Extracting data from commercial appraisals
Certified general appraisers write commercial appraisals as narrative reports, each firm with its own template and section order. A lender's appraisal review team receives one for most commercial mortgages, and auditors see them for fair value reporting. Each arrives as a PDF that can run past a hundred pages.
Underwriting needs the market value conclusion, the value from each approach, the capitalization rate and the net operating income. The commercial loan origination system sizes the loan from the value and the income, testing loan-to-value and debt service coverage against credit policy.
What makes commercial real estate appraisals hard
The same table, printed again
A narrative report prints its value conclusions in the transmittal letter, again in the salient facts summary and again in the reconciliation.
A template for every firm
Each firm uses its own word-processed template, setting the key facts in prose paragraphs and two-column summary tables.
Over a hundred pages
A full report can run past a hundred pages, and the figures underwriting needs sit on a handful of them among market analysis and addenda.
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 commercial real estate appraisal, end to end
About this document: A narrative appraisal by Integra Realty Resources, filed with the SEC as an exhibit. It values a single-tenant office building in Dallas leased to the federal General Services Administration. Eight portrait pages from a longer report: the cover, the transmittal letter, the summary of salient facts, the subject identification and the reconciliation. Page 4 is featured. The appraisers' signatures were removed, and the addressee's street line replaced.
Processing objective: The objective is to read the summary an underwriter checks first: the parcel, its size, occupancy and age, and the concluded fair value. An auditor ties that value to the property's carrying amount on the balance sheet. Each value grounds to its own cell of the Summary of Salient Facts.
Demonstration fields: Tax ID, Gross building area, Percent leased, Year built, Fair value as is
The document
4 of 8 pages
Page 1
Page 3
Page 4shown below
Page 5
Where each value came from
property.tax_id
{
"value": "001616000901A0000",
"page": 4,
"range": {
"end": 4337,
"start": 4320
},
"box": {
"xmin": 0.37961,
"ymin": 0.2192,
"xmax": 0.87546,
"ymax": 0.23788
}
}property.gross_building_area_sf
{
"value": 39329,
"page": 4,
"range": {
"end": 4451,
"start": 4442
},
"box": {
"xmin": 0.37968,
"ymin": 0.25394,
"xmax": 0.87547,
"ymax": 0.27032
}
}property.percent_leased
{
"value": 100,
"page": 4,
"range": {
"end": 4547,
"start": 4543
},
"box": {
"xmin": 0.37975,
"ymin": 0.28649,
"xmax": 0.87548,
"ymax": 0.30255
}
}property.year_built
{
"value": 1985,
"page": 4,
"range": {
"end": 4589,
"start": 4585
},
"box": {
"xmin": 0.37978,
"ymin": 0.30238,
"xmax": 0.87549,
"ymax": 0.32134
}
}value_conclusions[1].value_conclusion
{
"value": 4600000,
"page": 4,
"range": {
"end": 5236,
"start": 5226
},
"box": {
"xmin": 0.67899,
"ymin": 0.47547,
"xmax": 0.87582,
"ymax": 0.49651
}
}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.
