H1 Agentic AI in managing credit and residual value risk

Use cases driving improvements in how specialist lenders assess credit and residual value risk.

In commercial credit, rules-based and machine learning systems filter clear rejections at scale. But the grey area — which in larger transactions can account for 90% or more of applications — has always depended on underwriters spending most of their time assembling information rather than exercising judgement. Agentic AI changes this: by gathering data, contextualising risk signals, and producing a structured first-draft credit paper, it allows underwriters to focus on the decision. McKinsey estimates a 40–80% productivity uplift per use case. The underwriter remains the decision-maker — required under SM&CR and Consumer Duty, which demand that credit decisions are accountable, auditable, and explainable.

In residual value management, AI is replacing periodic price guides with real-time analysis — assessing vehicle condition, disposal channel pricing, and optimal remarketing routes continuously. LeasePlan's work on AI-assisted RV setting and Autobiz's deployment by Stellantis, which compresses the resale process by up to three weeks, illustrate what is now in production. For lenders with EV portfolio exposure, where residual values remain under pressure, real-time RV monitoring has become a risk management necessity.

Learn how agentic AI is being used in commercial credit underwriting — not to make decisions, but to assemble and contextualise the information that allows underwriters to make better ones.
Find out how lenders are using AI to move from periodic residual value guides to real-time monitoring — and what this means for portfolio risk management as EV residual values remain under pressure.
Explore the governance and regulatory requirements that apply to both use cases — including the FCA's expectations under Consumer Duty and SM&CR for AI-assisted risk decisions.

H2 AI vs AI: how AI changes fraud landscape

AI vs AI: how the fraud landscape is changing for specialist lenders

Understanding how AI is reshaping both the fraud threat and the response.

Cifas recorded 421,000 cases filed to the National Fraud Database in 2024 — a 13% increase and the highest number on record, with a case filed every two minutes. False documentation is the single most common filing category, directly relevant to equipment and asset finance lenders whose credit processes depend on the integrity of financial statements, identity documents, and business records. UK Finance data for the first half of 2025 shows over 2 million confirmed fraud cases — a 17% increase on the same period a year earlier. The consensus among practitioners is that this is not a cyclical blip: nearly 60% of companies reported an increase in fraud losses from 2024 to 2025.

The problem is not simply that fraud is more common. It is that the automation and speed that make digital lending efficient also reduce the friction that previously slowed fraudsters down. Faster decisioning and digital-first onboarding create a larger and more accessible attack surface. Fraudsters are now able to test lender controls systematically — starting with smaller applications to probe detection thresholds before scaling to larger exposures. Agentic AI introduces a further and more acute challenge: the point at which autonomous AI systems designed to transact on behalf of users become indistinguishable from the bots fraudsters deploy for the same purpose — running high-volume attacks at a scale and speed no human operation could sustain. Granting AI agents the ability to access and interact with internal systems also creates significant cybersecurity risks should malicious actors be able to control or influence AI agents via prompt injection attacks.

Learn what the current fraud threat landscape looks like for specialist lenders — which attack types are growing, where the vulnerabilities in lending processes are concentrated, and what the data shows about scale and trajectory.
Find out how the shift towards automated decisioning and agentic AI is changing the risk calculus — and what lenders need to put in place before automation creates exposure rather than efficiency.
Explore the detection and prevention tools that are demonstrating results in the specialist lending context, including approaches to document verification, behavioural analytics, and shared intelligence.