Floor plan lending is one of the few corners of financial services where 2005-era workflows are still considered normal. Spreadsheets. Manual audits. Underwriters reading PDF bank statements. Phone calls to dealers asking why a unit hasn't been paid off. The gap between what the industry does and what the technology now allows has grown wide enough that it's hard to ignore.

That gap is starting to close, though not in the way the hype would have you believe.

The Problem with "AI Will Fix Everything"

Before getting into real examples, it's worth being direct about the limits of the technology. Most floor plan lenders talking to their board about AI are not deploying autonomous underwriting systems that make credit decisions without human review. A significant share, including large captives and independent lenders, are still in a wait-and-see posture, cautious about regulatory exposure and skeptical that AI explanations can satisfy examiners.

Santander's underwriting team has openly flagged that AI-assisted credit decisioning creates a compliance problem: if you can't reproduce and explain how a decision was made in terms a regulator understands, you're exposed. Ford Motor Credit, despite running 10 AI projects focused on early risk detection, has designated explicit no-fly zones around anything that directly touches dealer or customer credit decisions. The industry's caution isn't unfounded.

Still, there's a lot of space between full automation and doing nothing. That space is where the interesting work is happening.

Where AI Is Already Working

Fraud detection and pattern recognition. Auto lending fraud losses run 21 times higher than credit card fraud losses, with synthetic identity fraudsters holding access to an estimated $1.8 billion in automotive finance credit as of mid-2025, according to TransUnion data. Traditional rule-based detection can't keep up with coordinated fraud rings that now use generative AI to fabricate pay stubs, bank statements, and employment records. Lenders using AI-powered document analysis can evaluate not just whether data looks plausible but whether the document itself has been manipulated. That's a different kind of check than a human reviewer can reliably run at volume.

Portfolio monitoring and risk scoring. Some of the largest independent lenders have moved toward dynamic dashboards and machine learning to rank dealer performance and flag accounts that warrant closer attention. The goal is a continuously updated view showing which dealers to prioritize by risk level, growth opportunity, or disengagement rather than waiting until the next scheduled review. Ford Motor Credit's analytics team is already analyzing 100% of portfolio data for early risk indicators instead of relying on sampling alone.

Digital auditing. SBS's Digital Audit solution completed more than 100,000 digital audits across U.S. dealerships by the end of 2025, processing between 2.5 and 3 million verification images annually. That's a volume physically impossible to achieve through traditional field audits. NextGear Capital released a self-audit feature inside its mobile app in late 2024, allowing dealers to conduct their own inventory audits, with AI and API integration handling reconciliation workflows on the back end.

MCA and sold-out-of-trust detection. Independent lenders have explored open banking tools like Plaid to get real-time visibility into dealer bank balances, which helps catch merchant cash advance activity before an NSF or a missed payoff. Detecting MCA involvement through bank statement review used to require a human going line by line. Automated transaction categorization and anomaly detection can surface those patterns far faster.

Underwriting efficiency, not replacement. Ford Motor Credit ran a formal study comparing its proprietary scoring model against a machine learning model built on the same loan data. The ML model predicted creditworthiness more accurately, held promise for higher approval rates, and showed potential for meaningfully lower credit losses. Ford's conclusion wasn't to hand the decision to the machine. The company said it would take years of parallel testing before any live deployment. The lesson is that ML improves the signal; humans still own the decision.

What the Industry Is Actually Thinking

Conversations with lenders across the captive, bank, and independent segments surface the same pattern. AI is most welcome when it reduces low-value manual work and surfaces clear signals so humans can make faster, better-informed decisions. It's resisted when it threatens to make decisions that are hard to explain or audit.

Risk officers at major captives have described wanting real-time analysis of credit decision models that can flag performance drifts before defaults rise, a capability the industry hasn't fully developed. Operations leaders at independent lenders have talked about turning noisy data into simple, action-triggering outputs: risk scorecards that fire specific workflows when thresholds are crossed. Dealer services teams at banks have described wanting to automate audit triage and SOT reporting, not to shrink headcount but to extend what the team can cover.

That distinction matters. An AI system that reads 800 bank transactions and tells an underwriter "this dealer has MCA exposure and three NSFs in 90 days, here's the data" is welcome. An AI system that says "deny this application" without traceability is not.

The Compliance Reality

Regulators are paying attention. The CFPB has stated it is building out its data science capabilities to identify fair lending violations at every stage of the credit lifecycle, and AI-derived decisions face the same scrutiny as any other underwriting criteria. The technology doesn't insulate you from fair lending obligations.

That's actually a reason to build AI into workflows carefully, not a reason to avoid it. Moody's surveyed 600 risk and compliance professionals and found that 91% are aware of AI's role in risk and compliance, and 53% are actively using or trialing it, up from 30% in 2023. The consensus from that group: 84% agreed AI has significant advantages, and 42% said human oversight is mandatory, not optional. One respondent put it plainly: "You can't outsource accountability."

Where This Is Headed

Floor plan lending has a staffing problem. Experienced credit operations people are expensive and hard to find. The work of monitoring a large dealer portfolio, chasing annual review packages, reconciling audits, tracking title expirations, flagging NSFs, is labor-intensive and mostly repetitive. At the same time, dealer consolidation is creating larger, more financially complex borrowers who expect faster responses and better service.

AI doesn't solve all of that. Spreadsheets will still exist at some institutions five years from now. The technology takes time to integrate, and many lenders don't have the data infrastructure to deploy modern ML models even if they wanted to.

What's shifting is the cost-benefit math. When banks can process documents in 30 to 60 seconds instead of 3 to 4 minutes, when audit images are reviewed by computer vision rather than a field team driving to dealerships, when risk dashboards surface the three accounts that need attention today instead of making a portfolio manager sort through 200 rows in Excel, the value becomes concrete enough that the hesitation starts to break down.

The Case for Optimism

Wholesale finance has been a technology laggard for a long time. That's documented fact, not a knock on the people in it. Legacy LMS platforms weren't built for real-time data. Audit workflows were designed for a world before mobile cameras and GPS. Credit underwriting relied on periodic snapshots, quarterly financials and annual reviews, rather than continuous monitoring.

The good news is that the category is finally getting serious attention from technology companies, and the infrastructure for open banking data, digital auditing, and real-time portfolio analytics exists and is being deployed at scale. Dealers are increasingly comfortable with self-service digital tools. The lenders who figure this out first are going to have a real competitive edge: lower loss rates, faster funding, better dealer experiences, and teams that spend more time on relationship work than data entry.

The technology is genuinely useful. The lenders who treat it as a tool for their people rather than a replacement for their people are going to get the most out of it. That framing isn't just good PR. It's also what works.

What VeroOS Is Doing About It

At Vero, we've built AI features into VeroOS because we think the human-in-the-loop model isn't a compromise. It's the right design.

Our AI initiative, launched in 2025, focuses on automating data ingestion, surfacing real-time portfolio signals, and reducing manual effort in risk and servicing operations. That means things like automated document tracking with expiration alerts, risk dashboards that flag dealer accounts based on live bank balance data and payment patterns, custom risk alerts built on any data element in the system, and MCA detection through bank transaction analysis. The platform integrates with Plaid for continuous cash flow visibility and with digital audit providers so lenders aren't waiting 48 hours for results.

None of that replaces a credit judgment call or a relationship conversation. What it does is clear the decks. Servicing teams shouldn't spend half their day chasing documents or sorting through spreadsheets to find which accounts need attention. That time belongs to work that actually requires human judgment, complex credits, dealer relationships, portfolio strategy.

The goal is a team that's faster and better-informed, not smaller. That's the version of AI we're building toward, and based on what we hear from lenders across the country, it's the version they're ready for.

Vero Technologies builds VeroOS, a cloud-native wholesale finance platform for banks, captives, and independent lenders. Learn more at vero-technologies.com.

‍