Extracting data from closing disclosures
A lender issues a Closing Disclosure for every consumer mortgage, on the five-page TRID form set by Regulation Z, with its own fees in each section. Post-closing quality control, loan buyers and servicers each review one per loan, as a PDF from the closing package or an e-disclosure portal.
Review needs the loan amount, the interest rate, the monthly payment, the closing costs and the cash to close. A TRID compliance check compares those with the Loan Estimate, and servicing boards the payment and escrow figures.
What makes closing disclosures hard
Fourteen lettered cost sections
Closing cost details run in sections A to N, with borrower-paid and seller-paid columns, each split between at and before closing.
The same totals on three pages
Closing costs and cash to close print on page 1 and again in the itemized pages behind it, each copy inside a different table.
Payments that change by period
Projected payments split into columns by year range, so mortgage insurance and the total monthly payment can differ between columns.
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 closing disclosure, end to end
About this document: The Consumer Financial Protection Bureau's filled sample Closing Disclosure for a purchase loan, published with its explainer for the form. Five portrait pages: loan terms and costs at closing, itemized closing costs, the cash-to-close calculation, loan disclosures and loan calculations. Page 1 is featured. Every party and address is a specimen placeholder.
Processing objective: The objective is to read what the loan costs, monthly and at closing, the way a borrower reads page 1. The rate produces the principal and interest payment, and the all-in payment adds insurance and escrow. The closing totals show what is due on the day.
Demonstration fields: Interest rate, Monthly principal and interest, Estimated total monthly payment (years 1-7), Closing costs, Cash to close
The document
4 of 5 pages
Page 1shown below
Page 2
Page 3
Page 4
Where each value came from
loan_terms.interest_rate_percent
{
"value": 3.875,
"page": 1,
"range": {
"end": 914,
"start": 908
},
"box": {
"xmin": 0.2941,
"ymin": 0.31223,
"xmax": 0.4709,
"ymax": 0.3407
}
}loan_terms.monthly_principal_and_interest
{
"value": 761.78,
"page": 1,
"range": {
"end": 1068,
"start": 1061
},
"box": {
"xmin": 0.29402,
"ymin": 0.34069,
"xmax": 0.47099,
"ymax": 0.40133
}
}projected_payments[0].estimated_total_monthly_payment
{
"value": 1050.26,
"page": 1,
"range": {
"end": 1792,
"start": 1783
},
"box": {
"xmin": 0.29441,
"ymin": 0.66738,
"xmax": 0.61808,
"ymax": 0.70581
}
}costs_at_closing.closing_costs
{
"value": 9712.1,
"page": 1,
"range": {
"end": 2569,
"start": 2452
},
"box": {
"xmin": 0.29455,
"ymin": 0.85306,
"xmax": 0.94217,
"ymax": 0.89858
}
}costs_at_closing.cash_to_close
{
"value": 14147.26,
"page": 1,
"range": {
"end": 2697,
"start": 2610
},
"box": {
"xmin": 0.29444,
"ymin": 0.89857,
"xmax": 0.94237,
"ymax": 0.944
}
}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.
