Extracting data from Form 941 returns
Most US employers file a Form 941 every quarter, on the IRS layout, and a payroll provider, an accountant or the owner fills it in. Lenders and acquisition diligence teams ask each company for several years of quarters, which arrive as payroll portal downloads, scanned signed copies and data room uploads.
Diligence needs the EIN, the quarter, the employee count, total wages, income tax withheld and the total tax after adjustments. A quality of earnings workpaper ties those figures to the payroll register and the general ledger, quarter by quarter, to confirm reported payroll cost.
What makes Form 941 returns hard
Two amounts on each line
Lines 5a to 5d print each taxable wage in one column and the tax at its rate in the next.
Lines added and retired
Revisions add lines when a payroll credit starts and mark them reserved when it ends, so the active lines differ between quarters.
Answers given by boxes
The EIN is entered one digit per box, and the quarter, depositor type and business status are answered by checking boxes.
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 Form 941, end to end
About this document: A three-page specimen Form 941 that Ocrolus, a document AI vendor, publishes in its developer documentation. Part 1's wage and tax lines fill page 1, followed by the deposit schedule and the signature and preparer blocks. The names and identification numbers are placeholder data. Its embedded text is garbled, so the values come from the page image. Page 1 is featured.
Processing objective: The objective is to follow the return from filer to tax owed. The fields run from the employer and quarter to the wages paid and the payroll tax after adjustments. The employer number sits in nine single-digit boxes, and the quarter is one checked box in a panel of four.
Demonstration fields: Employer identification number, Quarter, Wages tips and other compensation (line 2), Total taxes after adjustments (line 10)
The document
3 of 3 pages
Page 1shown below
Page 2
Page 3
Where each value came from
employer.ein
{
"value": "12-3456789",
"page": 1,
"range": {
"end": 442,
"start": 393
},
"box": {
"xmin": 0.06548,
"ymin": 0.08112,
"xmax": 0.63685,
"ymax": 0.10846
}
}quarter
{
"value": 1,
"page": 1,
"range": {
"end": 249,
"start": 217
},
"box": {
"xmin": 0.69116,
"ymin": 0.12062,
"xmax": 0.88445,
"ymax": 0.13727
}
}part_1.wages_tips_other_compensation
{
"value": 10000,
"page": 1,
"range": {
"end": 968,
"start": 957
},
"box": {
"xmin": 0.06825,
"ymin": 0.36915,
"xmax": 0.94698,
"ymax": 0.39299
}
}part_1.total_taxes_after_adjustments
{
"value": 10840,
"page": 1,
"range": {
"end": 2469,
"start": 2458
},
"box": {
"xmin": 0.06501,
"ymin": 0.80104,
"xmax": 0.94652,
"ymax": 0.82408
}
}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.
