Extracting data from employee censuses
Employers export a census from payroll or HR software, one row per employee, each system with its own columns. Benefits brokers and PEOs request one from every client group at each renewal, and acquirers request one during diligence. It arrives by email or data room, as a spreadsheet or a PDF report.
Underwriting needs each employee's birth date, gender, home state, hire date, status and annual pay. A carrier's rating engine prices the group from those, and a deal model uses the same rows to project payroll and benefits cost after close.
What makes employee censuses hard
Dozens of near-identical rows
A census runs one row per employee across a dozen or more columns, and each date and amount has to stay with its own employee.
Blank columns invite misreads
Termination date is blank on every active row. Any vendor solution reading by position can slide a neighbor's value into that gap.
Every system exports its own layout
Payroll and HR platforms each choose their own census columns: one prints age and birth date, another a coverage tier and home ZIP.
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 employee census, end to end
About this document: An Employee Census Report from AdvanStaff HR, a payroll provider, run against the demo company in its sales material. One landscape page with a native text layer, listing 34 active employees in 12 columns that run from employee ID to annual pay. Every employee is a fictional demo record. Page 1 is featured.
Processing objective: The objective is to lift one complete employee row and keep every value in its own column, as an underwriter would read it. The row identifies the employee by ID alone. It was chosen because its original hire date differs from its last hire date, a rehire.
Demonstration fields: Company, Employee ID, Birth date, Original hire date, Annual pay
The document
1 page
Page 1shown below
Where each value came from
company_name
{
"value": "ASI DEMO - GENERAL COMPANY",
"page": 1,
"range": {
"end": 115,
"start": 88
},
"box": {
"xmin": 0.44359,
"ymin": 0.06092,
"xmax": 0.63499,
"ymax": 0.0779
}
}employees[0].employee_id
{
"value": "E92815",
"page": 1,
"range": {
"end": 455,
"start": 449
},
"box": {
"xmin": 0.17458,
"ymin": 0.18725,
"xmax": 0.25208,
"ymax": 0.20582
}
}employees[0].birth_date
Printed on the page as 07/30/1968, normalized on extraction.
{
"value": "1968-07-30",
"page": 1,
"range": {
"end": 489,
"start": 479
},
"box": {
"xmin": 0.31481,
"ymin": 0.18728,
"xmax": 0.46571,
"ymax": 0.20589
}
}employees[0].original_hire_date
Printed on the page as 02/26/2024, normalized on extraction.
{
"value": "2024-02-26",
"page": 1,
"range": {
"end": 574,
"start": 564
},
"box": {
"xmin": 0.72971,
"ymin": 0.18739,
"xmax": 0.81386,
"ymax": 0.206
}
}employees[0].annual_pay
{
"value": 41601.6,
"page": 1,
"range": {
"end": 601,
"start": 592
},
"box": {
"xmin": 0.88094,
"ymin": 0.18743,
"xmax": 0.97574,
"ymax": 0.20605
}
}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.
