Extracting data from bills of lading
Ocean carriers and freight forwarders issue bills of lading on their own forms, each with numbered boxes for parties, vessel, ports and cargo. Importers, customs brokers and forwarders handle them for container shipments, and banks see them in letter of credit transactions. They arrive as scans, emailed PDFs or exports from a carrier's portal.
Operations needs the bill of lading number, the vessel and voyage, the ports of loading and discharge, and each container number. Those values feed a transport management system, which matches the shipment to its invoice and tracks the container to delivery.
What makes bills of lading hard
Captions share the value's cell
Each numbered box prints its caption and its typed value in one cell, so Port of Loading sits directly above the port it names.
A layout for every carrier
Every ocean carrier and forwarder prints its own bill, with the same parties, ports and cargo placed in different boxes.
Cargo runs onto riders
A long cargo description continues on a rider page, which repeats the bill's header and often carries the shipment totals.
Parse, extract, verify
Parse
Parse turns the document into reading-order Markdown and a hierarchy of blocks (text, tables, figures), and grounds each block to its exact location on the page.
Extract
You define the fields in a JSON schema, and Extract returns their values in a consistent, structured format, each linked back to where it appears in the Markdown.
Verify
Each value links back to where it appears, down to the character range in the Markdown, and the page and bounding box in the source document, so every result can be verified.
One bill of lading, end to end
About this document: A COSCO Container Lines bill of lading for one 40-foot container shipped from Beihai, China to New York, filed as an exhibit in a US court case. Two portrait pages, scanned on an office copier: the bill's face and a continuation rider that carries the totals. Page 1 is featured. A court reporter's exhibit sticker was erased from an empty box.
Processing objective: The objective is to identify the shipment: which bill, which vessel, from which port to which, and which container. A consignee matches those values against the arrival notice before the cargo is released. Each value grounds to its own numbered box on the face.
Demonstration fields: B/L number, Vessel, Port of loading, Port of discharge, Container number
The document
2 of 2 pages
Page 1shown below
Page 2
Where each value came from
bill_of_lading_number
{
"value": "COSU6030603280",
"page": 1,
"range": {
"end": 263,
"start": 227
},
"box": {
"xmin": 0.75441,
"ymin": 0.08072,
"xmax": 0.93223,
"ymax": 0.10249
}
}vessel
{
"value": "HUI JIN QIAO 181",
"page": 1,
"range": {
"end": 1252,
"start": 1206
},
"box": {
"xmin": 0.10156,
"ymin": 0.35112,
"xmax": 0.3624,
"ymax": 0.37685
}
}port_of_loading.city
{
"value": "BEIHAI, CHINA",
"page": 1,
"range": {
"end": 1293,
"start": 1261
},
"box": {
"xmin": 0.36239,
"ymin": 0.35135,
"xmax": 0.56632,
"ymax": 0.37705
}
}port_of_discharge.city
{
"value": "NEW YORK, NY",
"page": 1,
"range": {
"end": 1406,
"start": 1373
},
"box": {
"xmin": 0.10163,
"ymin": 0.37659,
"xmax": 0.36241,
"ymax": 0.40445
}
}containers[0].container_number
{
"value": "CBHU1842324",
"page": 1,
"range": {
"end": 1780,
"start": 1761
},
"box": {
"xmin": 0.10003,
"ymin": 0.42138,
"xmax": 0.27707,
"ymax": 0.44481
}
}How to try it yourself
Everything on this page is reproducible. The sample, its schema, and the Parse and Extract API outputs are published. You can repeat the same two calls from whichever surface you already build on.
- Claude Skilldescribe the job; the agent writes it
- CLIno code at all
- MCPexpose ADE docs as an agent tool
- REST APIany language
- Pythonlandingai-ade
- TypeScriptlandingai-ade
Documents into data
Accurate
The best accuracy on the difficult documents other tools get wrong, such as messy scans, complicated tables, checkboxes and handwriting.
Auditable
Every answer cites its sources. Values come back with an audit trail (page, bounding box, character range) so a human can review them.
Agentic
Applies multi-step reasoning, and plans before it acts. Handles documents it has never seen before and adapts to variations.
API-first
Composable APIs that drop into any tech stack and can be called from anywhere. Easy to build on and embed in your pipelines and products.
Affordable
Cost-competitive rates with complete price transparency. Spend controls before you run, and every job reports its exact cost.
Enterprise-grade
Built for production and scale from day one. Fully observable and monitored, running securely in our cloud or yours.
Built for high-volume, high-throughput production
- Zero data retention
- Priority rate limits
- Enhanced support
- VPC and on-prem deployment
Available on Team and Enterprise plans.
