• 

  Thumbnail

FlowGate — Psychology-Informed Fraud Prevention

By Chau Le

Inspiration

As an Entrepreneurship student minoring in Data Analytics and Psychology, I wanted to explore how behavioral science could improve financial safety. Romance scams stood out in the synthetic dataset, accounting for 40% of total scam-related losses. Unlike one-time fraud, these scams often involve gradual emotional attachment and escalating financial commitments, creating opportunities for earlier intervention.

What it does

FlowGate detects potentially risky payment patterns and uses psychology-informed interventions to help members reconsider financial decisions. Instead of generic fraud warnings, it uses perspective-taking questions, preserves member autonomy, and offers confidential human support and judgment-free recovery assistance.

How we built it

I analyzed Sound Credit Union's synthetic transaction data using Excel and Tableau, developed a rule-based detection hypothesis, and built an interactive prototype with AI-assisted coding. The prototype demonstrates both member-facing check-ins and employee-facing support workflows.

Challenges we ran into

My biggest challenge was translating behavioral psychology into a practical fintech solution without a technical background. Another was balancing fraud prevention with false positives, privacy, and a frictionless experience for legitimate members.

Accomplishments that we're proud of

Our proposed rule flagged all 12 constructed romance-scam cases while also identifying three legitimate look-alikes. More importantly, FlowGate goes beyond detection by translating psychological principles into actionable, nonjudgmental interventions.

What we learned

Fraud prevention isn't just about recognizing suspicious transactions. It's also about understanding how people make decisions under emotional influence. Effective technology must protect members without undermining their dignity, trust, or autonomy.

What's next for FlowGate by Chau Le

The next step is a 90-day credit union pilot to validate detection rules against real-world data and test whether behavioral interventions outperform standard warnings. Longer-term, FlowGate could scale as an embedded fraud-prevention solution for credit unions and community banks.

Built With

• canva
• claude
• tableau