Lenders are increasingly blending automation with human review as they prioritise faster decisions and more robust verification, leveraging AI-driven document forensics amid evolving borrower data needs.
Small business funding is being reshaped by a blunt but practical demand: borrowers want faster decisions, while lenders want stronger proof that the data behind an application is real and complete. According to the Federal Deposit Insurance Corporation’s 2024 Small Business Lending Survey, about 30% of banks can approve a small, straightforward loan within one day and 75% can reach a decision within five business days, yet only 10% have credit-scoring systems that can partly or fully automate underwriting for non-credit-card loans. That gap is helping push lenders towards a model that blends automation with targeted human review.
Cash-flow analysis is taking on greater weight in that shift. Traditional credit scores, collateral and personal guarantees still matter, but lenders are paying closer attention to deposits, balance swings, returned payments, negative balances and existing debt because those figures can offer a more current picture of repayment capacity. The FDIC survey also found that about half of banks were already using or considering financial technology in small business lending, pointing to demand for tools that can extract data, flag fraud and standardise exception handling rather than replacing underwriting entirely.
That is also changing how lenders treat bank statements. Open-banking links can provide live account data, but many applicants still rely on uploaded PDFs because they may refuse a connection, bank with an unsupported institution or need to submit older statements from a closed account. As a result, lenders increasingly need both paths to feed the same review process, so connected data and uploaded documents can be compared on equal terms.
The bigger risk is that a statement can look legitimate even when it has been altered. MoneyThumb says document forensics is becoming more important because editing tools and generative AI can create convincing but false PDFs. That is one reason verification is moving beneath the surface of the page, with systems checking metadata, fonts, object structure and other technical markers for signs that a file has been rebuilt or manipulated. The company’s Thumbprint product is designed to inspect uploaded financial PDFs for those warning signs.
This is pushing underwriting away from broad manual review and towards exception handling. In practice, that means routine files can be screened automatically for completeness, categorised data, cash-flow patterns and document anomalies before an underwriter sees them, while unusual, incomplete or higher-risk applications move into a smaller queue. The approach can reduce wasted effort, but it still depends on clear rules, good audit trails and staff judgement for cases that fall outside the standard pattern.
Lenders are also likely to segment applications more aggressively by size and complexity. The FDIC’s findings suggest banks already move faster on simple loans than on more complicated ones, and the same logic points towards distinct workflows for small working-capital requests, larger expansion loans and applications with policy exceptions or inconsistent records. Small Business Administration data released in 2024 also showed strong demand for smaller loans, with the agency backing $56 billion in financing and more than 100,000 financings in fiscal 2024, much of it through smaller-dollar lending.
For lenders, the next competitive edge will probably come from explainability as much as speed. A score or flag is not enough on its own; compliance teams, auditors and borrowers need to know what triggered the result and which transaction, document feature or policy rule supported it. That is why the most effective systems are likely to preserve the submitted file, the extracted values, the review signals and the final decision, while leaving room for underwriters to override automated outcomes when a business’s pattern is unusual but still sound.
Disclaimer: This article is intended to inform and educate, not to recommend or endorse any financial product, investment or strategy. Please consider your own financial circumstances and seek professional advice where appropriate before making financial decisions.
