View all speakers

Matt Trevathan

Chief Artificial Intelligence Officer
Nymbus

Roadblock Title:

How We Stopped AI from Hallucinating Against Our Banking Core

Time:

Tuesday - 2:00 PM (Tower A)

Abstract:

Every financial institution eventually encounters the same problem: critical systems are proprietary, poorly documented, or understood by only a handful of experts.

At Nymbus, our team faced this challenge while working with a proprietary query language used to access core banking data. Traditional AI approaches quickly exposed a familiar problem—models can generate answers that look correct even when they’re completely wrong. In financial systems, that isn’t just inconvenient; it creates operational, reporting, and compliance risk.

In this session, we’ll share how we built an AI-assisted query system designed around a simple principle: models can help reason, but they should never be trusted as the source of truth. Instead of relying on generated answers, we grounded every interaction in system metadata, validated outputs against live environments, and continuously captured verified patterns from expert users.

Attendees will learn practical techniques for:

  • Grounding AI outputs in authoritative system data rather than documentation alone
  • Validating model-generated queries before they reach production systems
  • Capturing and scaling institutional knowledge from domain experts
  • Identifying where AI can safely accelerate work—and where it can amplify mistakes
  • Building AI workflows that prioritize correctness, auditability, and human accountability

Whether you’re working with a banking core, legacy platform, vendor API, or internal system no one fully understands, you’ll leave with practical patterns for making AI useful without sacrificing control or trust.

Bio:

Matt Trevathan is a technology executive and prolific inventor with more than two decades of experience at the forefront of emerging technologies. His work has resulted in 271 patents spanning artificial intelligence, cloud computing, mobile technology, IoT, cybersecurity, and fintech.

As Chief Artificial Intelligence Officer at Nymbus, Matt leads the company’s AI strategy, governance, and innovation initiatives, helping community banks and credit unions responsibly adopt and scale AI. He oversees a team focused on transforming cutting-edge research into practical, production-ready solutions that drive operational efficiency, growth, and customer value.

Prior to Nymbus, Matt spent more than a decade at IBM, serving as Chief Architect for Mobile Solutions and helping pioneer technologies that anticipated industry standards. He later led IoT and Emerging Technologies at Kony (now Temenos) and founded CitizenTrader, a fintech platform focused on market sentiment and risk analysis. Matt is recognized for turning emerging technologies into practical, scalable solutions that drive business innovation.

Matt Trevathan