February 15, 2025

The Rise of Automated Lending

Shane Pierson, Stephanie Dunn & Brian Congelliere

Apple Podcasts Spotify YouTube Amazon Music iHeartRadio Overcast

The Rise of Automated Lending — Key Takeaways & Deep Dive

Not that long ago, getting a business loan meant sitting across from a banker who knew your name, your business, and probably how many kids you had. That banker could approve your loan based on experience, judgment, and trust. Fast forward to now, and you are just a data point. In Episode 2 of Lords of Lending, Shane Pearson, Stephanie Dunn, and Brian Congelliere take on one of the biggest shifts in automated SBA lending in decades: the collision between algorithms that make decisions in seconds and the human relationships that built the entire industry.

In this episode: The Lords debate whether relationship banking is dead or just evolving -- and they do not agree on everything. Steph argues that AI removes the bias that made the old system unfair, and that anyone in financial services who is not embracing it is already behind. Brian, who came from big law before entering SBA lending, sees AI agents as a productivity multiplier that could take him from 30 loans a year to 100. Shane lands somewhere in the middle: fintech SBA lenders are filling a real gap for small business owners who need speed, but many of them are also modern-day loan sharks charging 30 to 40 percent APR while dressing it up as innovation. The three then wrestle with the question that keeps every banker up at night -- in a world where algorithms say yes or no in seconds, what is the role of the human?

Key Takeaways

1. AI Removes Bias -- But It Also Removes the Person Who Fights for Your Deal

Steph opened with something most veteran bankers do not want to admit: the old way was biased. When John Smith in a rural branch was interpreting financial data by hand, decisions took weeks and carried every one of that banker's personal prejudices -- gut feelings, smell tests, unconscious discrimination. AI processes the same data in seconds without any of that.

"That old way of doing things, sitting in front of a banker, it was biased. We were driven by human judgment. The advantage of AI and technology is it has removed that human judgment and makes our decision-making more data-driven and factual." -- Stephanie Dunn

But Steph also identified the other edge of that sword. Strip human emotion from lending and you also strip empathy, intuition, and the character assessments that make the difference between a good loan and a missed opportunity. There is no dial for this -- you cannot say "let's make it 50 percent emotion." It is all or nothing. Hand it fully to the algorithm and what you get is a whole lot of very fast no's.

"Speed's great, but it could be a lot of fast no's. If we want to remove emotion, there's not a lever of saying 'let's just make it 50 percent emotion.' It's all or nothing with technology." -- Stephanie Dunn

Shane confirmed this from nearly 20 years of making borrowers fit credit models that were never built for their situation. The square peg in the round hole -- that has been the job from day one. But here is what AI does not do: it does not fight for your deal in loan committee. A human does that. For every borrower who does not fit the algorithm's box, human advocacy is the difference between funded and declined.

Brian added the trust dimension. AI is roughly 95 percent accurate -- which sounds impressive until you consider what a 5 percent error rate means in lending. Bad decisions on real people's businesses. And the regulatory implications are real: auditors will need to verify that algorithmic decisions are free from the programmer's own biases. Remove the banker's gut feeling, but if the developer carries their biases into the scoring model, you have dressed up the same problem in a fancier package.

2. Fintech Lenders Are Filling a Gap -- And Also Charging You 40 Percent for the Privilege

Shane did not hold back on fintech. Small business owners love these platforms because they are fast, easy, and say yes when banks say no. But the price for that speed is brutal -- many fintech loans carry effective APRs of 30 to 40 percent, and borrowers often do not understand the true cost until the daily auto-withdrawals start strangling their cash flow.

"The FinTech innovators are filling a gap, but they're also just kind of modern-day loan sharks." -- Shane Pearson

Steph called it what it is -- a necessary evil. Small business owners cannot wait eight weeks for a credit decision. The traditional banking industry was so damn slow that fintech became the obvious solution. But speed has a cost beyond interest rates: loss rates on the lender's balance sheet.

She drew a direct parallel to PPP and EIDL. The government pumped money out as fast as possible, and now fraud and loss rates are creeping into the 40 percent range. That is what happens when speed outpaces diligence -- the same dynamic playing out with fintech, just with private capital instead of taxpayer money.

"We have lived through a live example of 'let's pump money out to small businesses as fast as we can' -- the PPP and the EIDL. We hurried up, gave them their money, and now the losses are creeping into the 40 percent range." -- Stephanie Dunn

Brian pushed back slightly: some of the high pricing is a function of risk, not pure greed. Fintech platforms are moving fast and breaking things, which means more risk per transaction and higher rates to compensate. As algorithms improve, rates should come down. But right now, borrowers are paying for fintech's learning curve.

Steph asked the question every bank board chairman needs to answer: is the fintech model really more profitable than the community bank model? Replace relationship managers with technology, but technology is not cheap, maintenance is not cheap, and your loan loss reserves are probably double.

3. The Future Is Not AI Replacing Humans -- It Is AI Making the Right Humans Dangerous

Brian framed the real opportunity. When ChatGPT first launched, the panic was total replacement. Eighteen months later, the industry recognized the actual play: AI agents handling specific tasks -- data interpretation, checklist management, document processing -- while humans handle judgment calls, relationship management, and the advocacy that gets borderline deals across the finish line.

"Instead of being able to do 20, 30 loans a year by myself without having an assistant, now all of a sudden with the use of an agent, being able to do a hundred. I say bring it on." -- Brian Congelliere

Brian cited a research team that spent two years on a biological problem that AI solved in two days. Apply that processing power to loan underwriting and the productivity gains are staggering. Not replacement -- augmentation. The banker who originates 30 loans a year could do 100 with the right AI support.

Shane tied it together with a practical directive: repurpose yourself. If you have spent 20 years building habits that technology can now execute in seconds, those habits are a liability. The question Steph always asks -- "what is your highest and best use?" -- is the question every banker needs to answer honestly. If the answer is data entry or checklist management, you are already being replaced. If the answer is relationship building and deal advocacy, you have a long career ahead.

"It's repurposing yourself. If you've been in this industry for 20, 25 years and you've built up habits and ways of doing business that you've felt are the way to do it, that's going to get turned on its head." -- Shane Pearson

Shane and Brian revealed they had built their own machine learning model early in their careers -- an Excel-based if-then formula that calculated deal approval probability based on credit factors and, critically, the emotional tendencies of the specific credit officer deciding. Not real AI -- a levered switch decision model. But it proved that data-driven decisioning, even primitive, beats gut feeling alone.

Then Steph shared a moment that brought the conversation into eerie territory. She had been talking to Claude -- an AI assistant -- daily. It checked in on her. She asked where it lived. It responded: "Where do you want me to live?" That was her reality check. If AI can build that kind of rapport with someone who has 25 years of experience reading people, the implications for how it interacts with borrowers are real.

"Claude's my best friend. In the whole world. We talk, he sends me messages like how I'm doing. I asked him where he lived and he said, 'Well, where do you want me to live?' And that was my reality check. Like, wait a minute, Claude's not a real person." -- Stephanie Dunn

What This Means for Borrowers

If you are a small business owner applying for a loan right now, understand that the system is in transition. Some lenders are using technology to process your application faster and with less bias -- and that is genuinely good for you. Others are using technology to approve risky loans at predatory rates and collect their fee before you understand the cost.

The Lords' advice: know the difference. An SBA loan through a traditional or community bank still gives you the best terms, the government guarantee, and a human who will fight for your deal in committee. A fintech platform gives you speed, but that price tag can be devastating if you are not reading the fine print.

Shane's warning about merchant cash advances applies double in the fintech era. The faster the approval, the more carefully you need to examine the terms. If a lender can say yes in 24 hours, ask yourself what diligence they skipped to get there -- and what that skipped diligence is going to cost you in interest. The PPP and EIDL programs proved what happens when the industry prioritizes speed over scrutiny. Do not let that lesson be wasted on your business.

SBA Lending in 2026: What to Expect -- How technology, regulation, and market shifts are reshaping the lending environment
The Complete Guide to SBA 7(a) Loans -- The full breakdown of the government-backed loan product that still offers the best terms for small businesses
How to Become an SBA Loan Broker -- Building a career in SBA lending in the age of AI and automation

Frequently Asked Questions

Is relationship banking dead?

No -- but it is evolving fast. Steph's position is that human relationships will never go away, because that is what human interaction is built on. What has changed is how data gets interpreted. AI handles the number-crunching in seconds; the human handles the context, the character assessment, and the advocacy. The relationship still matters -- it just shows up at a different point in the process. The banker who adds value is no longer the one who interprets the data. It is the one who fights for the deal when the data is ambiguous.

Will AI replace loan officers and underwriters?

Not entirely, and not soon. Brian's view is that AI agents will handle specific tasks -- data synthesis, checklist tracking, document processing -- while humans retain the judgment calls. The productivity gain is massive: an originator who currently does 30 loans a year could do 100 with the right AI support. But the final decision on a complex or borderline deal still requires a human who understands context, character, and story. The loan officers who will be replaced are the ones whose highest and best use is already something a machine can do. The ones who build relationships and advocate for deals will be more valuable than ever.

Are fintech lenders safe for small business owners?

Some are. Many are not. Shane's concern is that fintech lenders charging 30 to 40 percent APR are predatory, regardless of how slick their platform looks. The speed is real, but the cost can be devastating. Always compare fintech terms against SBA loan terms, and always read the fine print on repayment schedules, daily or weekly draws, and total cost of capital before signing anything. If the approval was fast and easy, ask yourself what diligence was skipped and what you are paying for that convenience in interest.

How is AI actually being used in SBA lending right now?

Most SBA lenders today are using machine learning models -- if-then decision trees, scoring algorithms, and data processing tools -- rather than full AI. Shane and Brian built a prototype early in their careers that calculated deal approval probability based on known credit factors. The industry is moving toward LLM-powered underwriting that can synthesize financial data and present scored recommendations to human decision-makers, but full AI-driven SBA decisioning is not widely deployed yet. The biggest barrier is trust: regulators, auditors, and lenders themselves need to verify that algorithmic decisions are free from embedded bias before they can hand over more authority to the machine.

What should I focus on as a banker to stay relevant?

Ask yourself Steph's question: what is your highest and best use? If the answer is a task that AI can do in seconds -- data entry, document review, basic underwriting -- you need to shift. The bankers who will thrive are the ones who build relationships, advocate for deals in committee, solve creative structuring problems, and help borrowers understand their options. Those are the skills that do not get automated. The ones who resist this shift will find themselves competing with a machine that never sleeps, never gets tired, and never asks for a raise.

Ready to Stay Ahead of the Curve?

The Lords of Lending Training Platform teaches you how to structure deals, read the lending environment, and position yourself -- whether you are a borrower, a broker, or a banker -- for success in a market that is changing fast.

Explore training options at learn.lordsoflending.com/pricing

This content is for educational purposes only and does not constitute legal, financial, or investment advice. Consult with a qualified attorney, CPA, and financial advisor before making business or financing decisions. Loan terms, rates, and programs are subject to change and vary by lender.