Why Most Financial Chatbots Are Broken by Design — And What It Actually Takes to Fix Them

Most financial chatbots fail not because NLP is weak, but because their architecture ignores real user behavior, context persistence, failure handling, and compliance risk. Drawing from work at Goldman Sachs and OneMain Financial, Vijay Kumar Sridharan explains what it takes to build production-grade conversational AI systems that scale reliably in high-stakes financial environments.