Surcharge Sift
Surcharge Sift reads a small shipper's road-freight invoices and returns a dispute-ready Overbilling Dossier that flags every over-billed fuel surcharge, waiting-time charge, pallet extra and duplicate line, with a ready-to-send claim email for each finding.
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.
Transport/logistics manager or finance lead at a 20-200 person manufacturer, distributor or wholesaler in France, Belgium or the Netherlands shipping roughly 30-500 full truckloads a year through 5-30 regional carriers and reconciling carrier invoices in spreadsheets.
Monetization potential 4.9/10
Tie the price to a measurable money or time outcome and add a recurring or repeat-purchase path instead of a single low-value charge.
Why this could work in 30 days: one founder can ship the MVP in 5 days; the plan targets a first sale on day 9; 11 sales at $99 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
8 concrete actions. Tick them off as you go — nothing is saved or sent.
Product opportunity and positioning
Target customer
Transport/logistics manager or finance lead at a 20-200 person manufacturer, distributor or wholesaler in France, Belgium or the Netherlands shipping roughly 30-500 full truckloads a year through 5-30 regional carriers and reconciling carrier invoices in spreadsheets.
Why now
The buyer already owns the input files, the output is one document carrying a euro number, there is no integration, procurement cycle or regulatory approval, and the price sits far below any approval threshold.
Problem
Carriers add fuel surcharges, waiting-time charges, pallet extras and diverse lines that nobody verifies; the manager eyeballs totals, disputes only the obvious errors, and small overcharges are paid silently because a full line-by-line check against the rate card and fuel index takes hours nobody has.
Solution
Upload invoice PDFs plus an optional rate-card CSV; the pipeline extracts normalised line items, recomputes arithmetic, detects duplicates, compares fuel surcharge percentages to a maintained published index and flags rate deviations, then returns an Overbilling Dossier PDF with evidence quotes, euro deltas, confidence labels and one claim email per carrier, plus a findings CSV.
1. Product opportunity and positioning
Carrier invoices arrive as PDFs, get eyeballed for the total, and get paid. The fuel surcharge on line 4 is compared to nothing. An 85 EUR waiting-time charge on a 40-minute stop is compared to nothing. Duplicate pallet-exchange lines survive because nobody re-keys the arithmetic. A 20-200 person manufacturer shipping 30-500 truckloads a year across 5-30 regional carriers has the files and the motive but not the hours: a line-by-line check against the contract rate card and a published fuel index is half a day per carrier panel, every month, for a saving that is real but diffuse.
The alternatives are out of reach rather than missing. Freight-audit firms and Transport Management Systems price and integrate above this buyer. A consultant costs more than the overbilling. So the buyer pays the overcharge and moves on, and the overcharge repeats next month.
Surcharge Sift sells the check as a document, not as software. One upload, one PDF, one euro figure per finding, one claim email per carrier. No system to adopt, no ERP connector, no training, no seat licence. The buyer forwards the dossier to a carrier and waits for a credit note. That is the whole job to be done, and it happens the same afternoon the invoice batch lands.
Positioning sentence: Surcharge Sift is the invoice auditor a mid-sized shipper can buy this afternoon, and it returns a dispute-ready Overbilling Dossier rather than a dashboard.
The buyer is narrow on purpose: a named human with a spreadsheet, in France, Belgium or the Netherlands, already drowning in carrier PDFs. The wedge is narrow on purpose too: no marketplace, no carrier network, no payments, no tracking, no emissions module. Everything that made the original freight platform two-sided and slow is deliberately excluded, because a solo founder cannot conjure carrier liquidity inside 30 days but can read a PDF and produce a document with a euro number on it.
JEV scores this concept 61.4/100 (band C) and names monetization as the weak axis at 4.92, precisely because a single $99-$249 one-off charge carries no repeat path. The plan answers that rather than ignoring it. Every delivered dossier ends with a next-quarter offer, the Deep Audit tier carries the rate-card comparison that justifies the top price, and day 18 is spent asking each buyer to book the next batch before the contest closes. Conversion, not feature count, decides whether that answer holds.
Smallest sellable MVP
A paid Overbilling Dossier for one uploaded batch of carrier invoice PDFs: upload, automated line-item extraction, deterministic overbilling findings, a PDF dossier with euro deltas and claim emails, unlocked by Stripe payment.
Must-have features
- Magic-link upload page for a batch of invoice PDFs plus optional rate-card CSV, no accounts or passwords
- LLM structured extraction of every invoice line into a fixed schema: invoice no, date, carrier, lane, weight, base rate, fuel %, waiting time, extras
- Deterministic rules engine: arithmetic recomputation, duplicate invoice and duplicate line detection, fuel surcharge % against a maintained FR/BE/NL published index, rate deviation against the uploaded rate card
- Server-rendered Overbilling Dossier PDF with the verbatim source line, expected vs billed value, euro delta and a confidence label per finding
- One generated claim email per carrier, copy-paste ready
- Free Sample Sift: first 10 invoices produce a one-page findings summary automatically
- Findings CSV export and Stripe Checkout with webhook that unlocks the full dossier inside the job page
Explicitly out of scope
- No TMS, ERP or accounting integration
- No carrier-side product, carrier onboarding or load marketplace
- No payment execution, factoring or guaranteed carrier payment
- No legal advice and no automated sending of claim emails
- No multi-seat accounts, roles or approval workflows
- No live shipment tracking or emissions reporting
- No subscription billing inside the 30-day window
Demo flow (what the buyer sees)
- Buyer receives the LinkedIn offer and uploads 10 recent carrier invoice PDFs via the magic-link page
- Pipeline extracts line items and the rules engine computes findings
- Free one-page Sample Sift summary is returned, showing euro deltas found
- Buyer pays $99 or $249 through Stripe Checkout on the job page
- Webhook unlocks the full Overbilling Dossier PDF plus findings CSV
- Buyer copies a claim email per carrier and sends it from their own mailbox
Technical implementation plan
Stack
Architecture
Upload writes PDFs and the rate-card CSV to storage and creates a job row with a magic-link token; no accounts. A serverless job route extracts text from each PDF and calls the LLM once per invoice with a strict JSON schema, then discards the model's own arithmetic. A deterministic rules engine recomputes every line, detects duplicates, compares fuel surcharge percentages to the committed index CSV, and diffs rates against the uploaded rate card. Findings are stored as rows and rendered twice from the same data: a free one-page summary and the full dossier PDF gated behind payment. Stripe Checkout is created server-side against the job id; the webhook verifies the signature and flips the job to paid, which changes the download route from teaser to full dossier plus CSV.
Build steps
- Scaffold Next.js on Vercel, create Supabase project, storage buckets and the jobs/findings tables
- Build the upload page and magic-link job page with token access
- Add PDF text extraction and the LLM structured-extraction call with a fixed line-item schema
- Commit the FR/BE/NL fuel-surcharge index CSV with source dates and a loader
- Implement the rules engine: arithmetic, duplicates, fuel % comparison, rate-card deviation
- Render the free one-page Sample Sift summary
- Render the full Overbilling Dossier PDF with evidence quotes and confidence labels
- Generate claim email text per carrier from findings
- Add Stripe Checkout, webhook and paid-state gating on the download route
- Test end to end on 10 real anonymised invoices from at least 3 carriers
- Deploy to the production domain and run a live paid test order followed by a refund
AI components
- LLM structured extraction converting raw PDF text into a normalised line-item JSON array per invoice
- LLM drafting of the per-carrier claim email from the already-computed findings, never from raw invoice text alone
Pricing and monetization
Quarter Audit
Entry offerFor Small shipper auditing one quarter of invoices from up to three carriers
- Overbilling Dossier PDF for up to 150 invoices
- Findings CSV
- Copy-paste claim email per carrier
Deep Audit
For Shipper auditing a full year or a multi-carrier panel who wants the rate-card comparison
- Overbilling Dossier PDF for up to 500 invoices and up to eight carriers
- Contract rate-card deviation analysis
- Prioritised claim list ranked by euro value
- One re-run after corrections
Model: Two one-time audit packages priced by batch size, with a free 10-invoice sample as the sales instrument and a next-quarter re-order as the repeat path.
First revenue offer: Send 10 recent carrier invoices and I will return a free one-page findings summary within an hour. If it shows real overbilling, the full dossier covering your whole quarter is $99 and lands the same day.
Expected ACV: $144
Upsell path: Delivered dossiers end with a next-quarter offer at $99, converting a one-off purchase into a quarterly audit cycle; buyers with multi-carrier panels are moved to the $249 Deep Audit with rate-card deviation and a prioritised claim list.
First-customer acquisition
Direct LinkedIn outreach to named transport, logistics and finance leads at 20-200 person manufacturers and distributors in France, Belgium and the Netherlands, offering a free automated 10-invoice Sample Sift.
Where to find them
- LinkedIn search on 'responsable transport' at French manufacturers and distributors with 20-200 employees
- LinkedIn search on 'transport manager' and 'supply chain manager' in Belgium and the Netherlands
- Company pages showing a small logistics team and multi-site distribution
- European freight and supply-chain LinkedIn groups where shippers ask invoice and carrier questions
- Supply-chain Slack and Discord communities with active FR/BE/NL membership
- Small accounting firms serving transport and wholesale SMEs
- Anonymised invoice batches volunteered by network contacts for pipeline proof runs
Exact outreach motions
- Send a 25-message wave offering to run ten recent carrier invoices through the free Sample Sift and return a one-page summary, no meeting required
- Follow up once after four days with the sample summary re-attached and a single-line ask for the full batch
- Return every sample within one hour of receiving files so the offer reads as a service, not a demo
- Convert any sample with a euro finding into a paid Quarter Audit ask with the delta already visible in the message
- Post the same sample-audit offer in three European freight and supply-chain LinkedIn groups
- Message finance leads with a 'we found euro X on your invoices' angle and an anonymised findings page
- Ask every delivered buyer to book next quarter's batch at $99 before the contest closes
Message template
Hi [Name] - I run carrier invoice audits for shippers your size. Send me 10 recent carrier invoice PDFs and I will return a one-page summary of fuel surcharges, waiting time, pallet extras and duplicate lines that do not match your rate card or the published index. Free, back within an hour, no call needed. If it finds nothing I will tell you that plainly.
30-day execution roadmap
Week 1 · Build & validate · 7 days
Skeleton and upload
- Scaffold Next.js on Vercel with Supabase Postgres and storage buckets.
- Ship the multi-PDF upload page with optional rate-card CSV.
- Create job rows with magic-link tokens, no passwords.
Deliverable: Deployed upload page issuing a working magic-link job URL.
PDF to line items
- Wire PDF text extraction and the LLM structured-extraction call to a fixed schema.
- Normalise invoice no, date, carrier, lane, weight, base rate, fuel %, waiting time and extras.
- Run 5 real anonymised invoices and fix the worst parse failures.
Deliverable: Normalised line-item JSON for a 5-invoice test batch.
Findings logic
- Implement line arithmetic recomputation and duplicate invoice and line detection.
- Commit the FR/BE/NL published fuel-surcharge index CSV and compare billed percentages against it.
- Add rate deviation checks against the uploaded rate-card CSV.
Deliverable: Findings JSON with a euro delta and confidence label per finding.
Dossier rendering
- Render the Overbilling Dossier PDF with verbatim evidence quote, expected vs billed, delta and confidence.
- Generate one claim email per carrier from the findings.
- Build the free one-page Sample Sift summary from the first 10 invoices.
Deliverable: Sample dossier PDF and findings CSV from the test batch.
Stripe and go live
- Add Stripe Checkout and the webhook that unlocks the full PDF on the job page.
- Run a live paid test order end to end and refund it.
- Publish the pricing page and confirm the TrustMRR-connected payment account is active.
Deliverable: Live paid product on a real domain with one tested order.
Named buyers
- Build a 60-company list from LinkedIn on 'responsable transport' and 'supply chain manager' in FR/BE, 20-200 employees.
- Capture name, role, company, city and an invoice-pain signal per row.
- Finalise the free Sample Sift offer copy.
Deliverable: 60-row prospect CSV plus final offer copy.
Proof runs
- Run Sample Sift on 10 anonymised invoices sourced from two test contacts.
- Fix extraction failures and tighten confidence thresholds.
- Draft the day-12 follow-up message for non-responders.
Deliverable: Two clean sample summaries and a hardened extraction prompt.
Week 2 · Launch & first sale · 7 days
First 25 sends
- Send 25 personalised LinkedIn messages with the free 10-invoice Sample Sift offer.
- Log every send and reply in one sheet.
Deliverable: 25 sends logged and first replies in the inbox.
Deliver samples fast
- Run and return every sample within one hour of receiving files.
- Convert findings into a paid Quarter Audit ask with the euro delta visible.
Deliverable: First paid dossier requested or paid.
Volume and communities
- Send 25 more messages.
- Post the sample-audit offer in three European freight and supply-chain LinkedIn groups.
Deliverable: 50 total sends and three community posts live.
Deliver and fix
- Deliver all paid dossiers within 24 hours of payment.
- Fix every extraction error a buyer reports the same day.
Deliverable: Delivered dossiers and zero open extraction bugs.
Second touch
- Follow up day-8 non-responders with the sample summary re-attached.
- Ask each delivered buyer for next quarter's batch.
Deliverable: One follow-up per prospect and one re-order ask per buyer.
Fuel surcharge table
- Update the FR/BE/NL fuel index rows for the current month.
- Add a source-date line to the dossier so buyers can audit the comparison.
Deliverable: Refreshed index CSV deployed.
Funnel numbers
- Compute reply rate, sample-to-paid rate and revenue to date.
- Kill the worst-performing message angle.
Deliverable: One-page funnel snapshot.
Week 3 · Repeat & systematize · 7 days
NL segment
- Send 30 messages to Dutch and Belgian prospects.
- Localise the offer line and claim email template into Dutch.
Deliverable: 30 NL/BE sends and an NL claim-email template.
Rate-card deviation
- Improve rate-card matching for lane naming and weight-break variance.
- Rank findings by euro value in the Deep Audit dossier.
Deliverable: Deep Audit dossier with a prioritised claim list.
CFO and DAF buyers
- Send 20 messages to finance leads with a 'we found euro X on your invoices' angle.
- Attach an anonymised findings page as proof of format.
Deliverable: 20 finance-lead sends logged.
Quarterly cycle
- Offer each past buyer a standing next-quarter audit at $99 booked before December 16.
- Add the re-order ask to the dossier's final page.
Deliverable: Re-order ask shipped and outcome data recorded per buyer.
Community posts
- Post an anonymised findings summary with amounts redacted in two communities.
- Reply to every comment with the sample offer.
Deliverable: Two posts live and all comment replies handled.
Trust repair
- Review every finding a buyer disputed and downgrade or delete that rule.
- Recalibrate confidence labels so high means arithmetic-certain.
Deliverable: Updated rule set and a calibration note.
Profit check
- Reconcile Stripe payouts against TrustMRR records.
- Decide whether the final week pushes Deep Audit or Quarter Audit.
Deliverable: Reconciled revenue statement.
Week 4 · Scale what works · 7 days
Year-end close
- Send 30 messages framed on year-end invoice close and budget recovery.
- Follow up every open thread once.
Deliverable: 30 sends plus follow-ups logged.
Large batches
- Add ZIP upload and per-invoice progress so 150-invoice batches do not stall.
- Cap upload size and show a clear error on unreadable PDFs.
Deliverable: ZIP upload live in production.
Accountants serving logistics SMEs
- Message 10 small accounting firms serving transport SMEs with a referral-style sample offer.
- Give them a resendable one-page summary of what the dossier contains.
Deliverable: 10 accounting-firm conversations opened.
Clear the queue
- Batch-render all outstanding dossiers and send them.
- Chase any payment that landed without files attached.
Deliverable: Zero dossiers pending.
Conversion
- Rewrite the pricing page around the euro-delta outcome and the two batch limits.
- Add an FAQ on invoice confidentiality and data deletion.
Deliverable: Updated pricing page and FAQ live.
Deadline
- Send 20 messages with a December 16 delivery guarantee.
- Post the last-call offer once in each community.
Deliverable: 20 sends and two community posts.
Segment numbers
- Recompute conversion by message angle and by segment.
- Stop the lowest-yield segment entirely.
Deliverable: Segment performance table.
Days 29-30 · Consolidate · 2 days
Close everything
- Follow up every unpaid job page with a single-line reminder.
- Deliver same-day for anything paid before 13:00 UTC.
Deliverable: All open threads closed or explicitly dropped.
Reconcile and learn
- Reconcile final revenue, refunds, fees and ad spend against TrustMRR.
- Write a one-page post-mortem with the three numbers that mattered.
Deliverable: Final reconciled profit figure and post-mortem.
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
- Blended ARPU of $144 assumes 28 Quarter Audits at $99 and 12 Deep Audits at $249 (28x99 + 12x249 = 5760); this mix is a projection, not observed behaviour (Base/Upside)
- Conservative assumes 150 outreach messages converting at roughly 2% paid (Conservative)
- Base assumes the free Sample Sift converts at 15-20% once a euro finding is visible (Base)
- Upside assumes 25% sample-to-paid conversion and two re-orders per ten buyers, matching the concept's own 40-buyer projection (Upside)
- Payment fees modelled at 4% blended on collected revenue, assuming Stripe card processing plus currency conversion on some EUR cards
- Ad spend is zero across all scenarios because the acquisition channel is organic LinkedIn outreach and community posting
- Operating cost covers Vercel, Supabase, domain and LLM extraction at roughly $2 per delivered order plus fixed hosting
- No customers, revenue or partnerships exist at the time of writing; these are projections to be falsified in week one
Conservative
3 customers- Refunds
- $0
- Chargebacks
- $0
- Payment fees
- $17
- Paid marketing
- $0
- Hackathon profit
- $415
- Operating costs
- $25
- Business profit
- $390
Base case
12 customers- Refunds
- $0
- Chargebacks
- $0
- Payment fees
- $69
- Paid marketing
- $0
- Hackathon profit
- $1,659
- Operating costs
- $70
- Business profit
- $1,589
Optimistic
40 customers- Refunds
- $0
- Chargebacks
- $0
- Payment fees
- $230
- Paid marketing
- $0
- Hackathon profit
- $5,530
- Operating costs
- $180
- Business profit
- $5,350
Model your own numbers
Runs in your browser · nothing is saved- Collected revenue
- $1,188
- Payment fees (3.9%)
- $46
- Paid marketing
- $0
- Hackathon profit
- $1,142
- Business profit
- $1,072
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.
All three scenarios are projections derived from stated assumptions, not verified revenue. They describe hypothetical outcomes for a product with no customers at the time of writing.
Main risks and validation experiments
Risk 1Carrier invoice PDFs vary so widely that extraction misreads lines and produces false findings.
Every finding quotes the source line verbatim, arithmetic findings are computed in code not by the model, and low-confidence fields are labelled as needing the buyer's check rather than asserted.
By day 7 run 20 invoices from 5 different carriers through the pipeline and count invoices that fail to yield usable line items.
If more than one in five invoices fails to parse, narrow the offer to specific carrier formats rather than shipping guesses.
Risk 2False-positive overcharge claims damage the buyer's carrier relationships and trigger refunds.
Conservative rule thresholds, explicit confidence labels, and a day-20 review that deletes any rule generating disputed findings more than twice.
Track every finding a buyer disputes during weeks two and three and attribute it to a named rule.
If refund or dispute rate exceeds 10% of paid orders, remove the offending rule class or stop selling the tier that uses it.
Risk 3Buyers refuse to upload commercially sensitive invoices to an unknown solo-founder domain.
Free 10-invoice sample, redaction tolerated, a plain data-deletion promise in the FAQ, no accounts, and a deletion after 30 days default.
Measure how many of the first 25 prospects actually send invoice files within five days of the offer.
If fewer than 3 of 25 send files, change the entry point to screenshot or CSV upload before spending more outreach time.
Risk 4A free-first funnel generates sample runs and API cost with no payments.
Cap samples at 10 invoices, cap total free runs at 25, and always attach the euro delta before asking for payment.
After 20 completed samples, compute sample-to-paid conversion against the 25% assumption.
If sample-to-paid conversion is under 10% after 20 samples, stop the free tier and sell the paid audit directly.
Risk 5Payment processing availability, account review or a blocked payout prevents collecting revenue inside the contest window.
Confirm the TrustMRR-connected payment account and complete a live charge on day 5, not day 25, and keep a second processor configured but unused.
Run one live paid test order and one refund on day 5 and confirm the payout settles.
If the account is under review and cannot settle by day 7, switch processors immediately rather than continuing outreach.
Hour-by-hour: the first 24 hours
Hours 1-2
- Create the GitHub repo, Vercel project, Supabase database and storage buckets.
- Deploy an empty Next.js app to a staging domain.
Output: A live deployed URL and a working Supabase connection.
Hours 3-4
- Build the upload page for multiple PDFs plus optional rate-card CSV.
- Create job rows with magic-link tokens and a job page route.
Output: Files land in storage and a tokenised job page renders.
Hours 5-6
- Wire PDF text extraction and the LLM structured-extraction call against the fixed line-item schema.
- Test on one real anonymised invoice and inspect the JSON.
Output: Line-item JSON from one real carrier invoice.
Hours 7-8
- Commit the FR/BE/NL fuel-surcharge index CSV with current-month rows and source dates.
- Write the loader that the rules engine will call on day 3.
Output: Versioned index file committed and loadable.
Full 30-day launch report
Read the full generated report
Surcharge Sift - 30-Day Launch Report
Contest window: 16 Nov 2026 13:00 UTC to 16 Dec 2026 13:00 UTC. One founder. The ranking metric is verified profit, so every decision below optimises for money collected inside the window rather than product breadth.
1. Product opportunity and positioning
Carrier invoices arrive as PDFs, get eyeballed for the total, and get paid. The fuel surcharge on line 4 is compared to nothing. An 85 EUR waiting-time charge on a 40-minute stop is compared to nothing. Duplicate pallet lines survive because nobody re-keys the arithmetic. A 20-200 person manufacturer shipping 30-500 truckloads a year across 5-30 regional carriers has the files and the motive but not the hours: a line-by-line check against the contract rate card and a published fuel index is half a day per carrier panel, every month, for a saving that is real but diffuse.
The alternatives are out of reach rather than missing. Freight-audit firms and Transport Management Systems price and integrate above this buyer, and a consultant costs more than the overbilling. So the buyer pays the overcharge and moves on.
Surcharge Sift sells the check as a document. One upload, one PDF, one euro figure per finding, one claim email per carrier. No system to adopt, no ERP connector, no training. The buyer forwards the dossier and waits for a credit note.
Positioning: the invoice auditor a mid-sized shipper can buy this afternoon - a dispute-ready dossier, not a dashboard.
JEV scores the concept 61.4/100 in band C and flags monetization as the weak axis at 4.92, because a one-off $99-$249 charge has no repeat path. The plan answers that rather than ignoring it: every dossier ends with a next-quarter offer, the Deep Audit tier carries the rate-card comparison that justifies the top price, and day 18 is spent asking each buyer to book the next batch before the contest closes.
2. Smallest sellable MVP
The sellable unit is a paid Overbilling Dossier for one uploaded batch, and nothing else ships by day 5: magic-link upload, LLM extraction into a fixed line-item schema, a deterministic rules engine (arithmetic, duplicates, fuel percentage against the index, rate-card deviation), a dossier PDF with evidence quote and euro delta and confidence label, one claim email per carrier, and Stripe Checkout that unlocks the full PDF.
The free Sample Sift - first 10 invoices, one-page summary - is the same pipeline with a different renderer. It is the sales instrument, not a second product. Out of scope for 30 days: TMS or ERP integrations, any carrier-side feature, dispute representation, automated claim sending, team seats, tracking and emissions.
3. Technical implementation plan
Next.js on Vercel, Supabase Postgres and storage, one LLM structured-extraction call per invoice, server-side PDF rendering, Stripe Checkout plus webhook, and a versioned CSV of published FR/BE/NL fuel-surcharge indices committed to the repo.
Upload writes files to storage and creates a job row with a magic-link token. A worker route extracts text, calls the model with a strict JSON schema, then hands off to the rules engine where all arithmetic is recomputed in code. The job page shows the free one-pager; payment flips the job state and the webhook releases the full dossier and CSV.
Hardest risk is layout variance producing extraction errors. The mitigation is architectural: every finding quotes the source line verbatim, arithmetic is deterministic, and any field below a parse-confidence floor is labelled as needing the buyer's check. A wrong euro claim costs the buyer a carrier relationship, so the product must be able to say nothing found here cleanly.
4. Pricing and monetization
Two one-time tiers: Quarter Audit at $99 for up to 150 invoices across three carriers, and Deep Audit at $249 for up to 500 invoices, eight carriers, rate-card deviation, a prioritised claim list and one re-run. Blended planning ARPU is $144, modelled as 28 Quarter Audits plus 12 Deep Audits.
Delivery cost per order is roughly $0.30 to $2 in extraction and rendering, so every sale clears its delivery cost by a wide margin - the constraint the contest enforces with disqualification.
No subscription is sold inside the window; promising one before there is evidence would be fiction. The repeat path is behavioural and cheap: the last page of every dossier offers next quarter's batch at $99, and day 18 asks each buyer to book it before the contest closes.
5. First-customer acquisition
One channel: direct LinkedIn outreach to named transport and finance leads, with a free automated sample audit as the reason to reply. 150 prospects over four weeks from searches on responsable transport, transport manager and supply chain manager at 20-200 person manufacturers and distributors in France, Belgium and the Netherlands, supplemented by posts in European freight and supply-chain groups and a small accounting-firm angle on day 24.
The message never asks for a meeting. It offers to run ten recent invoices and return a one-page findings summary within an hour. When the sample finds nothing, say so plainly and stop - that honesty is the differentiator against every audit firm that would manufacture urgency.
Week-one goal: 25 sends, invoice files from at least 3 prospects, 1 paid dossier.
6. 30-day execution roadmap
Days 1-5 build, with a hard gate: on day 5 a real card buys a real dossier end to end and the payment is refunded. If checkout is not live on day 5, selling time is stolen from the only resource that cannot be re-bought.
Days 6-14 sell. Day 6 builds the 60-row prospect CSV, day 7 proves the pipeline on anonymised invoices, days 8, 10 and 12 send and follow up. Day 9 is the first-sale target. Day 13 refreshes the fuel index; day 14 reviews reply rate and kills the weakest message angle.
Days 15-21 convert and defend. Dutch localisation, Deep Audit rate-card depth, a finance-lead angle, the next-quarter re-order ask, and day 20 spent fixing every false positive a buyer disputes. Day 21 reconciles Stripe against TrustMRR.
Days 22-30 close. Year-end framing, ZIP upload for large batches, an accounting-firm angle, a fulfilment sprint, a rewritten pricing page and two final outreach waves. Day 29 is last call; day 30 reconciles profit and writes the post-mortem.
7. Revenue and profit scenarios
These are projections built from stated assumptions, not verified revenue. No customers exist at the time of writing.
| Scenario | Buyers | ARPU | Gross | Fees 4% | Refunds | Ad spend | Operating | Hackathon profit | Business profit | | Conservative | 3 | $144 | $432 | $17 | $0 | $0 | $25 | $415 | $390 | | Base | 12 | $144 | $1,728 | $69 | $0 | $0 | $70 | $1,659 | $1,589 | | Upside | 40 | $144 | $5,760 | $230 | $0 | $0 | $180 | $5,530 | $5,350 |
Payment fees assume Stripe at roughly 4% blended on card volume. Ad spend is zero because the channel is organic outreach - a deliberate risk reduction, not an oversight. Operating covers Vercel, Supabase, domain and LLM extraction at about $2 per order plus fixed hosting.
The conservative case is the honest floor: 150 messages at roughly 2% paid conversion. The upside case matches the concept's own 40-buyer projection and should be treated as a ceiling that requires 25% sample-to-paid conversion.
8. Risks and validation experiments
Extraction accuracy is existential. Test 20 invoices from 5 carriers by day 7; if more than one in five fails to parse, narrow the offer to two carrier formats instead of shipping guesses.
False positives destroy trust faster than a missed overcharge. Day 20 rule review plus per-finding confidence labels; delete any rule that generates disputed findings more than twice.
Buyers may refuse to upload commercial invoices. The free sample on 10 redacted invoices is the test; if fewer than 3 of 25 prospects send files in week one, change the entry point.
Free-first funnels can burn API cost with no revenue. Cap samples at 10 invoices and 25 runs; if sample-to-paid conversion is under 10% after 20 samples, stop sampling and sell the paid audit directly.
A blocked payment account would end the contest. Verify the TrustMRR-connected checkout with a live charge on day 5, not day 25.
9. First 24 hours
Hours 1-2: repo, Vercel project, Supabase database and storage, empty app deployed. Output: a live URL.
Hours 3-4: upload page and job creation with magic-link tokens. Output: files in storage, tokenised job page rendering.
Hours 5-6: PDF text extraction plus the LLM extraction call against the fixed schema, tested on one real invoice. Output: line-item JSON.
Hours 7-8: commit the FR/BE/NL fuel-surcharge index CSV skeleton with current-month rows and source dates. Output: a versioned index file the day-3 rules engine can load.
Anything unfinished in these eight hours moves to day 2. The day-5 gate does not move.
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 →RateCon Rapid
Turns any broker load email into a signed-ready rate confirmation PDF in 30 seconds, with an automatic profit check for small trucking companies.
VTC Profit Lens
VTC Profit Lens turns weekly Uber/Bolt/Heetch CSV exports into a one-page true-profit report for independent drivers, showing which platform and shifts actually pay after fuel and commission.
Preorder Cap
Preorder Cap gives independent bakeries and florists a hosted live-capacity page that self-decrements as preorder slots are claimed, exports sold-out days as a calendar feed, and generates the Instagram and WhatsApp reply text the owner retypes a hundred times a week.