Extracting data from auto insurance applications
Each auto insurer prints its own application and files a separate edition for every state it writes in. A carrier's underwriting team receives one with every new policy its agents and direct channels write. They arrive through agency portals and email, as signed PDFs or scanned paper.
Underwriting needs each listed driver and their history, each vehicle and its VIN, and the limits and premium for every coverage. The underwriting system checks those answers against motor vehicle and CLUE reports, and policy administration issues the declarations page from them.
What makes auto insurance applications hard
Each carrier and state differs
Each carrier files its own form state by state. Massachusetts forms list coverages as Parts 1 to 12, numbered by the state's standard policy.
Drivers and vehicles repeat
Each household driver and each auto gets its own block of answers, and a family policy repeats those blocks across several pages.
Limits print as split pairs
A liability limit prints as two figures in one cell, $25,000 each person and $50,000 each accident, and appears elsewhere as 25/50.
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 auto insurance application, end to end
About this document: A completed Progressive application for Massachusetts motor vehicle insurance, written through an independent agency, six portrait pages long. Pages 1 and 2 carry the premium summary and the household's drivers. Page 3, the outline of coverage, is featured, and the driving history and a signed agreement follow it. Every personal and agency detail, including the VIN and policy number, was replaced with a stand-in before publication.
Processing objective: The objective is to tie the insured vehicle to one complete coverage line and to the total premium. Personal Injury Protection is the only line on the page with a deductible, so it fills every column of the coverage table. The five coverage premiums add up to the 12-month total printed beneath them.
Demonstration fields: VIN, Coverage, Deductible, Coverage premium, Total 12-month premium
The document
4 of 6 pages
Page 1
Page 2
Page 3shown below
Page 5
Where each value came from
vehicles[0].vin
{
"value": "3FA6P0HD2FR401300",
"page": 3,
"range": {
"end": 5320,
"start": 5298
},
"box": {
"xmin": 0.1741,
"ymin": 0.15197,
"xmax": 0.34674,
"ymax": 0.16558
}
}coverages[1].coverage
{
"value": "Personal Injury Protection",
"page": 3,
"range": {
"end": 5797,
"start": 5762
},
"box": {
"xmin": 0.17336,
"ymin": 0.28164,
"xmax": 0.4715,
"ymax": 0.29645
}
}coverages[1].deductible
{
"value": 8000,
"page": 3,
"range": {
"end": 5839,
"start": 5833
},
"box": {
"xmin": 0.73532,
"ymin": 0.28166,
"xmax": 0.83615,
"ymax": 0.29644
}
}coverages[1].premium
{
"value": 129,
"page": 3,
"range": {
"end": 5851,
"start": 5848
},
"box": {
"xmin": 0.83607,
"ymin": 0.28166,
"xmax": 0.92468,
"ymax": 0.29644
}
}total_12_month_premium
{
"value": 3559,
"page": 3,
"range": {
"end": 6624,
"start": 6615
},
"box": {
"xmin": 0.83429,
"ymin": 0.41233,
"xmax": 0.92437,
"ymax": 0.43062
}
}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.
