Editor’s Note — This article is sponsored by Parlay. As with all sponsored content in Fintech Takes, this article was written, edited, and published by me, Alex Johnson. I hope you enjoy it!
Two dentists. A dozen restaurants. Four large manufacturers. Maybe a couple of oyster farms.
This is what “specialization” can look like if you’re a community bank.
Put another way, community banks are stuck with an especially complex part of the lending market.
Big banks flank them on consumer lending (a business that benefits from massive scale) and large commercial lending (a business that is attractive because of its wide margins). Meanwhile, vertical SaaS companies are increasingly competing in the opposite direction; they pick a single industry and build only for it. Community banks have neither luxury. Their mission is broad by nature and serious in intent: Serve the businesses in their communities, especially the ones left behind by other financial institutions. And the businesses that tend to get left behind for community banks to serve, somewhat by default, are the small and midsize ones.
Community banks may have a narrow place to win, but “SMB” is a funny thing to call a segment when it contains businesses that have almost nothing in common beyond not being large. Underwriting them well means knowing what normal seasonality looks like for an HVAC installer, or an ice cream shop, what counts as useful collateral, and which risks are actually just symptoms of not understanding the business well enough.
That’s why vertical SaaS is such an appealing model. Learn one industry deeply enough, and you can tell the owner, essentially, this was built for you. In lending, that specialization is even more meaningful because understanding the business changes the quality of the resulting credit decisions.
But a community bank doesn’t have quite the same freedom. It can’t become the world’s greatest lender to daycares and preschools and call it a day when there are still two dentist offices, twelve restaurants, four manufacturers, and a smattering of oyster farms in town looking for a loan.
And historically, that’s where community banks get stuck because specialization isn’t something you can simply decide to have. You have to accumulate it.
This reminds me of the old joke about a tourist at Wimbledon asking the groundskeeper how he gets the grass so green. The groundskeeper chuckles and says, “You start with 2,000 years of rain.”
Vertical lending expertise has worked more or less the same way: Rinse and repeat until the pattern becomes second nature. But that’s not the case for community banks. If a community bank has two dentists in town and has underwritten one of them, they are already 50% of the way to knowing everything that they will know — through direct experience — about underwriting dentist offices.
The 2,000 years of rain is really 2,000 years of accumulated experience.
What AI can change now is who gets to retain it.
Historically, a human underwriter would accumulate that experience over the course of their career. Now that experience can accumulate in a system built to retain what each deal teaches — a silicon underwriter, if you will — that has more capacity than any one person.
A skill can encode a procedure once, whether that’s spreading a franchise’s P&L, testing SBA affiliation, or sizing a loan against seasonal cash flow, and make it reusable across every relevant deal after. Every override, every no, we don’t count that revenue, becomes institutional context instead of something that leaves with the person who personally experienced it. Funded loans and eventual defaults feed the same loop, so the 400th SBA deal informs the 401st.
The silicon underwriter wouldn’t be making the final decisions, and no examiner would want it to. A business owner still wants a person accountable for the final answer, and regulators feel the same way. But a silicon underwriter could eliminate, say, 80% to 85% of the underwriting work on an SMB loan application, and reserve human judgment for the points in the decision that require it.
This points to an argument some banks still haven’t made to themselves. The obvious AI pitch in financial services is operational efficiency: Less paperwork, fewer manual steps.
Sure, fine.
But operational efficiency isn’t what boxes community banks in. Learning efficiency is. How many kinds of businesses can a community bank learn to underwrite well, and how quickly?
AI opens a new path.
With AI, community banks can build specialized context across a broader set of businesses that reflects their own market: Manufacturing here, oyster farming there, a completely different mix two counties over.
Say a community bank processes four small business loans a month. If forty showed up tomorrow, could it handle them? What about four hundred?
Probably not.
On the flip side, could the bank find more borrowers? Hire the loan officer with the largest rolodex from the competitor across the street? Open another branch?
Probably. Community banks are obviously resource constrained, but they know how to build relationships and generate loan demand with the resources that they do have. The much more pressing constraint is knowing that success on the front end creates more underwriting than the back end can absorb.
But if AI can scale context, specialization, and much of the work behind underwriting, think about how much faster a community bank can grow loan volume by shifting more of its energy into customer acquisition. The relationship advantage is already there. In the Federal Reserve Banks’ latest Small Business Credit Survey, 61% of small bank applicants said an existing relationship with the lender influenced where they chose to apply for financing (compared with 32% of applicants at online lenders who said the same).
What community banks have struggled to scale is the specialized underwriting capacity behind it, and that’s the larger promise of the silicon underwriter. Niche expertise used to live and die with the person who held it. Now each deal can leave the institution with more context than it had before, and that knowledge can support the next vertical, borrower, and a hundred loans.
The existence of AI doesn’t make it any easier to be a community bank. What it does create is a theoretical path to specialize across the strange mix of businesses that define a community; one that funnels more people into the relationships community banks already shine at and lets the bank take on more of the borrowers it already wants to serve.
For a closer look at how community banks can start building their own silicon underwriter, check out Parlay’s latest whitepaper.
