Extracting data from retail installment contracts
A car dealer signs a retail installment sale contract for every financed sale, usually on a forms vendor's state edition, then assigns it to a lender. A bank or credit union's indirect lending team funds one for each deal it buys. They arrive as dealer portal scans or as paper sent overnight.
A funder needs the APR, the amount financed, the finance charge, the payment schedule and each itemized charge. The loan origination system checks those against the credit approval before funds are released, then books the contract into the servicing system.
What makes retail installment sale contracts hard
Every state has its own edition
Forms vendors publish a separate edition for each state's law. Lenders accept several vendors' forms, so line numbering and panel order shift between contracts.
Dozens of lines feed seven totals
The itemization nests lettered sub-lines for price, tax, fees and add-on products under numbered totals, ending in an amount financed the disclosure box repeats.
Typed values sit in printed blanks
Dealer software prints each figure into a preprinted blank, between a printed dollar sign and an (e) marker for an estimated amount.
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 retail installment sale contract, end to end
About this document: A copy of Reynolds and Reynolds' LAW 553 retail installment sale contract, California edition, filled in by an unknown party. Page 1 of the seven portrait pages is featured. It carries the parties, the vehicle, the Truth-in-Lending box and the payment schedule. Page 2 itemizes the amount financed, and pages 3 to 7 hold the terms and the arbitration provision. Personal details and reference numbers were replaced with fictitious equivalents.
Processing objective: The objective is to capture the financing terms a funder checks against the credit approval, from the Truth-in-Lending box and the payment schedule beneath it. The five values reconcile: sixty payments of 503.32 make the 30,199.20 total of payments, and the 2,500.00 down payment brings it to the 32,699.20 total sale price.
Demonstration fields: Annual percentage rate, Finance charge, Total of payments, Total sale price, Monthly payment
The document
4 of 7 pages
Page 1shown below
Page 2
Page 4
Page 5
Where each value came from
truth_in_lending.annual_percentage_rate
{
"value": 23.99,
"page": 1,
"range": {
"end": 2061,
"start": 2053
},
"box": {
"xmin": 0.02922,
"ymin": 0.4416,
"xmax": 0.15319,
"ymax": 0.46036
}
}truth_in_lending.finance_charge
{
"value": 12706.64,
"page": 1,
"range": {
"end": 2086,
"start": 2070
},
"box": {
"xmin": 0.15318,
"ymin": 0.44158,
"xmax": 0.27689,
"ymax": 0.46037
}
}truth_in_lending.total_of_payments
{
"value": 30199.2,
"page": 1,
"range": {
"end": 2135,
"start": 2119
},
"box": {
"xmin": 0.40039,
"ymin": 0.44154,
"xmax": 0.52416,
"ymax": 0.46038
}
}truth_in_lending.total_sale_price
{
"value": 32699.2,
"page": 1,
"range": {
"end": 2160,
"start": 2144
},
"box": {
"xmin": 0.52414,
"ymin": 0.44153,
"xmax": 0.64781,
"ymax": 0.46038
}
}payment_schedule[0].amount_of_payment
{
"value": 503.32,
"page": 1,
"range": {
"end": 2569,
"start": 2562
},
"box": {
"xmin": 0.23575,
"ymin": 0.59334,
"xmax": 0.44167,
"ymax": 0.62372
}
}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.
