
For the credit and collection industry, the developments carry particular significance. AI-assisted underwriting can change how institutions evaluate applicants, while analytical tools can help staff identify at-risk loans and organize follow-up. But adopting the technology also raises questions about fair lending, data privacy, vendor oversight and the reliability of automated recommendations.
Faster Lending Decisions
One of the clearest applications is underwriting, where institutions are using AI models to evaluate credit information and automate more lending decisions.
At iQ Credit Union, a custom underwriting model developed with Zest AI helped increase the share of automated personal loan approvals from 3% to 17%, according to a case study published by the vendor. Zest also reported that faster decisioning helped the credit union quadruple indirect loan approvals. The model uses features derived from credit bureau information, including tradelines, inquiries, collections and bankruptcy data.
Suncoast Credit Union offers another example. In a separate Zest AI case study, the credit union reported that 60% of applications were being decided without staff intervention, freeing thousands of staff hours annually. Those figures are vendor-published customer results, rather than independent assessments, but they illustrate how AI is being deployed in operating lending programs.
For lenders, speed is only part of the equation. An underwriting model must also fit the institution’s risk tolerance and lending policies. NCUA guidance emphasizes that credit unions using third-party lending technology should understand the models, explain how they work and evaluate whether the algorithms and underlying data create discriminatory outcomes.
Virtual Assistants Handle Routine Questions
Customer service is another area where smaller institutions are finding practical uses for AI.
American Heritage Credit Union in Philadelphia has introduced Gigi, a virtual associate that handles voice and chat interactions. According to a November 3 Forbes report, the assistant helps members with tasks such as password resets and balance inquiries, with its use expanding to account openings and loan products. Its responses come from a predefined set of answers prepared by credit union employees, and an internal team reviews conversations to keep the information current.
That design highlights an important distinction: deploying AI does not necessarily mean allowing a chatbot to generate unrestricted answers. American Heritage’s approach combines automated interpretation of member questions with institution-prepared responses and continuing employee review.
Other deployments keep the scope narrower still. In an anonymous community-bank case study, provider CustomGPT.ai described a public website assistant grounded in the bank’s published product information, rates and terms. The assistant has no core banking connection and does not receive customer account data, according to the company.
Fraud Prevention And Account Opening
AI-related technology is also supporting identity verification and fraud prevention.
Amerant Bank, based in Coral Gables, Florida, introduced Alloy’s identity and fraud prevention platform to help verify customers, automate account-opening decisions and detect suspicious behavior, according to Forbes. The platform connects institutions with multiple data sources used in identity verification and transaction monitoring.
For smaller institutions, these deployments demonstrate a route to more advanced capabilities through outside providers rather than entirely in-house development. That reliance, however, makes vendor oversight especially important. NCUA identifies data privacy, operational resilience, model risk and the ability to understand algorithmic decisions among the challenges credit unions face when partnering with AI providers.ncua+1
Bringing Borrower Risk Into Focus
For collection departments, a particularly relevant application is using AI to make portfolio information easier to analyze.
Natco Credit Union has implemented Zest AI’s LuLu lending intelligence tool to automate reporting, develop internal compliance-auditing tools and identify at-risk loans, according to the vendor. The case study describes an effort to reduce manual data gathering and give staff quicker access to information used in lending and risk management.
That is different from turning collection decisions over to an autonomous system. The documented application centers on helping employees find and interpret information—not demonstrating that AI independently negotiates payment arrangements or reduces charge-offs. The published case study does not establish those collection outcomes.
New products suggest where the market is heading. On October 28, Kinective announced Cortive, an AI-native banking platform, along with planned tools for deposit retention, lending and cash management. Its Lending Excellence product is designed to combine loan, deposit, payment and household information, flag borrower risk and recommend policy-aligned next steps.
The timing matters: Kinective said the new agentic solutions would become available in the first quarter of 2027. They represent an announced capability, not evidence of completed deployments or demonstrated collection results. The company also said employees would retain control over decisions, including approval of loan modifications.
More Personalized Digital Banking
AI is also changing how institutions present products and information online.
On November 2, Finalytics.ai announced that Travis Credit Union had launched a personalized website experience that adjusts content and next steps based on visitor behavior. Someone researching a first credit card, for example, may receive credit-building guidance, while an auto-loan shopper may see payment information and a path toward preapproval.
According to the announcement, the credit union will measure performance through outcomes such as accounts opened and loans funded. That connects personalization to specific business results, although the announcement did not provide measured improvements from the deployment.
Accountability Remains With The Institution
The expanding use cases do not eliminate existing compliance obligations.
NCUA says it has not issued AI-specific rules or regulations. Instead, examiners assess institutions’ use of the technology through existing expectations covering safety and soundness, applicable laws, internal controls and ongoing risk monitoring. Credit unions are expected to identify AI-related risks, measure them regularly and implement appropriate controls.
For institutions using third-party tools, that includes understanding the risks introduced by the technology rather than treating vendor selection as the end of the review process. Lending applications also require attention to fair lending and the potential for bias in models and data.
For credit and collection professionals, the most useful distinction is between AI that supports an employee and AI that makes or executes a decision. Current examples show institutions using the technology to automate selected approvals, answer bounded service questions and surface portfolio risks. The opportunity is greater speed and better access to information; the continuing requirement is accountability for how those tools affect borrowers and members.






