Extracting data from pension statements
Public retirement systems and corporate plan administrators send retirees a pension statement with every payment, each in the system's own layout. Mortgage lenders may request the latest statement when applicants use pension income to qualify, as a portal download or a scan of the mailed advice.
Underwriting needs the gross monthly annuity, the tax withholding, the other deductions and the net payment. The loan origination system counts the gross annuity as qualifying income, once documents show it continuing for three more years.
What makes pension statements hard
Current beside year to date
Each amount prints for the current period and again as a year-to-date total, so every figure has a twin in the next column.
Deductions in several tables
Taxes, before-tax deductions and after-tax deductions each sit in their own small table, and a summary band totals them again.
Each system's own layout
State systems, municipal plans and corporate administrators each design their own payment advice, with different labels for the same amounts.
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 pension statement, end to end
About this document: The specimen pension pay stub that the New York State and Local Retirement System shows its retirees in a tutorial on reading their payment advice. One landscape page with no text layer: a header, earnings and tax tables, deduction tables, a summary band and the net pay distribution. Page 1 is featured. Every personal value is a placeholder.
Processing objective: The objective is to follow one monthly payment from gross to net. The service annuity is the gross amount, withholding and other deductions come off it, and what remains is the net pay deposited. Each figure grounds to its own cell.
Demonstration fields: Monthly service annuity, Federal withholding, Total deductions, Net pay
The document
1 page
Page 1shown below
Where each value came from
earnings[0].current
{
"value": 2148.58,
"page": 1,
"range": {
"end": 808,
"start": 800
},
"box": {
"xmin": 0.19324,
"ymin": 0.37344,
"xmax": 0.38231,
"ymax": 0.48169
}
}taxes[0].current
{
"value": 257.82,
"page": 1,
"range": {
"end": 877,
"start": 871
},
"box": {
"xmin": 0.69224,
"ymin": 0.37342,
"xmax": 0.82902,
"ymax": 0.48161
}
}summary.total_deductions_current
{
"value": 802.5,
"page": 1,
"range": {
"end": 1807,
"start": 1801
},
"box": {
"xmin": 0.69487,
"ymin": 0.71663,
"xmax": 0.82938,
"ymax": 0.74401
}
}summary.net_pay_current
{
"value": 1088.26,
"page": 1,
"range": {
"end": 1824,
"start": 1816
},
"box": {
"xmin": 0.82934,
"ymin": 0.71663,
"xmax": 0.96184,
"ymax": 0.74399
}
}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.
