The customs invoice you should have been getting all along.

Drop a PDF or photo. Multi-page documents are fine -- invoice, packing list, bill of lading, shipping bill, all in one file. The reader handles each section in turn.

We extract every line: HS code as printed, quantity, unit value, weight, origin country, freight, insurance, Incoterm, buyer, seller, ports, vessel, container number.

Anything we are not 100% certain about is flagged for you to verify before submit. The form will not silently smooth over an uncertain field.

US imports run against USITC HTS; the UK runs against the HMRC Trade Tariff; the EU against TARIC; everywhere else uses the destination's published schedule, falling back to the World Bank WITS bound rate when needed.

The classifier applies the WCO General Rules of Interpretation against the line description, cross-references CBP CROSS rulings for similar products, and proposes a 6 or 10 digit code with the reasoning attached. When confidence is below 85% we run three independent passes and either reach agreement or surface the disagreement.

Every alternative code we considered is recorded, so when you disagree, you can see exactly what was on the shortlist and why.

Every layer of the landed cost is named: MFN duty, Section 232, Section 301, Section 122, AD/CVD, VAT or GST or IGST, MPF, HMF, AFRMM, IPI, ICMS, country-specific levies.

Each line links to the primary source row that produced the rate and shows when that source was last verified. If a source has not refreshed in 14 days we say so, plainly, with an amber chip on the row.

The total comes with a range, not a single point. A typical +/- 5% uncertainty so you have a budget you can plan against. There is a PDF, there is a JSON receipt, and the math is identical to what a customs broker computes by hand.