Dealer floor-plan lending
A field auditor walks a dealer lot on a fixed cycle. The auditor photographs units and checks serials against what the dealer still owes. Most days it is clean. Then a unit is not where the schedule says it should be. The dealer's explanation sounds plausible.
The gap
The auditor decides on the spot, with a checklist and a phone. Escalate, and spend the dealer relationship on what may be a timing gap. Or write it down and recheck it next cycle. A full cycle is exactly how long a sold-out-of-trust unit can sit unflagged. Either way, the reasoning stays in the auditor's head. Six months later the loss is real, and someone asks why nobody escalated. There is a photo, a checklist, and no record of the judgment.
Where the tools stop
Tracking a unit is a record. Calling one out of trust is an accusation against a dealer. If the call is wrong, it carries the liability. So the software runs the clock, collects the photo, and reports the status. Then it hands the decision back to a person.
Logs verification status per unit, collects dealer-submitted photos, and triggers scheduled fees or maturity events when a unit stays unresolved past day 7 or day 15. That is a clock. It does not read the photo and the dealer's explanation and decide whether the gap is credible.
Pairs the platform with an in-house team of human auditors and reports status per unit through Wholesale Intelligence. The software records what a person decided; it does not decide.
Describes its own audit process in plain language: if there is no reasonable explanation, a unit may be considered sold out of trust. That is stated as a human call, not a system output, which is the most honest description in the category.
Markets an AI-driven fraud prevention platform using machine learning for portfolio-level anomaly and pattern detection. That is a real claim and the closest thing to a competitor here, though nothing published shows it reasoning across one unit's photos, serial history, dealer statement and payment history to a cited verdict on that unit.
What changes
tenday.ai builds an agent that reads a cycle's audit worksheet and photos. It weighs them against the curtailment schedule and the dealer's prior audit history. For each flagged unit it drafts one of three calls: explainable, needs follow-up, or likely out of trust. Each call carries a confidence score and the evidence it read. Explainable units clear, with the reasoning attached to the file. Everything else queues for the risk manager. The risk manager signs the escalation with the evidence already assembled, not rebuilt from scratch.
Drafted recommendation, awaiting signature
What the engagement needs
Audit worksheets and photos, the curtailment schedule, and the dealer's prior audit history. They arrive as files or by email. No regulated data class is involved. The risk manager or portfolio risk manager stays the signature on every escalation. The agent drafts. It never files a claim and never moves against a dealer's line.
The whole thing
Honest limits
This sells hardest in the weeks after a lender eats a sold-out-of-trust loss. That lender wants the next cycle to catch it faster. Without a recent loss in memory, the auditor's judgment call is a cost nobody prices. The build then competes against problems that already have a line item.
CheckVentory / uVerifi already sells AI-driven fraud prevention language to this exact buyer. Nothing public shows it producing a citable, unit-level verdict rather than a portfolio anomaly score. If a buyer conversation shows otherwise, the opening this rests on is narrower than it looks.
The vendor descriptions above come from published material, not hands-on trials.
More use cases
A borrower's renewal folder arrives as two years of tax returns, a bookkeeper's statements and a scanned rent roll. No two folders arrive in the same shape. The credit analyst retypes all of it by hand before any credit thinking starts.
Three weeks from the Form 5498 deadline, an analyst works down a stack of valuation files. In March it was a trickle. An appraisal, a K-1, a sponsor letter, a purchase agreement nobody has updated in three years.
A regional downturn does not send one default notice. It sends a run of them inside the same few weeks. Every one waits for an examiner to read the bond. Until then, nobody knows which are fast denies and which are full investigations.