Extracting data from financial models
Sponsors and their advisers build financial models in Excel, each with its own sheet structure and timeline. Lenders and investment committees receive one with every deal, often dozens of sheets long. They arrive as workbooks or as PDF printouts, in a data room or appended to a report.
A credit or investment analyst needs total project cost, loan balances, the debt service coverage ratio by year, the exit value and the IRR. Those figures load into the reviewer's underwriting model, which tests them against a coverage minimum and a return hurdle.
What makes financial models hard
Wide grids with stacked headers
A cash flow sheet sets a total beside ten or more annual columns, under stacked headers naming the calendar year and the project year.
The same template, sheet after sheet
Models repeat one template for each product type or scenario, so the office sheet and the retail sheet share every row label.
Printed to fit one page
Models are printed to fit each sheet on one page, so a fifteen-column grid shrinks to small type with negatives in red parentheses.
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 financial model, end to end
About this document: The Aloha Development Model by JLL, a real estate pro forma for redeveloping Aloha Stadium. The State of Hawaii published it as an exhibit to a report by Foley & Lardner. Its 16 pages are landscape spreadsheet printouts at eleven page sizes. A returns summary and assumption sheets precede cash flow sheets for office, retail, residential and hotel. Page 6 is the office own-operate cash flow sheet.
Processing objective: The objective is to follow one development from cost through debt to exit. The sheet's columns are a total and thirteen years, three of construction and ten of hold. The loan and coverage figures sit in the first operating year, after three empty columns.
Demonstration fields: Total costs, Permanent loan beginning balance, First-year DSCR, Residual value, Owner IRR
The document
4 of 16 pages
Page 1
Page 6shown below
Page 8
Page 10
Where each value came from
cash_flow_scenarios[0].total_costs
{
"value": 64299010,
"page": 6,
"range": {
"end": 24848,
"start": 24837
},
"box": {
"xmin": 0.26918,
"ymin": 0.20511,
"xmax": 0.31583,
"ymax": 0.22559
}
}cash_flow_scenarios[0].permanent_loan_beginning_balance
{
"value": -50971154,
"page": 6,
"range": {
"end": 27921,
"start": 27908
},
"box": {
"xmin": 0.45063,
"ymin": 0.44356,
"xmax": 0.49623,
"ymax": 0.45472
}
}cash_flow_scenarios[0].first_year_dscr
{
"value": 1.25,
"page": 6,
"range": {
"end": 32592,
"start": 32588
},
"box": {
"xmin": 0.4506,
"ymin": 0.66227,
"xmax": 0.49616,
"ymax": 0.67277
}
}cash_flow_scenarios[0].residual_value
{
"value": 83029620,
"page": 6,
"range": {
"end": 31007,
"start": 30996
},
"box": {
"xmin": 0.26846,
"ymin": 0.5893,
"xmax": 0.31552,
"ymax": 0.59997
}
}cash_flow_scenarios[0].aloha_irr
{
"value": "14.25%",
"page": 6,
"range": {
"end": 34492,
"start": 34486
},
"box": {
"xmin": 0.26808,
"ymin": 0.77782,
"xmax": 0.31537,
"ymax": 0.79733
}
}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.
