I underwrite small businesses for a living. I build the models that decide who gets capital and who does not. Six months ago I added an input nobody at our scale was using. Six months from now most of my competitors will be using it. The input is AI visibility.
This is not a marketing column. It is a credit column.
Here is the structural shift. A decade ago a small business walked into a bank. A loan officer pulled a credit report, looked at two years of returns, made a call. That market is gone. Banks retreated from the segment after 2008, the SBA tightened, regional lenders consolidated, and a non-bank category — alternative lenders, fintechs, revenue-based platforms — picked up the slack. The share of small business applicants seeking financing from online fintech lenders has nearly doubled in five years — from 17% in 2020 to 29% in 2025, per the Federal Reserve's most recent Small Business Credit Survey. We deploy capital where the bank used to. The decision is no longer made by a loan officer. It is made by a model.
Models read inputs. The classic ones still matter — cash flow, processor data, deposit history, sector risk, prior defaults. What changed is the inputs available in 2026 that did not exist in 2020. Processor APIs are richer. Bank-data feeds are real-time. And — quietly, and increasingly — the AI engines have become a credit signal.
Here is what I mean. Pull two businesses with identical revenue, identical processor mix, identical deposit balances. Ask ChatGPT, Claude, Perplexity, and Google AI Overviews the same prompt — "best provider of [category] in [city]" — and watch what comes back. One business is named, sourced, cited across trade press and third-party directories. The other is silent. Same P&L. Different futures.
The named business has a moat. Customer acquisition cost is lower because the AI engine sends warm leads. Pipeline durability is higher because the brand is legible to the model and the model is now the first stop for buyers. The silent business is one algorithm change away from a revenue cliff. As an underwriter, I have to price that.
I know the objection. Credit decisions should not factor brand presence — that is the consultant's instinct, not the lender's. Brand presence has always been a credit signal. The question was whether you could measure it. For decades you could not. Reviews were noisy. Press coverage was episodic. SEO was gamed. Citation Share — the rate at which a business is named in answers from AI engines on prompts its customers actually use — is the first measurable, comparable, real-time signal of brand presence that exists at the small-business level. It is cheap to query. It is hard to fake. It moves slowly enough to be a stock variable, not a flow variable. That is what a credit input looks like.
The implications run in both directions.
For lenders: the firms that build this into their models first will mispricing-trade their competitors for a window — maybe twelve months, maybe twenty-four. After that it gets repriced into the market and becomes table stakes, the same way bank-data feeds did between 2017 and 2021.
For small business owners: the answer engines are no longer just where customers find you. They are increasingly where capital finds you. A business that is invisible to ChatGPT is becoming a business that pays more for working capital — or does not get it at all.
Three things SMB operators should do this quarter:
Pull your Citation Share now. Run the prompts your customers run inside ChatGPT, Claude, Perplexity, and Google AI Overviews. Note where you are named. Note where you are silent. That gap is the work.
Treat trade coverage as infrastructure. Earned media in credible publications is what the AI engines pull from. Ad spend does not move this. Third-party authority does.
Make your business legible to the model. Consistent naming, structured data, clean category placement, sourced facts — Generative Engine Optimization (GEO) is the discipline. The model is reading you. Make yourself easy to read.
The capital market is a leading indicator. When lenders start paying attention to a signal, it is because that signal predicts outcomes. Citation Share predicts the next five years of small business performance. I am pricing it now. So will everyone else.
The model is the underwriter. The answer engine is the credit file. Show up there.
Alex Shvarts is the Founder and Chief Technology Officer of FundKite, a small business funding platform that has deployed more than $200 million in capital to U.S. small businesses since 2015. He writes on financial technology, alternative lending, and AI's impact on the SMB economy.
Sources
Federal Reserve Banks, 2026 Report on Employer Firms: Findings from the 2025 Small Business Credit Survey — https://www.fedsmallbusiness.org/reports/survey/2026/2026-report-on-employer-firms
PYMNTS, Earnings Show AI Lending Platforms Scaling Originations (March 2026) — https://www.pymnts.com/news/b2b-payments/2026/earnings-show-ai-lending-platforms-scaling-originations/
Federal Reserve Bank of Chicago, Small Firm Borrowing in Seventh District States: Results from the 2024 Small Business Credit Survey — https://www.chicagofed.org/publications/chicago-fed-insights/2025/2024-small-business-credit-survey
