Wired to Win: Outtakes from Tearsheet’s AI Innovation Awards

A few years ago, “AI in banking” meant fraud models running quietly in the background and a chatbot that could reset your password. Now it’s resolving a dispute end-to-end, reasoning through a lending decision, carrying on a conversation that actually gets something done.

This evolution happened bank by bank, fintech by fintech, each one solving a narrow, painful problem – a flood of dispute emails, a fraud model that couldn’t keep up, a customer assistant that kept routing people to a human. The companies that got it rebuilt workflows to make AI native to their infrastructure. 

• Fifth Third‘s AI assistant, Jeanie, now resolves 42% of customer requests without a human – up from 3% – while cutting cost per interaction by two-thirds.
• Casap turned dispute resolution, one of banking’s most painful back-office processes, into an automated workflow that cut resolution times by 87% for one credit union.
• Plaid rebuilt its data models to read transaction sequences like a story, improving fraud detection and underwriting accuracy across its network.
• Titan trained its AI specifically on banking regulation and examiner expectations. Since launching in October 2025, it’s tripled its revenue, with compliance officers preferring its responses over ChatGPT, Gemini, and Opus more than 70% of the time.

The full report breaks down 4 case studies on what separates AI that works from AI that just launches.

Download now