How Klarna's AI assistant redefined customer support at scale for 85 million active users

Fintech company Klarna built an AI assistant using LangGraph and LangSmith to handle customer support for its 85 million users. The system functions as a multi-agent architecture to manage tasks like payments, refunds, and escalations, performing the work of 700 full-time employees. By using a controllable agent framework and test-driven development, Klarna created a more reliable and cost-effective solution. This resulted in an 80% reduction in customer resolution time and the automation of about 70% of repetitive support tasks.