Extracting data from certificates of occupancy
A city issues a certificate of occupancy when a building may lawfully be used, and every buildings department prints its own form. A lender will not close on new construction without one, and a landlord files it for each tenancy. Title companies read thousands a year, arriving as scans from municipal records.
A closer needs the address, the certificate number, whether it is temporary or final and the dates it runs between. A title system files it against the property record, and a temporary certificate goes on a diary to be chased before it expires.
What makes certificates of occupancy hard
A scan with no text
A certificate is often filed as an image of paper, so every value has to be read from the picture.
Values packed into ruled cells
Lettered sections run down the form, each a grid of label and value pairs squeezed into narrow boxes.
Use tables run columns together
The permitted use table sets a floor number beside the number of persons allowed on it, close enough that position is what separates them.
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 certificate of occupancy, end to end
About this document: A New York City Department of Buildings temporary certificate of occupancy for a school building in the Bronx. Four portrait pages, scanned with no text layer: the certificate itself, two pages of the permissible use and occupancy table, and a closing page. Page 1 is featured. It names no private individual.
Processing objective: The objective is to read the certificate's identity and the window it covers: which building, which certificate, whether it is temporary and between which dates. Both dates are returned in ISO form, normalized from the month-first dates the certificate prints.
Demonstration fields: Property address, CO number, Certificate type, Effective date, Expiration date
The document
4 of 4 pages
Page 1shown below
Page 2
Page 3
Page 4
Where each value came from
property.address
{
"value": "192 EAST 151ST STREET",
"page": 1,
"range": {
"end": 651,
"start": 621
},
"box": {
"xmin": 0.10356,
"ymin": 0.22887,
"xmax": 0.46013,
"ymax": 0.25006
}
}co_number
{
"value": "220152652T005",
"page": 1,
"range": {
"end": 84,
"start": 60
},
"box": {
"xmin": 0.55484,
"ymin": 0.12076,
"xmax": 0.8768,
"ymax": 0.13901
}
}certificate_type
{
"value": "Temporary",
"page": 1,
"range": {
"end": 602,
"start": 575
},
"box": {
"xmin": 0.66601,
"ymin": 0.2079,
"xmax": 0.89152,
"ymax": 0.22889
}
}effective_date
Printed on the page as Effective Date: 04/16/2019, normalized on extraction.
{
"value": "2019-04-16",
"page": 1,
"range": {
"end": 711,
"start": 685
},
"box": {
"xmin": 0.666,
"ymin": 0.22889,
"xmax": 0.8915,
"ymax": 0.25011
}
}expiration_date
Printed on the page as Expiration Date: 07/15/2019, normalized on extraction.
{
"value": "2019-07-15",
"page": 1,
"range": {
"end": 862,
"start": 835
},
"box": {
"xmin": 0.66596,
"ymin": 0.25008,
"xmax": 0.89147,
"ymax": 0.32101
}
}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.
