AI is already in asset-based lending (ABL). It can sort receivables agings, flag unusual invoices and assemble in minutes a data set that once cost an examiner most of a day. The open question is how much of the examination it should be trusted with.
The principals at Asset Based Lending Consultants answer that question from different positions. Donald Clarke has watched field examination evolve over roughly five decades, from paper to spreadsheets to remote review. His son Dwight Clarke came up in a more technology-driven environment and sees AI as a way to make examinations faster and broader. We asked them to talk it through.
Where Does AI Help First?
Dwight: An examination starts with a mountain of material: agings, inventory reports, borrowing base certificates, financial statements, bank activity and backup. Much of the early time goes to getting that material into a form where analysis can begin. That is where AI helps first. If it can prepare the data, review the reports, make selections and flag exceptions, the examiner starts further along. It should not perform the examination. Reviewing collateral, understanding the borrower and making credit judgments stay with qualified people.
Don: I have watched this work move from manual processes to spreadsheets to remote reviews, so AI is one more step in a long progression. If it can review a large data set, group related accounts or pull samples against set criteria, that is useful. But finding something and understanding it are different things. AI may tell you where to look. The examiner still has to go look. My concern is efficiency becoming a substitute for verification.
Can AI Make the Review Broader and More Effective?
Dwight: Sampling is predictable. Borrowers know examiners tend to start with the larger invoices. AI can look across far more of the population and surface duplicate billing, unusual month-end activity or invoice amounts outside the normal pattern. The examiner begins with a short list of exceptions instead of thousands of records.
Don: An exception is not a conclusion. Unusual activity often has a perfectly good explanation. It becomes a credit concern when the facts show the lender’s collateral or position may be at risk, and the examiner has to establish the cause and the significance.
What Should Happen With the Time AI Saves?
Dwight: It should go into analysis. Investigate the inconsistencies, talk with the borrower, communicate with the lender and understand what is happening behind the numbers. If technology frees the examiner for that work, efficiency improves the quality of the examination as well as its speed.
Don: Agreed. When the system flags an exception, go investigate it. Ask the question, request the support, follow the lead.
Where Does Human Intelligence Remain Essential?
Don: Documents do not tell the whole story. An examiner can watch how inventory is handled, open a box, talk with the people on the floor and compare what they say with what is in front of them. Sometimes the explanation does not match the records. Hesitation leads to another question. AI can organize the evidence, but the examiner has to interpret it and decide where to go next.
Dwight: Some things cannot be verified remotely. Physical counts and the condition of inventory still require someone to see them. AI also makes mistakes. It can give conflicting answers or explanations the facts do not support. Where it helps is in pointing the examiner toward something worth attention. If receivables are rising while sales are falling, AI might suggest checking whether customers are paying more slowly. That is a useful prompt, but the examiner still has to determine whether it is correct.
Is There a Risk of Relying Too Heavily on AI?
Don: Yes. Newer examiners may accept an output without understanding how the conclusion was reached. The examiner has to question the finding, verify the evidence and reach an independent conclusion. Skip that, and you also skip the experience that builds sound judgment.
Dwight: That is why fundamentals matter more, not less. Someone who does not understand ABL or credit risk could paste an incorrect AI output into a report a lender relies on. The future examiner needs a strong grounding in ABL and enough knowledge of AI to use it critically. They have to know what to question and what to double-check.
AI may change the examiner’s skill set. It does not reduce the need for expertise. It raises it.
Who Remains Accountable?
Don: I think of the examiner as the cop on the beat. You follow leads, ask questions, apply experience and common sense, and reach a conclusion the facts support. AI does not remove that responsibility. Independence, verification and professional skepticism stay part of the job.
The same holds beyond field examination. AI can organize information, track missing items and prepare preliminary summaries. Decisions on reserves, advance rates, risk ratings and collateral availability still belong to accountable professionals.
How Far Apart Are the Two Generations?
Dwight: My question is usually what this technology can help us do better. Can we review more, find issues sooner, improve reporting and give examiners more time to analyze and communicate?
Don: Mine is what responsibility cannot be delegated. You can automate a great deal, but you cannot automate accountability. Somebody still has to apply knowledge, experience and common sense.
One starts with what AI makes possible. The other starts with what due diligence cannot afford to lose. They meet at the same boundary: AI takes on more of the repetitive work, and the examiner spends more time on the borrower and the concerns that matter.
What Comes Next?
Dwight: Too much time is still spent requesting reports and waiting for files. Authorized connections to borrowers’ accounting systems could let software retrieve the required information and prepare it for review. The examiner would still verify the results, but another layer of administrative work would disappear. The role shifts toward evaluating information and away from assembling it.
Don: Whatever comes next, the principle holds. Let AI carry the quantitative work: go through every record, tie out every number and flag whatever looks off, faster than any person could. Then let the examiner carry the qualitative work: ask why, judge whether the answer holds up and understand how the business really runs. The data shows what happened. Experience explains why it matters.
The future of ABL due diligence is unlikely to be a contest between artificial and human intelligence. It is more likely to be artificial intelligence guided by human intelligence. AI can help examiners see more and get there sooner. Someone still has to understand what it means.
Donald Clarke is president of Asset Based Lending Consultants and Don Clarke Enterprises. A senior instructor with the Secured Finance Network (SFNet) who has taught asset-based lending for more than 30 years, he received the organization’s 2019 Harry H. Chen Memorial Award of Excellence and 2021 Lifetime Achievement Award, and was inducted into its Hall of Fame. He is the author of “Asset Based Lending Disciplines,” the first textbook on the subject.
Dwight Clarke, CFE, is vice president of Asset Based Lending Consultants. He conducts field examinations and financial due diligence for lenders worldwide. He holds a master’s degree in finance and management from the University of Central Lancashire, earned the Certified Fraud Examiner designation in 2020 and was named to SFNet’s 40 Under 40 in 2021.
