AI Underwriting
Last updated October 1, 2026
AI underwriting means using artificial intelligence to help judge whether a borrower can repay a loan. It covers two different jobs.
The first is preparing the file. AI reads tax returns, bank statements and other documents, places the numbers into a financial spread, calculates ratios, lists missing items and flags figures that don't match across documents. It can also draft a summary of the borrower and the request. This use is practical today and easy to defend, because a person still reviews the work.
The second is making the decision. Here, a model scores the application or approves and declines on its own. For most commercial, small business and community lenders, it raises legal and supervisory questions they are not set up to answer.
Why it matters for lenders
Underwriting is often the slowest stage of lending, and much of that time goes to assembly: sorting files and keying numbers. AI can cut that time sharply and let underwriters spend their hours on judgment.
The decision is different. In the US, Regulation B requires that a statement of reasons for a credit denial be specific and give the principal reasons. A lender must be able to say truthfully why it declined. The Consumer Financial Protection Bureau withdrew its 2022 and 2023 circulars on complex algorithms in May 2025, but that rule in the regulation still applies. Fair lending laws also apply to any model, so lenders should be able to test results across groups of applicants on their own data.
Supervisory guidance is thin. In April 2026, the federal bank regulators replaced their model risk guidance, and it states that generative and agentic AI are outside its scope. That is one more reason to keep a human in the loop on each credit decision.
When choosing a tool, ask for an audit trail of every AI output, a named person's sign-off on decisions, reasons you can trace, links from each number back to its source page, logged overrides, and fair lending testing support.
How it works in practice
A small business lender gets a 90-page application package. The AI tool sorts the documents, spreads three years of financials and a year of bank statements, and drafts a summary. It flags that reported revenue is much higher than bank deposits.
The underwriter checks the spread, adjusts for a one-time equipment sale and asks the borrower about the revenue gap. The answer, a second bank account, resolves it. She writes and signs the credit memo, and the loan goes to the approver. The AI saved her most of a day, and the decision is hers. The AI lending page covers this split in more depth.
Related terms
Underwriting
Underwriting is the work of deciding whether a borrower can and will repay a loan, and on what terms. The underwriter gathers facts about the borrower, its finances, any collateral and any guarantors, measures them against the lender's written policy, and recommends whether to approve, decline or restructure the deal.
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.
Credit Memo
A credit memo, short for credit memorandum, is the underwriter's written case for or against a loan. It describes the request, the borrower, the repayment source, collateral and guarantors, the main risks and any policy exceptions, then makes a recommendation. Approvers use it to decide, and auditors and examiners use it later to see how the decision was made.
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.
Adverse Action Notice
An adverse action notice is the notice a lender must send when it denies a credit application or takes certain other negative actions. In the US, the Equal Credit Opportunity Act requires it for business and consumer credit. It must name the action and give specific reasons, or explain how to request them.
Further reading
AI Underwriting: What AI Should Prepare and What Humans Should Decide
Which underwriting tasks AI can take on today, which decisions should stay with a credit officer, and what to require from an AI vendor.
Cash Flow Underwriting: How It Works for CDFIs and Community Lenders
What cash flow underwriting is, the bank data it uses, how it compares to credit scores, and what a CDFI needs in place to use it fairly.
See how lenders put this into practice
FUNDINGO runs origination, servicing and reporting on one loan record inside Salesforce.