Trust But Verify: SBA Credit Memos and AI

AI can be a useful tool for SBA Lenders, helping reduce the time spent organizing information and drafting credit memos. It can summarize documents, identify trends, and create first (emphasis on first, please) drafts in minutes. 

But the rule everyone should remember is: TRUST BUT VERIFY.

(We’ve all seen news reports of various professionals relying on AI and not verifying, only to be caught out with fictional references and other career-ending problems!)

There are five SBA Credit Memo sections that AI can write surprisingly well – and then there are five sections that you should never let your AI tool touch.

Let’s look at the good ones first.

Five SBA Credit Memo Sections to delegate to AI (and verify)

Business History

AI excels at organizing information from business plans, websites, and borrower interviews into a clear business history.

Trust but Verify: dates, ownership, and key facts, and ensure the most-recent online information is what’s accessed.

Management Summary

AI is also good at summarizing resumés and highlighting management experience.

Trust but Verify: experience, credentials, and roles.

Historical Financial Performance

AI is, at its core, a pattern recognition tool. This means it’s great at trend identification for revenue, expenses, profitability, and debt service coverage.

Trust but Verify: calculations, ratios, and conclusions against financial statements.

You don’t want to let it draw conclusions without strict supervision!

Industry Overview

AI can quickly summarize publicly available information on industry trends and market conditions.

Trust but Verify: economic conditions change rapidly, so confirm that the information is current (it’s known for pulling out-of-date data), and that it’s relevant to the Borrower’s market and consistent with SBA lender requirements.

Collateral Summary

AI can organize appraisal information, collateral values, and lien positions into a concise summary. Just be sure to ask it for a concise summary; some models are well known for being rather verbose.

Trust but Verify: values, loan-to-value calculations, and appraisal conclusions. 

Again, watch what conclusions it’s putting forward!

Five Sections Requiring Human Judgment

These sections should not be entrusted – even with verification – to AI. They require human experience, interpretation, nuance, and judgment.

SBA Eligibility

SBA eligibility and program rules require interpretation. AI can help with research, but Lenders should always rely on their own expertise to confirm eligibility requirements and current guidance from the most-recent SBA SOPs.

Repayment Analysis

While AI can summarize the numbers, only human Lenders can determine whether cash flow is sufficient, sustainable, and supported.

Credit Recommendations

Approvals, conditions, and policy exceptions require Lender expertise based on the complete credit picture. AI cannot replicate the balance of quantitative analysis with qualitative judgment.

Policy Exceptions

AI might be able to identify potential exceptions, but it falls to experience Lender staff to determine if the exceptions are acceptable and appropriately documented to support SBA compliance.

Overall Risk Assessment

Management quality and Borrower character are human characteristics requiring human evaluation. And documentation inconsistencies and mitigating factors require a human to review the complete loan file.

In summary…

Before relying on AI-generated content, confirm:

• Financial calculations and ratios

• SBA guidance and SOP references

• Eligibility conclusions

• Appraisal and collateral information

• Names, dates, and ownership details

• Assumptions used in the analysis

AI can save time by producing an organized first draft, but in the end, the Lender is responsible for ensuring the final analysis is complete, accurate, and supported by the loan file. The goal isn’t to replace experienced underwriters (perish the thought!); it’s to give those underwriters more time to focus on the decisions that require their expertise and their human judgment.

That said, AI will undoubtedly become more common in SBA Lending, and this will require an extra level of care and caution from your team. Combining the careful use of technology with experienced underwriting, ongoing SBA lender training, and guidance from an experienced SBA consultant when needed, can help Lenders improve efficiency and grow their portfolios. 

We’re excited about the opportunities available for our clients using AI, and we’re here to help, whether through SBA lender training or as the SBA consultant you can call when you need some guidance. Curious? Give us a call at 877-576-0819, or contact us through this link.