Extracting data from driver's licenses
Every US state and territory designs its own license card, and redesigns it every few years, so dozens of layouts circulate at once. A bank opening an account and a hospital registering a patient both take one as proof of identity. Onboarding teams verify thousands a month, arriving as phone photographs through a mobile app.
A verifier needs the license number, the issuing state, the expiry date and the date of birth. A KYC platform matches those against the application, screens the holder against watchlists, then scores the applicant for identity risk before an account opens.
What makes driver's licenses hard
Every state designs its own card
A license redesigned in one state moves where every field sits, and older cards stay valid for years alongside the new design.
Values sit behind numbered codes
Each value follows a small code from the AAMVA standard, so a card carries around twenty labels and twenty values at once.
Security art crowds the fields
A guilloche background, a ghost portrait and microtext sit among the printed values, on a card the size of a credit card.
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 driver's license, end to end
About this document: The official specimen of a Texas driver license design, published by the state's Department of Public Safety in a news release. One landscape card image, the front only. Around twenty fields sit over a guilloche background with a star, a seal and a ghost portrait. Every value is specimen data, so no redaction was needed.
Processing objective: The objective is to read a card's identity and its holder's physical description from one image. Each value sits behind a small numbered code, so extraction has to return the value and leave the code. Height comes back in words, normalized from the form the card prints.
Demonstration fields: License number, Issuing state, Sex, Height, Eye color
The document
1 page
Page 1shown below
Where each value came from
document_info.document_number
{
"value": "12345678",
"page": 1,
"range": {
"end": 56,
"start": 42
},
"box": {
"xmin": 0.30245,
"ymin": 0.26016,
"xmax": 0.48774,
"ymax": 0.31276
}
}document_info.state_issuing
{
"value": "Texas",
"page": 1,
"range": {
"end": 5,
"start": 0
},
"box": {
"xmin": 0.02769,
"ymin": 0.06977,
"xmax": 0.35308,
"ymax": 0.2016
}
}personal_info.sex
{
"value": "F",
"page": 1,
"range": {
"end": 214,
"start": 206
},
"box": {
"xmin": 0.30386,
"ymin": 0.60194,
"xmax": 0.38034,
"ymax": 0.63992
}
}personal_info.height
Printed on the page as 16 Hgt 5'-10", normalized on extraction.
{
"value": "5 feet 10 inches",
"page": 1,
"range": {
"end": 228,
"start": 215
},
"box": {
"xmin": 0.3042,
"ymin": 0.64085,
"xmax": 0.42424,
"ymax": 0.682
}
}personal_info.eye_color
{
"value": "BRO",
"page": 1,
"range": {
"end": 240,
"start": 229
},
"box": {
"xmin": 0.30434,
"ymin": 0.68076,
"xmax": 0.40927,
"ymax": 0.72153
}
}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.
