Custom AI for cargo
damage documentation

Photos at pickup, transit, and delivery come back with damage localized and labeled, ready for your dispute file.

Production API in four to six weeks. You walk with the discovery deliverables if V1 misses the targets we set together.

Free assessment See how it works

How it works

From first call to a production API in four to six weeks. Precision and recall targets agreed in writing before any build work begins.

1. Discovery sprint ($2,500)

Two-week, fixed-fee discovery. We run a baseline analysis on your actual photos, scope what to detect, and write a SOW with precision and recall targets in your language. You keep these deliverables whether or not the build proceeds.

2. You send what you have

Even a small dataset works. We handle annotation and training, then bridge the gaps with synthetic data for rare defects or edge cases your photos can't cover.

3. V1 in 4–6 weeks

We deliver a working API with a precision and recall report measured against the targets we set during discovery. No further invoices yet.

4. Hit targets, or walk

If V1 hits the agreed targets, the build balance and first month of service license are invoiced and you go live. If V1 misses, you walk with the discovery deliverables and owe nothing further. No termination fee, no clawback.

Want to see it run first?

Try It with your own photo

What's in your packet

A claims-ready bundle, not a research demo. Per inspection, available as a PDF for your dispute file, a JSON payload to your TMS, or an API call.

Annotated evidence

Original photo plus a copy with damage localized in severity-colored boxes, plus class, confidence, and coordinates per detection.

Capture metadata

Container or BOL ID, timestamp, GPS, inspector handle. Pre-filled from capture context, no manual entry.

Delivery

PDF for your dispute file, JSON to your TMS, or email straight to claims. Direct API access for everything.

Real customer case studies coming as our first three projects deliver.

Transparent pricing

One-time build, monthly service license, per-image inference. The build and license are only invoiced once V1 has been accepted against the targets in your SOW.

Build (one-time)

From $5,000 total

$2,500 discovery (weeks 1-2, upfront) + build phase from $2,500 (weeks 3-6, invoiced only on acceptance). Simple single-class detectors with small datasets land near $5,000 total; most multi-class builds land at $10,000. Final build price is scoped in the SOW with your sign-off before any build work starts.

Feasibility analysis on your actual photos
Written SOW with precision and recall targets
Annotation, synthetic data, and model training
Deployed API endpoint and acceptance test

Ongoing (after V1 acceptance)

From $500 / mo + $0.02 / image

Service license: hosted API, drift monitoring, support, quarterly retrain. Simple single-class models start near $500/month; most multi-class deployments land at $1,500/month. Scoped in the SOW alongside the build.

Inference: $0.02 per image up to 50,000/month, $0.01 above. Billed monthly in arrears.

Both start at V1 acceptance, not at kickoff.

Start here

Free feasibility check on your photos.

Send 20 to 50 of your existing photos. I'll send back a one-page report on what's reliably catchable today, what's hard, and what a custom build would unlock. No commitment, no follow-up unless you want one.

Free assessment

Where we're focused

Our current build focus is cargo and loading-dock damage documentation. The same platform supports manufacturing defects, vehicle damage, PPE compliance, and more. See all use cases