Ten thousand records.One drop. Every row shows its work.

Drop what you have — repair orders, lot sheets, DVI exports, title feeds, photo folders, walk-around videos. WIN by VIN hashes it, parses it, maps the columns, validates every VIN by check digit, quarantines anything a machine read, and publishes the rest at your origin tier onto the VIN Passports that need it.

Use the CLI

250,000 rows per file500 MB per videoSHA-256 dedupe before upload0 machine reads published without a person

WHAT A ROW BECOMESEVIDENCE

VIN *17 characters; check digit verified per row
Event date *ISO, US or EU formats accepted
Title / description *What happened
Odometer (mi)Numeric; adds a mileage anchor
AmountInvoice total, sale price
Reference #RO, invoice, stock or lot number
Photo URL / filenameMatches uploaded images by filename

Columns are auto-mapped from common headers (VIN, service_date, odometer, description, invoice…) and can be corrected before commit.

FOUR STEPS · NOTHING IS WRITTEN UNTIL COMMIT

How a batch works.

Open a batch

Name the supplier and the evidence channel. Up to five batches open at a time.

Drop files or folders

Hashed in your browser, uploaded with progress, parsed or extracted on the server.

Review

Correct column mappings, read every row verdict — accepted, quarantined, rejected, duplicate — and the machine-read candidates.

Commit

Accepted rows publish at your origin tier. Quarantined items wait for a person. Win Points vest in 14 days.

TEMPLATES

Start from a template.

Any header names work — the mapper recognises the usual synonyms — but these templates map with zero corrections.

Service recordsvin, service_date, odometer, description, invoice, total, shopDownload service-records.csv →Auction lotsvin, sale_date, lot, odometer, announcement, hammer_price, auction_houseDownload auction-lots.csv →Inspectionsvin, inspected_at, odometer, findings, inspector, report_refDownload inspections.csv →Title eventsvin, event_date, jurisdiction, event, brand, odometer_disclosureDownload title-events.csv →Photo manifestvin, captured_at, angle, filename, odometerDownload photos-manifest.csv →JSON linesone object per line — any of the columns above, nested objects flattenedDownload records.jsonl →

AUTOMATION · REST + CLI

The executable.

The winbyvin CLI ships in the repo (bin/winbyvin.mjs) and talks to the same API as the studio. Resume-safe, hashed, concurrent, honest about every file.

<b>$</b> npx winbyvin import ./exports/*.csv ./photos/2026-08/ \
    --api https://winbyvin.com --key $WBV_PARTNER_KEY \
    --supplier "Northline Vehicle Lab" --channel inspection

<i>▸</i> batch bat_m0k3… opened · channel inspection
<i>▸</i> hashing 214 files … 3 identical (skipped)
<i>▸</i> uploading 211 files · 3 at a time ……………… done
<i>▸</i> dvi-export-2026-08.csv   1,204 rows · 1,188 accepted · 11 quarantined · 5 rejected
<i>▸</i> IMG_4471.HEIC             quarantined · captured 2026-08-18 · GPS · VIN candidate 4T1G11AK4MU123456
<i>▸</i> committed · 1,188 evidence items across 402 VINs · report written to winbyvin-report.json

<b>$</b> npx winbyvin status bat_m0k3… --api https://winbyvin.com --key $WBV_PARTNER_KEY

POST /api/ingest/batchesOpen a batch · { channel, supplier }

POST /api/ingest/batches/:id/filesmultipart · up to 50 `file` parts

POST /api/ingest/batches/:id/mapping{ fileId, mapping } · re-validate

POST /api/ingest/batches/:id/commitPublish accepted rows

GET /api/ingest/batches/:idStatus + verdicts

x-partner-key headerIssued to verified suppliers

QUESTIONS FROM SUPPLIERS

Straight answers.

What happens to a PDF or a photo?

It is hashed, stored, and machine-read: VIN candidates (check-digit verified), dates, odometer readings and amounts from PDFs; dimensions, capture date, camera and GPS presence from image EXIF; size and duration hints for video. Everything machine-read waits in quarantine at zero weight until a person confirms it against a VIN — the network never lets extraction overwrite identity.

What weight do my rows get?

Your origin tier. Verified shops and inspectors publish at 36–40/40, auction houses and jurisdictions at 34/40, dealers at their verified credential. Rows are labelled with your name on every VIN Passport they touch.

How are duplicates handled?

Files are hashed with SHA-256 in your browser before upload; identical files never leave your machine. Rows that repeat a VIN, date, title and reference are marked duplicate and earn nothing.

Can I automate it?

Yes. The same endpoints power the web studio, the REST API and the winbyvin CLI. Open a batch, POST files, correct a mapping if needed, commit. Partner keys are issued to verified suppliers.

Do rows earn Win Points?

Signed-in suppliers earn 2 Win Points per accepted row (bulk rate, vests after 14 days). Partner-key integrations are commercial agreements and are not pointed.

LOADINGVP / ···

ASSEMBLING RECORD

Pulling evidence, grading it, and showing its work.

Ten thousand records.One drop. Every row shows its work.

250,000 rows per file500 MB per videoSHA-256 dedupe before upload0 machine reads published without a person

$ npx winbyvin import ./exports/*.csv ./photos/2026-08/ \ --api https://winbyvin.com --key $WBV_PARTNER_KEY \ --supplier "Northline Vehicle Lab" --channel inspection ▸ batch bat_m0k3… opened · channel inspection ▸ hashing 214 files … 3 identical (skipped) ▸ uploading 211 files · 3 at a time ……………… done ▸ dvi-export-2026-08.csv 1,204 rows · 1,188 accepted · 11 quarantined · 5 rejected ▸ IMG_4471.HEIC quarantined · captured 2026-08-18 · GPS · VIN candidate 4T1G11AK4MU123456 ▸ committed · 1,188 evidence items across 402 VINs · report written to winbyvin-report.json $ npx winbyvin status bat_m0k3… --api https://winbyvin.com --key $WBV_PARTNER_KEY