Inside Hunt Score, the AI Underwriting Engine Behind HL Hunt’s Credit Products

SPONSORED CONTENT — Underneath HL Hunt’s consumer-facing credit products sits Hunt Score, the company’s in-house underwriting engine. HL Hunt describes it as an AI system that scores credit risk, screens for fraud and runs predictive analytics, and positions it as part of a broader internal suite that also includes the company’s Metro 2 reporting infrastructure and its merchant payment platform.

The pitch for any underwriting model built this way is speed and reach: a system that can weigh bureau data alongside transaction history and business financial signals in real time can, in principle, extend credit to files that a traditional bureau-score cutoff would reject outright — the same thin-file population HL Hunt’s credit builder products are aimed at. Whether Hunt Score does that in practice is not something HL Hunt has published data to demonstrate.

The company markets Hunt Score on two attributes in particular: fairness and explainability. Those are the two properties regulators and researchers have spent the most effort trying to define and measure in credit-scoring models generally, precisely because they are difficult to verify from the outside. A lender can assert that a model is fair without publishing the disparate-impact testing that would substantiate it, and can assert that a model is explainable without publishing the documentation that would let an outside reviewer check the explanation against the model’s actual behavior.

That gap is not unique to HL Hunt. It is the standard condition of proprietary underwriting models across the industry, and it is the reason bank regulators built a specific supervisory framework — the Federal Reserve’s SR 11-7 guidance on model risk management — requiring regulated lenders to validate models internally even when they will not disclose them externally. HL Hunt has not stated whether Hunt Score has undergone a comparable internal validation process, or whether one has been reviewed by an outside party.

The fair-lending stakes are concrete rather than abstract. Any model used to make or influence a credit decision in the United States is subject to the Equal Credit Opportunity Act, which requires specific and accurate reasons when an applicant is denied — an obligation that becomes harder to satisfy, not easier, as a scoring model’s internal logic becomes more complex. A system marketed as explainable should, in principle, make those adverse-action notices more precise. HL Hunt has not published examples of the notices Hunt Score generates or described how its stated explainability translates into the specific reason codes a rejected applicant receives.

None of this means Hunt Score performs poorly, or unfairly. It means the claims are, at this point, the company’s own, in the same way the credit builder’s outcome statistics were company-reported figures when this publication examined that product last week. HL Hunt has been more forthcoming than many private lenders about its compliance posture generally — it discloses its NMLS registration and describes state licensing as a work in progress rather than obscuring the gap — and that candor is worth noting even where the underwriting model itself remains a closed system.

For a prospective borrower, the practical takeaway is narrow: a real-time, alternative-data underwriting engine can plausibly open credit access that a pure bureau-score model would not. Whether Hunt Score does so fairly and consistently is not something an applicant, or this publication, can verify from published materials alone.