Case Study: Flipkart improves customer support and AI traceability with Arize AI

How Flipkart Leverages Generative AI for 600 Million Users

Flipkart, a massive e-commerce platform serving close to 600 million registered users, needed a better way to scale customer support and improve conversational commerce with generative AI. As the company explored GenAI use cases, it found that some customer-facing scenarios, especially for highly anxious users, still required a human touch. Flipkart turned to Arize AI to help measure and connect model performance with product outcomes, using observability and traceability to understand where its AI systems were succeeding or failing.

With Arize AI, Flipkart was able to stitch together LLM metrics, data science metrics, and product metrics to create a clearer performance storyline and identify the right levers to improve the experience. The company shifted from a direct customer-support approach to an agent-assist solution called AgentServe, empowering support agents with AI-generated assistance instead of replacing human interaction. This helped reduce call times and drive incremental GMV impact, while Arize AI provided the traceability needed to tune modeling and policy decisions more effectively.

Flipkart

Anusua (Anu) Trivedi

Head of Applied AI