Extracting data from life insurance policies
An issued life policy runs from a few pages to dozens, in a layout each carrier designs: a policy data page, premium schedules, riders and the contract provisions. Life settlement providers and lenders taking a collateral assignment each review hundreds a year. Copies arrive as carrier PDFs, scans of the policyholder's original and portal downloads.
Reviewers need the policy number, product, face amount, premium, risk class and the guaranteed premium schedule. A life settlement pricing model loads them to project premiums against the death benefit, and a lender's system records the cover pledged behind a loan.
What makes life insurance policies hard
Terms buried in contract prose
The grace period, the reinstatement window and the suicide exclusion period are each stated once, in running text across several pages of provisions.
Premium tables span pages
Guaranteed premium schedules run one row per policy year, often for decades, and continue across consecutive pages with a separate table for each rider.
Every carrier designs its own contract
Carriers file their own policy forms, so the face amount and risk class sit in a differently arranged data block on each contract.
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 life insurance policy, end to end
About this document: A specimen annual renewable term life policy published by Pacific Life, 14 portrait pages with a text layer. The cover carries the policy data block. Pages 3 to 7 hold the policy specifications and the guaranteed premium tables, and pages 8 to 13 set out the contract provisions. Page 1 is featured. The insured is a placeholder, and every data page is stamped SAMPLE.
Processing objective: The objective is to read the coverage at a glance, as an underwriter or a lender would. That means the kind of policy, its number, the cover, what it costs each month and the rating behind that price. Each value grounds to its own line of the cover's two-column data block.
Demonstration fields: Product, Policy number, Face amount, Initial monthly premium, Risk class
The document
4 of 14 pages
Page 1shown below
Page 5
Page 7
Page 9
Where each value came from
product_type
{
"value": "TERM LIFE INSURANCE",
"page": 1,
"range": {
"end": 882,
"start": 863
},
"box": {
"xmin": 0.11494,
"ymin": 0.60191,
"xmax": 0.41894,
"ymax": 0.62331
}
}policy_number
{
"value": "VF99999990",
"page": 1,
"range": {
"end": 1136,
"start": 1110
},
"box": {
"xmin": 0.11823,
"ymin": 0.77541,
"xmax": 0.37647,
"ymax": 0.78866
}
}face_amount
{
"value": 500000,
"page": 1,
"range": {
"end": 1222,
"start": 1200
},
"box": {
"xmin": 0.55903,
"ymin": 0.77506,
"xmax": 0.82887,
"ymax": 0.78915
}
}initial_modal_premium.amount
{
"value": 36,
"page": 1,
"range": {
"end": 1199,
"start": 1173
},
"box": {
"xmin": 0.55872,
"ymin": 0.7606,
"xmax": 0.81092,
"ymax": 0.77493
}
}insured.risk_class
{
"value": "STANDARD NONSMOKER",
"page": 1,
"range": {
"end": 1109,
"start": 1078
},
"box": {
"xmin": 0.11832,
"ymin": 0.7604,
"xmax": 0.48479,
"ymax": 0.77379
}
}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.
