How AI Is Changing Loan Underwriting
AI underwriting is approving more small businesses and catching more risk. Here is what it means for your next loan application.
The Limits of Traditional Credit Scoring
Traditional small business credit underwriting relies on a narrow set of inputs: personal FICO score, business credit scores, time in business, annual revenue, and financial ratios derived from tax returns. These inputs were selected because they were measurable, comparable, and predictive of loan performance — and they still are. But they are also a limited window into a business's actual creditworthiness. A seasonal business with excellent revenue 8 months per year looks distressed on a simple debt-service-coverage ratio calculated from annual revenue. A startup with no credit history may have a highly creditworthy founding team. A business in an industry with unusual revenue recognition patterns may look unhealthy by conventional metrics despite strong underlying economics. Traditional models systematically deny credit to many businesses that would be good borrowers.
What AI Models Add to Underwriting
Modern AI underwriting models consume far more data than traditional scorecards. With open banking integration, lenders can analyze 12 or 24 months of actual bank transaction data — not just annual totals, but the pattern of deposits, the stability of revenue, the timing of payroll, the presence or absence of overdrafts, and the trajectory of growth. This granular data reveals the health of a business far more accurately than a tax return summary. Beyond bank data, AI underwriting can incorporate payment processing data (if you use a processor that shares data), public data about the business (reviews, web presence, license status), industry benchmarks (how does your performance compare to similar businesses in your market), and even macroeconomic signals that affect your industry's outlook. The result is a holistic picture of business health that traditional scorecards cannot produce.
Which Businesses Benefit Most from AI Underwriting
The businesses that benefit most from AI underwriting are those that traditional models systematically undervalue. Seasonal businesses with strong but uneven revenue. Businesses in underserved communities where traditional banking relationships are thinner. Businesses in industries with unconventional accounting (like agriculture or construction, with project-based revenue recognition). Newer businesses without long credit histories but with demonstrably strong operations. For these businesses, AI underwriting is not just faster — it can be the difference between accessing credit and being denied entirely. Independent studies have found that AI underwriting models approve 15–25% more applicants than traditional scorecards for the same target loss rates, because the additional data allows more accurate identification of truly creditworthy businesses.
The Risks and Limitations of AI Underwriting
AI underwriting is not without risk for borrowers. Models trained on historical data can encode historical biases — if certain types of businesses were systematically denied credit in the past, the model may learn to deny them in the future, perpetuating discrimination. Regulators are increasingly focused on this risk, requiring lenders to test AI models for disparate impact and provide explainable adverse action reasons when applications are declined. AI models can also be brittle in novel economic conditions. A model trained on data from 2015–2023 may not perform well in economic environments that differ significantly from the training period. The COVID pandemic exposed this brittleness in many models. The best AI underwriters combine machine learning with human oversight and model monitoring.
What This Means for Your Application
For borrowers, the AI underwriting revolution has practical implications. Connecting your bank accounts and payment processing data to lenders (with your permission) gives AI models the data they need to evaluate you favorably — if your underlying cash flow is strong, more data generally helps. Maintaining clean, consistent banking habits — avoiding overdrafts, keeping revenue through consistent accounts, maintaining adequate balances — improves your profile in ways that AI models specifically detect. If you are denied credit by a lender using AI underwriting, ask for a specific reason. Regulatory requirements increasingly mandate explainable AI for credit decisions — you should be able to understand what factors drove an adverse decision. Sometimes the reason reveals a fixable issue (a bank account with unusual activity, a period of low deposits) that can be explained or addressed.
