Extracting data from CP 575 notices
The IRS sends a CP 575 notice once, when it assigns a business its Employer Identification Number, in versions that follow the entity's filing requirements. Banks, payment processors and health plans collect a copy from each business they onboard, as a scan or PDF uploaded with the application.
Onboarding needs the EIN, the legal name exactly as the IRS holds it, the address and the forms the business must file. Business verification checks match the name and EIN against the business's W-9 before the account goes live.
What makes CP 575 notices hard
Filings stated in a sentence
The forms a business must file sit inside the letter's prose, after the line telling it to file the forms by the dates shown.
Copies arrive marked up
Organizations pass along annotated copies, with highlights and callout boxes added over the notice by whoever handled it before.
A letter issued once
The IRS issues a CP 575 only once, so many copies in circulation are scans of an original kept for years in a company's files.
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 IRS CP 575, end to end
About this document: A specimen CP 575 A notice published by a health plan on its provider forms page, to show providers which IRS letter to submit. Three portrait pages: the letter, reminders and a tear-off stub. The publisher added a red example header, callout boxes and highlights. Page 1 is featured. Every value is placeholder data.
Processing objective: The objective is to read the assignment in one page. It names the notice and the application it answers, which business received which number, and what that business must now file. Each value grounds to its own line of the letter's header or body.
Demonstration fields: Employer Identification Number, Legal name, Notice number, Application form, Required filing
The document
3 of 3 pages
Page 1shown below
Page 2
Page 3
Where each value came from
ein
{
"value": "88-8888888",
"page": 1,
"range": {
"end": 201,
"start": 158
},
"box": {
"xmin": 0.61161,
"ymin": 0.14067,
"xmax": 0.91523,
"ymax": 0.17036
}
}entity.legal_name
{
"value": "ABC Behavioral LCSW PC",
"page": 1,
"range": {
"end": 428,
"start": 405
},
"box": {
"xmin": 0.17538,
"ymin": 0.21728,
"xmax": 0.38484,
"ymax": 0.2313
}
}notice.notice_number
{
"value": "CP 575 A",
"page": 1,
"range": {
"end": 246,
"start": 214
},
"box": {
"xmin": 0.61581,
"ymin": 0.20746,
"xmax": 0.91757,
"ymax": 0.22157
}
}notice.form
{
"value": "SS-4",
"page": 1,
"range": {
"end": 213,
"start": 202
},
"box": {
"xmin": 0.61655,
"ymin": 0.18206,
"xmax": 0.72552,
"ymax": 0.1959
}
}required_filings[0].form
{
"value": "Form 1120",
"page": 1,
"range": {
"end": 1686,
"start": 1677
},
"box": {
"xmin": 0.30443,
"ymin": 0.66193,
"xmax": 0.38882,
"ymax": 0.67244
}
}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.
