Extracting data from Form 1065 returns
Partnerships file more than four million Form 1065 returns a year, each printed by its preparer's tax software on that year's IRS layout. Mortgage lenders collect two years of them from every borrower who owns 25% or more of a partnership, uploaded to a borrower portal as PDFs and scans.
An underwriter needs the partnership's EIN and tax year, its ordinary business income, and the depreciation and depletion it deducted. A cash flow analysis such as Fannie Mae Form 1084 adds those non-cash deductions back to arrive at the income a borrower qualifies on.
What makes Form 1065 returns hard
Lines change between tax years
The IRS revises the form each year, and line 20 has carried both other deductions and the energy efficient commercial buildings deduction.
Checkbox rows in the header
Rows G to K set return type, accounting method and partnership elections with checkboxes, several to a printed line.
Page 1 opens a long filing
The balance sheet on Schedule L and a Schedule K-1 for every partner follow page 1, all in one filing.
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 1065, end to end
About this document: Page 1 of a Form 1065, filled with placeholder data and published by a document AI vendor as its sample. One portrait page: the partnership's header, checkbox rows G to K, 32 numbered amount lines, and the signature and paid preparer blocks. Page 1 is featured. The data is invented and shown unredacted.
Processing objective: The objective is to capture who filed the return and where it ended up, along with the state of one checkbox. The checkbox comes back as a boolean, and the ordinary business loss comes back as a negative number read from the printed minus sign.
Demonstration fields: Partnership name, EIN, Schedules C and M-3 attached (row J), Total income (line 8), Ordinary business income (line 23)
The document
1 page
Page 1shown below
Where each value came from
partnership.name
{
"value": "Sample Software Service",
"page": 1,
"range": {
"end": 494,
"start": 448
},
"box": {
"xmin": 0.24714,
"ymin": 0.10588,
"xmax": 0.8003,
"ymax": 0.13611
}
}partnership.ein
{
"value": "12-3456789",
"page": 1,
"range": {
"end": 804,
"start": 761
},
"box": {
"xmin": 0.80123,
"ymin": 0.10609,
"xmax": 0.94312,
"ymax": 0.13403
}
}schedules_c_and_m3_attached
{
"value": true,
"page": 1,
"range": {
"end": 1270,
"start": 1218
},
"box": {
"xmin": 0.05646,
"ymin": 0.25953,
"xmax": 0.93916,
"ymax": 0.27264
}
}income.total_income
{
"value": 15000,
"page": 1,
"range": {
"end": 2653,
"start": 2644
},
"box": {
"xmin": 0.82376,
"ymin": 0.4092,
"xmax": 0.94309,
"ymax": 0.42416
}
}deductions.ordinary_business_income
{
"value": -76000,
"page": 1,
"range": {
"end": 4561,
"start": 4551
},
"box": {
"xmin": 0.82365,
"ymin": 0.65121,
"xmax": 0.94319,
"ymax": 0.66649
}
}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.
