Intelligent Document Processing
Last updated October 1, 2026
Intelligent document processing (IDP) is software that reads documents and turns them into data a system can use. It goes further than optical character recognition (OCR), which only turns an image of text into characters. IDP also understands what kind of document it is reading and which values matter.
Most IDP tools work in four steps:
- Classify. Decide what each file is: a tax return, a bank statement, a driver's license, a lease.
- Extract. Pull out the key values, such as net income, monthly deposits or a name and date of birth.
- Validate. Check those values against rules and other documents. Does the name on the ID match the application? Do the statement's ending balance and next month's starting balance agree?
- Deliver. Send clean, labeled data into the loan file.
Modern tools use machine learning and large language models, so they can handle layouts they have not seen before. Most also give each value a confidence score. Values below a set score go to a person to check.
Why it matters for lenders
Lenders handle large volumes of paper. Staff open each file, rename it, decide what it is and type numbers into a spreadsheet. That work is slow, and every keystroke is a chance for error.
IDP speeds up intake and financial spreading, and it can flag mismatches that a tired person might miss. It also makes it easier to see which stipulations are still outstanding.
It is not perfect. Scans are blurry, statements come in odd formats, and models misread numbers. Lenders should keep a person responsible for the final values, and the system should log every change that person makes. Each number in a spread should link back to the page and line it came from.
How it works in practice
A commercial real estate lender receives a borrower package with three years of tax returns, rent rolls and twelve months of bank statements, all in one 140-page PDF. The IDP tool splits it into separate documents, labels each one and files them on the loan.
It then pulls rental income and expenses into the spread and adds up monthly deposits. One value is flagged: the rent roll shows higher income than the tax return. The analyst reviews the flag, finds that two new leases started after year-end and notes the reason. She also corrects a misread expense line. The system records both actions.
Before IDP, the same analyst spent most of a day sorting and keying that package before she could start the real analysis.
Related terms
Financial Spreading
Financial spreading is entering a borrower's balance sheets, income statements and tax returns into a lender's standard format, so several years can be compared side by side and key ratios computed. The result, called a spread, is the base for commercial underwriting, from debt coverage to leverage, and for tracking the borrower's results after closing.
Bank Statement Analysis
Bank statement analysis is the review of a borrower's bank statements to measure true revenue, balances, overdrafts, returned payments and existing debt. Underwriters use it to check what the application says, size the loan or advance, and find obligations that don't show on a credit report, such as daily payments to other funders.
Stipulations (Stips)
Stipulations, or stips, are the documents and conditions a funder requires before it will release money, such as recent bank statements, the owner's ID, a voided check or proof of ownership. The term is common in small business lending and merchant cash advance funding. Each stip is typically tracked as requested, received or cleared.
Human-in-the-Loop
Human-in-the-loop means a person reviews and approves what an AI system prepared before it takes effect. The AI can gather data, draft a summary or suggest an action, but a named person decides and is accountable. In lending, it keeps credit decisions, adverse action notices and money movement under human control, with each review and override recorded.
AI Underwriting
AI underwriting uses artificial intelligence to help assess a loan. Most often, the AI prepares the file: it reads documents, pulls financial figures into a spread, checks for gaps and mismatches, and drafts a summary for the underwriter. A narrower use lets a model score or decide the loan, which raises fair lending and explanation questions that most lenders must answer first.
Further reading
See how lenders put this into practice
FUNDINGO runs origination, servicing and reporting on one loan record inside Salesforce.