Extracting data from vehicle registrations
State motor vehicle agencies print registration cards from their own systems, and every state uses a different card. Auto lenders and insurers collect one for each vehicle they finance or cover, and a fleet manager tracks hundreds at once. Most arrive as phone photos or scans uploaded to a portal.
Servicing needs the plate number, the title number, the vehicle's year and make, the expiration date and the fees paid. A loan servicing or policy administration system matches the vehicle to its collateral or policy record, and flags a registration due to lapse.
What makes vehicle registrations hard
A borderless grid of values
Labels and values sit in loose columns across the card, and the table has to be rebuilt from their positions alone.
Every state prints its own card
Each state agency prints a different registration, with its own field order, label abbreviations and card size, and redesigns it over time.
Small cards, photographed
Registrations reach lenders and insurers as phone photos and low-resolution scans of a pocket-sized card, often creased or shot at an angle.
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 vehicle registration, end to end
About this document: A Florida vehicle registration printed as a system test by the state's Department of Highway Safety and Motor Vehicles, then republished in a city parking permit handout. One landscape page, an image-only scan at low resolution. It carries the plate, decal, vehicle, title and an itemized fee block. The card was cropped from the handout, and one license number was replaced with a stand-in.
Processing objective: The objective is to identify the vehicle and confirm what was paid. The plate, title number and model year tie the card to one vehicle, and the registration tax plus a 3.00 county fee gives the 71.15 grand total. Each value grounds to the part of the card that prints it.
Demonstration fields: Plate number, Title number, Model year, Registration tax, Grand total
The document
1 page
Page 1shown below
Where each value came from
registration.plate_number
{
"value": "429MAI",
"page": 1,
"range": {
"end": 123,
"start": 117
},
"box": {
"xmin": 0.07987,
"ymin": 0.16255,
"xmax": 0.19086,
"ymax": 0.21669
}
}registration.title_number
{
"value": "100020689",
"page": 1,
"range": {
"end": 412,
"start": 396
},
"box": {
"xmin": 0.43737,
"ymin": 0.21667,
"xmax": 0.55501,
"ymax": 0.33546
}
}vehicle.year
{
"value": 2005,
"page": 1,
"range": {
"end": 309,
"start": 278
},
"box": {
"xmin": 0.07986,
"ymin": 0.21668,
"xmax": 0.19075,
"ymax": 0.33523
}
}fees.registration_tax
{
"value": 68.15,
"page": 1,
"range": {
"end": 483,
"start": 466
},
"box": {
"xmin": 0.70308,
"ymin": 0.21666,
"xmax": 0.79539,
"ymax": 0.33562
}
}fees.grand_total
{
"value": 71.15,
"page": 1,
"range": {
"end": 1063,
"start": 1058
},
"box": {
"xmin": 0.70314,
"ymin": 0.43164,
"xmax": 0.79549,
"ymax": 0.47143
}
}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. VPC and on-prem deployment require an Enterprise plan.
