Extracting data from closing protection letters
A title insurer issues a closing protection letter to the lender on a real estate closing, on its own version of the ALTA form. A lender's closing department receives one for every loan it closes, usually as a PDF emailed by the settlement agent or issued through the underwriter's portal.
The closing team needs the lender named, the settlement agent, the file and loan numbers and the coverage limit. Funding checks those against the loan file in the loan origination system, and releases funds once the letter matches.
What makes closing protection letters hard
A few values in legal text
The transaction details fill a short header block, followed by numbered requirements, conditions and exclusions that run to a second page.
Two parties in the header
The letterhead block names the protected lender and the settlement agent side by side, each with an address, and both roles have to stay apart.
Letters from cancelled agents
Underwriters have received claims on letters from agents they had cancelled, so many lenders check each letter on the underwriter's portal.
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 protection letter, end to end
About this document: A sample closing protection letter from Covenant Real Estate Services, a Virginia settlement agent, issued on Fidelity National Title's form to a specimen lender. Two portrait pages: the transaction block and the requirements on page 1, then the conditions, exclusions and signature on page 2. Page 1 is featured. The borrowers and property are specimen data, and the page carries a SAMPLE watermark.
Processing objective: The objective is to capture who the letter protects, against whose handling of which transaction, and up to what amount. The coverage limit is read from inside a numbered requirement, and comes back as a number while grounding points at the sentence that sets it.
Demonstration fields: Lender protected, Settlement agent, File number, Loan number, Maximum funds covered
The document
2 of 2 pages
Page 1shown below
Page 2
Where each value came from
addressee.name
{
"value": "Lending Bank, LLC",
"page": 1,
"range": {
"end": 72,
"start": 55
},
"box": {
"xmin": 0.029,
"ymin": 0.07136,
"xmax": 0.13677,
"ymax": 0.08323
}
}settlement_agent.name
{
"value": "Covenant Real Estate Services LLC",
"page": 1,
"range": {
"end": 409,
"start": 375
},
"box": {
"xmin": 0.58782,
"ymin": 0.11785,
"xmax": 0.78576,
"ymax": 0.12834
}
}file_number
{
"value": "CRES-23-003S",
"page": 1,
"range": {
"end": 194,
"start": 168
},
"box": {
"xmin": 0.02935,
"ymin": 0.12894,
"xmax": 0.19613,
"ymax": 0.13948
}
}loan_number
{
"value": "2003001",
"page": 1,
"range": {
"end": 315,
"start": 294
},
"box": {
"xmin": 0.02925,
"ymin": 0.17361,
"xmax": 0.16558,
"ymax": 0.18576
}
}max_aggregate_funds
{
"value": 5000000,
"page": 1,
"range": {
"end": 1459,
"start": 1304
},
"box": {
"xmin": 0.02726,
"ymin": 0.37911,
"xmax": 0.8992,
"ymax": 0.39068
}
}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.
