Extracting data from certified payroll reports
Contractors on prevailing wage public works file a certified payroll each week, on the federal WH-347, a state form or their payroll software's own layout. Prime contractors and awarding agencies review one from every contractor and subcontractor on the job, every week. They arrive as emailed PDFs and compliance portal uploads.
Compliance review needs each worker's classification, hours, hourly rate, fringe benefits and gross wages. A labor compliance system compares each rate against the prevailing wage determination for that classification, and flags any shortfall before the agency releases the contractor's next payment.
What makes certified payroll reports hard
Each worker spans stacked rows
Every worker gets a block of overtime and straight-time rows, with one set of classification and pay cells spanning the whole block.
Gross cells split diagonally
Many payroll forms split the gross cell diagonally, with earnings on this project above the rule and earnings from all work below.
A form for every jurisdiction
Contractors file on the federal WH-347, a state agency's form or their payroll software's own printout, each arranging the columns differently.
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 certified payroll report, end to end
About this document: A filled specimen of Oregon's WH-38 certified payroll form, published by the state's Bureau of Labor and Industries in its prevailing wage training. Two landscape pages: the payroll grid of hours and wages, then the certified statement on its reverse. Page 1 is featured. The workers are fictional, and the contractor's name and contact details were replaced with stand-ins.
Processing objective: The objective is to tie one worker's week to the project and payroll it was reported under, as an agency reviewer does. The worker's gross for all work, $1,163.75, sits in the lower half of a diagonally split cell, beneath the gross earned on this project.
Demonstration fields: Project name, Payroll number, Straight-time hours, Gross wages (all work), Net wages paid
The document
2 of 2 pages
Page 1shown below
Page 2
Where each value came from
project.name
{
"value": "Rock Creek Middle School",
"page": 1,
"range": {
"end": 493,
"start": 383
},
"box": {
"xmin": 0.01948,
"ymin": 0.12078,
"xmax": 0.98022,
"ymax": 0.1537
}
}payroll_number
{
"value": "3",
"page": 1,
"range": {
"end": 190,
"start": 178
},
"box": {
"xmin": 0.47716,
"ymin": 0.08339,
"xmax": 0.55536,
"ymax": 0.09789
}
}workers[2].total_straight_time_hours
{
"value": 3,
"page": 1,
"range": {
"end": 5353,
"start": 5352
},
"box": {
"xmin": 0.43734,
"ymin": 0.66587,
"xmax": 0.47605,
"ymax": 0.70352
}
}workers[2].gross_wages_all_work
{
"value": 1163.75,
"page": 1,
"range": {
"end": 4998,
"start": 4973
},
"box": {
"xmin": 0.58349,
"ymin": 0.62818,
"xmax": 0.64845,
"ymax": 0.72237
}
}workers[2].net_wages_paid
{
"value": 872.81,
"page": 1,
"range": {
"end": 5054,
"start": 5047
},
"box": {
"xmin": 0.71159,
"ymin": 0.62818,
"xmax": 0.78084,
"ymax": 0.72237
}
}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.
