating on $249.

Deliverable: First payment collected and receipt delivered.

Day 13Follow-up

Specific-number nudges

  • Send day-3 and day-7 follow-ups quoting finding count and dollars found.
  • Log reply reasons in a single sheet for message iteration.

Deliverable: Follow-up log with reply taxonomy.

Day 14Polish

Fix and re-demo

  • Ship the top three extraction fixes from live audits.
  • Post the Loom in two advocate communities where workflow is discussed.

Deliverable: Changelog plus two community posts.

Week 3 · Repeat & systematize · 7 days
Day 15Outreach wave 3

Remaining list

  • Send the remaining 60 emails from the 150-row list.
  • Follow up all non-responders from days 10 and 12.

Deliverable: Full list contacted once.

Day 16Outlier feature

Benchmark paste

  • Ship the paste-a-price-column outlier comparison with a clear source label.
  • Add a per-line note field for the advocate's own wording.

Deliverable: Benchmark feature live.

Day 17Pricing test

Case pass emphasis

  • Reorder the pricing page to lead with the $149 case pass.
  • Add a one-line comparison of pass versus plan usage.

Deliverable: Revised pricing page live.

Day 18Referrals

Second cases

  • Email every buyer asking for one peer introduction in-thread.
  • Prompt case-pass buyers to send a second case for the plan upgrade.

Deliverable: Referral requests sent to all buyers.

Day 19Letter quality

Template variants

  • Add two more appeal letter variants selectable per finding type.
  • Let the advocate save a letterhead and signature block once.

Deliverable: Three letter templates with saved letterhead.

Day 20Robustness

Bad scans

  • Handle rotated scans, multi-column layouts and low-DPI photos with a retry prompt.
  • Add a page thumbnails view to spot skipped pages.

Deliverable: Retry path shipped for failed pages.

Day 21Wave 4

Secondary list

  • Build a 60-row secondary list from billing-advocate practices.
  • Send 40 emails to the secondary list.

Deliverable: 60-row secondary list, 40 sent.

Week 4 · Scale what works · 7 days
Day 22Trust

Data handling clarity

  • Publish a plain-language data-handling page with retention and deletion specifics.
  • Add a deletion confirmation email on retention expiry.

Deliverable: Data-handling page and deletion email live.

Day 23Retention

Second-case nudges

  • Send a usage summary email to active accounts with statements remaining.
  • Nudge idle case-pass buyers with the second-case offer.

Deliverable: Usage email sent to all buyers.

Day 24Metrics

Funnel review

  • Compute emails sent, free audits accepted, paid conversions and reply reasons.
  • Rewrite the subject line and first sentence based on the best replying segment.

Deliverable: One-page funnel table and a revised message.

Day 25Self-serve

Remove founder from loop

  • Make upload, delivery and unlock fully self-serve without founder action.
  • Add an in-app empty state that tells the advocate what to upload.

Deliverable: Unassisted first paid case completed start to finish.

Day 26Upsell

Pass to plan

  • Add an in-app upgrade prompt at the statement cap.
  • Email every case-pass buyer the plan math in three lines.

Deliverable: Upgrade prompt live, buyers emailed.

Day 27Proof

Permission-based quotes

  • Ask each paying advocate for a written, attributed quote with explicit permission.
  • Publish only quotes actually received and approved.

Deliverable: Zero or more approved quotes on the landing page.

Day 28Final wave

Last outreach push

  • Send 40 final emails and follow up every warm thread.
  • Offer the $149 pass with a 48-hour turnaround promise.

Deliverable: Final outreach batch sent.

Days 29-30 · Consolidate · 2 days
Day 29Reconciliation

Money and fees

  • Reconcile every payment, refund and processing fee in the TrustMRR-connected account.
  • Confirm no customer or payment existed before launch day.

Deliverable: Reconciliation sheet matching the payment account.

Day 30Wrap

Close the loop

  • Deliver every open case and clear the free-audit backlog.
  • Write the post-contest plan: keep plan pricing, drop manual audits, keep the list warm.

Deliverable: All cases delivered and a written post-contest plan.

Section 6 of 8

Revenue and profit scenarios

How the numbers are defined

Hackathon profit — Collected revenue − refunds − chargebacks − payment processing fees − paid marketing spend.

Business profit — Hackathon profit − operating costs (hosting, domains, tooling, AI/API delivery costs).

The contest ranks entrants on hackathon profit only. Hosting, tooling and AI/API delivery costs are not deducted by the organisers, but each sale still has to cover them — so we show both. Definitions verified on 2026-11-15 against the official rules.

Stated assumptions

  • Base case assumes 12 paying customers in 30 days from 150 cold emails plus a 60-row secondary list; this is a hypothesis, not a measured conversion rate.
  • Blended ARPU reflects a mix of $149 one-time case passes and $249 monthly plans sold in the window, not a renewal assumption.
  • Payment fees assume 4.5% blended on collected revenue: Stripe domestic card processing at 2.9% plus $0.30, plus Stripe Billing recurring fees.
  • Ad spend is zero in conservative and base because acquisition is cold email and organic community posts; upside includes $200 for one paid community sponsorship.
  • Operating costs cover Vercel, Supabase, domain, Resend and vision/LLM inference per statement, priced at roughly $0.60-$1.00 of inference per 30-page audit plus fixed hosting.
  • Every sale is assumed to cover its own delivery cost; no scenario prices below delivery cost.
  • Refunds: only the upside scenario carries one refunded $149 case pass. Chargebacks assumed zero across all scenarios.
  • All figures are 30-day projections for the contest window, not verified revenue or existing traction.

Conservative

4 customers
$796
collected revenue · avg payment $199
Refunds
$0
Chargebacks
$0
Payment fees
$36
Paid marketing
$0
Hackathon profit
$760
Operating costs
$45
Business profit
$715

Base case

12 customers
$2,688
collected revenue · avg payment $224
Refunds
$0
Chargebacks
$0
Payment fees
$121
Paid marketing
$0
Hackathon profit
$2,567
Operating costs
$85
Business profit
$2,482

Optimistic

30 customers
$7,470
collected revenue · avg payment $249
Refunds
$149
Chargebacks
$0
Payment fees
$336
Paid marketing
$200
Hackathon profit
$6,785
Operating costs
$240
Business profit
$6,545

Model your own numbers

Runs in your browser · nothing is saved
Collected revenue
$1,788
Payment fees (3.9%)
$70
Paid marketing
$0
Hackathon profit
$1,718
Business profit
$1,633

Hypothetical arithmetic, not a forecast. The fee rate is a flat 3.9% assumption, refunds and chargebacks are left at zero here, and none of this is tracked data.

These are scenario projections built from stated assumptions. They are not verified revenue, existing customers, traction or commitments of any kind.

Section 7 of 8

Main risks and validation experiments

Risk 1Advocates refuse to upload client documents without a signed business associate agreement, stalling free-audit conversion.

Mitigation

Lead the first touch with a free audit on a de-identified or sample statement, and publish a plain-language data-handling page with retention and deletion specifics.

Validation experiment

Ask the first 20 repliers one question: would you send a de-identified statement today, and what would you need in writing first? Count the answer split.

Kill criteria

If fewer than 2 of 20 repliers will send any statement, even de-identified, the upload path is the blocker and the delivery model needs rework.

Risk 2Poor scans or unusual statement layouts produce wrong code extractions and destroy trust on the first case.

Mitigation

Page-by-page extraction with page anchors, a blocking reconciliation warning when line charges do not equal the printed page total, and confidence flags on uncertain cells.

Validation experiment

Run ten pages of varied real-world statement scans and measure the share of rows the advocate corrects versus accepts without edits.

Kill criteria

If more than 15% of rows require correction after the day-20 fixes, the product is not yet time-saving and should not be sold as 20 minutes.

Risk 3Advocates read findings as coding advice and fear liability for acting on a flagged line.

Mitigation

Label every finding as a pattern flag with the line, code and amount shown, state explicitly that the tool does not adjudicate codes, and keep the advocate's signature on the letter.

Validation experiment

Include the liability question in the first five free-audit reply emails and record objections verbatim.

Kill criteria

If 3 of the first 5 free-audit recipients cite liability as a reason not to buy, rework the finding language before more outreach.

Risk 4Cold-email conversion lands far below the assumed rate, making the customer target unreachable in 30 days.

Mitigation

Pre-build a 150-row primary list plus a 60-row secondary list, and send in three waves so message changes can be tested inside week two.

Validation experiment

Send 40 emails, then compare free-audit acceptance rate against the next 50 with a rewritten subject line.

Kill criteria

If 150 emails produce fewer than 3 free-audit acceptances, stop volume outreach and switch to therapist and billing-consultant referral asks.

Risk 5Subscription churn is high because solo advocates audit occasionally rather than weekly.

Mitigation

Keep the $149 case pass as the default entry and position the plan as a 10-statement bucket, not an unlimited promise.

Validation experiment

Track how many pass buyers return with a second statement inside 14 days.

Kill criteria

If under 25% of pass buyers return with a second case within 14 days, keep only per-case pricing and drop the monthly plan.

Section 8 of 8

Hour-by-hour: the first 24 hours

Hours 1-2

  • Create Supabase project with cases, documents, line_items and findings tables.
  • Create Vercel project, buy the domain and deploy the placeholder page.
  • Create Stripe test products for $149 one-time and $249 monthly; connect the account to TrustMRR before any sale.

Output: Live domain, four database tables, two Stripe products.

Hours 3-5

  • Wire the upload form to private Supabase storage with a signed URL return.
  • Send one sample statement page to the vision model with a strict JSON schema and inspect the raw output.

Output: One page extracted into structured line items.

Hours 6-8

  • Loop extraction over all pages with page numbers attached.
  • Add the reconciliation check comparing summed charges to the printed page total.

Output: Full 30-page statement table with a pass or fail on totals.

Hours 9-11

  • Build the minimal editable line-item table showing code, description, units, charge and confidence.
  • Persist manual corrections back to Postgres.

Output: Working review grid on the sample statement.

Hours 12-14

  • Ship the duplicate and quantity-mismatch rules with dollar-at-issue math.
  • Export a first SheetJS workbook to confirm the artifact opens cleanly in Excel.

Output: Requirement: a workbook download containing at least three findings.

Hours 15-18

  • Write the three-sentence positioning line and the free-audit intake copy.
  • Draft the 90-second Loom script and the first 20 rows of the advocate list.

Output: Intake copy, Loom script, 20-row outreach list.

Complete report

Full 30-day launch report

Read the full generated report

1. Product opportunity and positioning

Chargemaster X-Ray sells to one named professional: the owner-operator of a 1-5 person independent patient-advocacy or medical-billing-advocacy practice in the US, typically a certified patient advocate or an ex-hospital biller working hourly or on contingency. Her trigger is concrete. A client hands over a 20-80 page hospital itemized statement and a payer or provider deadline is days to weeks away.

Her first two hours on every case are identical and unbillable: retype revenue codes, CPT/HCPCS codes, descriptions, units and charges into a spreadsheet, then eyeball the sheet for duplicates, unbundled component charges, quantity mismatches and outliers. Then she retypes the surviving findings into appeal letters. None of that is judgement work; it is transcription, pattern matching and typing, and it caps how many cases she can carry.

JEV scores this concept 62.6/100 (band C, confidence 0.66). The strongest criteria are the ones that decide a 30-day contest: speed to first revenue (8.0) and reachable customers (7.06). The buyer has a published email, a live client file and a deadline, so one cold email with a Loom can produce a paid case inside two weeks. The weakest criterion is operational simplicity (3.03), and the build plan answers it directly: no payer submission, no fax, no portal integration, no coding advice, no human review queue, no 24/7 support promise. Upload in, workbook and letter out, documents auto-deleted after the retention window.

Differentiation is against a blank Excel template and three hours of the advocate's own evening, not against another vendor. The wedge stays narrow on purpose: it never touches a superbill, never touches a consumer, never submits a claim and never bills on contingency.

2. Smallest sellable MVP

One upload, two artifacts. The advocate drags a statement PDF or photo into the intake form. The product OCRs page by page into an editable line-item table that keeps page numbers and flags low-confidence cells. A deterministic rules pass labels each line as duplicate, quantity mismatch, same-day repeat, component-pair or outlier, with the dollar amount at issue and a suggested argument. Export produces an Excel/CSV workbook; Generate produces a print-ready appeal packet with the advocate's letterhead. Stripe gates the download at $149 for one case or $249 per month.

Out of scope for the full 30 days: payer submission, fax, portals, clinical coding interpretation, patient-facing anything, EHR integrations, team seats and a public API.

3. Technical implementation plan

Next.js on Vercel, Supabase Postgres plus private storage with row-level access, a vision-capable LLM API with JSON-schema structured output, SheetJS for the workbook, pdf-lib for the packet, Stripe Checkout and Billing, Resend for delivery, and a Vercel cron for retention deletion.

Flow: upload to private bucket, one vision call per page with the page number pinned into every returned row, persist rows, run the deterministic rules module (no model involved in flagging), render exports server-side, deliver by email. The single hardest risk is extraction drift on skewed, rotated or multi-column scans. The guard is arithmetic, not hope: summed line charges must equal the printed page total, and a mismatch surfaces as a blocking warning so the advocate never ships a silently corrupted workbook. Inference cost per 30-page audit is roughly $0.60-$1.00, which keeps every sale above its own delivery cost.

4. Pricing and monetization

Two prices, both immediately approvable by a solo practitioner: $149 one-time Single Case Pass (one audit up to 150 line items, workbook, one packet, 7-day retention) and $249 per month Practice Plan (10 audits, unlimited packets, case workspace with finding statuses, 30-day retention, email support). The plan exists so a $149 test converts into recurring revenue on the second case; the pass exists so nobody has to decide on a subscription to try it.

The offer this week is the free audit: send a de-identified statement, get the workbook and one letter back in 24 hours, unlock the full artifact for $149. Expected blended ARPU in the window is about $224, reflecting a mix of passes and plans.

5. First-customer acquisition

One channel: cold email to named solo advocates with a 90-second Loom. Build a 150-row primary list from public advocate directories, association member listings, independent practice websites and LinkedIn owner-title searches. Send in three waves so subject lines can be tested inside week two. The ask is never a demo call; it is one statement, returned as an audit within 24 hours.

Every free audit reply contains only the top two findings and their dollar amounts, with the full workbook behind the $149 unlock. Day-3 follow-ups quote finding count and total dollars found. Day-7 follow-ups offer the case pass to anyone hesitating on the plan. Every converted advocate is asked for one peer introduction in the same thread. Week-one target: 150 emails sent, 10 statements accepted, 3 paid by day 12.

6. Thirty-day execution roadmap

Days 1-5 build the artifact: skeleton and Stripe on day 1, page-level extraction on day 2, the editable grid on day 3, rules and dollar math on day 4, workbook and packet export on day 5. Day 5 is the MVP deadline and the first day a stranger could pay.

Day 6 adds the paywall, metering and retention cron. Days 7-9 build the 150-row list, the Loom and the intake form. Days 10-14 send the first 90 emails, run every free audit by hand, fix extraction bugs same day, and target the first payment on day 12. Days 15-21 send the remaining list plus a 60-row secondary list, ship the outlier benchmark feature, test pass-first pricing, and harden rotated and multi-column scans. Days 22-28 add data-handling clarity, usage nudges, a funnel review, self-serve completion without founder action, pass-to-plan upsell prompts and a final outreach wave. Days 29-30 reconcile every payment and fee against the TrustMRR-connected account, deliver all open cases and write the post-contest plan.

7. Revenue and profit scenarios

These are projections from stated assumptions, not verified revenue, customers or traction.

Conservative: 4 customers at $199 blended ARPU, $796 gross, $36 processing, $45 operating. Base: 12 customers at $224, $2,688 gross, $121 processing, $85 operating. Upside: 30 customers at $249, $7,470 gross, one $149 refund, $336 processing, $200 paid sponsorship, $240 operating.

Because hosting, domains and inference are not deducted from hackathon profit, contest profit is gross minus refunds, chargebacks, processing and paid marketing: $760 conservative, $2,567 base, $6,785 upside. Business profit subtracts operating cost: $715, $2,482 and $6,545 respectively. Payment fees assume Stripe at roughly 4.5% blended. No scenario prices below delivery cost.

8. Main risks and validation experiments

Data-handling refusal is the biggest killer, not competition. Test it in week one: ask the first 20 repliers whether they would send a de-identified statement today and what they need in writing first. If fewer than 2 of 20 will send anything, the upload path is the blocker.

Extraction accuracy is second. Run ten pages of varied real scans and measure the share of rows corrected versus accepted. Above 15% correction after the day-20 fixes, the 20-minute promise is false and pricing must change.

Liability perception is third: advocates may read flags as coding advice. Put the question in the first five free-audit replies and record objections verbatim; three of five citing liability means rework the flag language before more outreach.

Channel conversion is fourth: if 150 emails produce fewer than 3 free-audit acceptances, stop volume outreach and switch to referral asks from billing consultants. Churn is fifth: if under 25% of pass buyers return with a second case inside 14 days, keep per-case pricing only and drop the monthly plan.

9. First 24-hour action plan

Hours 1-2: Supabase project with four tables, Vercel deploy with the domain, Stripe test products, account connected to TrustMRR before any sale. Hours 3-5: upload to private storage and a single-page vision extraction with strict JSON schema. Hours 6-8: full-document extraction with page anchors and the reconciliation check. Hours 9-11: editable review grid with persisted corrections. Hours 12-14: duplicate and quantity rules with dollar-at-issue math, plus a first SheetJS workbook that opens cleanly. Hours 15-18: intake copy, Loom script and the first 20 rows of the advocate list.

This brief was generated from public data about established startups, then narrowed and scored for a solo 30-day launch. Revenue and profit figures are scenario models, not verified results, guarantees or evidence of existing customers. Validate demand yourself before you spend a month on it — and check the contest rules for what counts toward your score.

Related ideas

All ideas →