Trius Federal Credit Union automates 20% of lending decisions and cuts delinquency rate by 42% with Zest AI

Trius Federal Credit Union, a $130M-asset institution serving roughly 16,000 members, faced two compounding pressures common to smaller credit unions: a delinquency rate of 1.72% that eroded portfolio quality, and a fully manual underwriting process that slowed loan decisions and created operational drag. With no loan application automation in place, staff handled every credit decision by hand, limiting throughput and making it difficult to offer the same-day service that larger banks and credit unions could deliver. In a competitive local lending market, the gap in decisioning speed and accuracy posed a direct threat to member retention and growth.

Trius FCU partnered with Zest AI to deploy AI-powered automated underwriting across auto loans, personal loans, and credit cards. The implementation centered on machine learning models that evaluate creditworthiness using a broader set of predictive features than traditional scorecards, enabling more accurate risk differentiation at the point of decision. Integration was handled directly with Trius's existing loan origination system, Origence, allowing the AI layer to slot into established workflows without rebuilding core infrastructure. Zest AI provided hands-on implementation support throughout, and the full integration was completed in three months. Trius also began adopting LuLu, Zest AI's generative AI lending intelligence tool, giving leadership instant access to compliance data, industry benchmarks, and peer comparisons to inform strategy.

Within 14 months of going live, Trius FCU achieved measurable gains across both portfolio quality and operational efficiency:

• Delinquency rate on Zest AI-decisioned loans fell from 1.72% to 1% — a significant reduction
• 20% of all consumer lending decisions are now fully automated, enabling same-day outcomes for eligible applicants
• Integration completed in 3 months, delivering rapid time-to-value without a prolonged implementation cycle

The automation gains allowed the underwriting team to handle higher application volumes without adding headcount, and surfaced cross-sell opportunities at the point of decision. Amy Demkey, SVP of Lending, noted the technology effectively lets a small institution operate with the capabilities of a much larger one.

• Small credit unions with under $1B in assets can adopt enterprise-grade AI underwriting without extensive technical infrastructure, provided the vendor supports LOS integration directly.
• A three-month integration timeline is achievable when AI tooling is designed to layer onto existing systems rather than replace them — minimizing disruption during rollout.
• Delinquency reduction and automation gains are not a tradeoff; the same ML models that speed decisions also improve credit quality by identifying risk signals traditional scorecards miss.
• Extending AI adoption incrementally — starting with underwriting, then adding fraud detection and generative AI tools — allows teams to build confidence and expand use cases without overextending at launch.