Case Study: Khan Academy achieves fast, flexible LLM observability with Langfuse

Khan Academy uses Langfuse’s LLM Engineering platform to build Khanmigo AI

Khan Academy, a nonprofit educational platform serving over 150 million learners, faced a challenge integrating AI observability into its primarily Go-based infrastructure. Most LLM tracing tools required specific SDKs that were not compatible with their architecture, creating a roadblock for their flagship AI product, Khanmigo. They needed a solution that would provide deep observability without compromising their technical stack.

Langfuse provided a solution through its open API, which allowed Khan Academy's engineering team to build a custom Go client for seamless integration. This gave multiple teams across the organization a unified platform for tracing and debugging AI interactions. The results included maintained stack flexibility, increased development speed, and the avoidance of engineering overhead from self-hosting. Langfuse became a fundamental tool for rapid iteration and collaborative debugging across the entire product.

Khan Academy

Walt Wells

Staff Software Engineer / ML & Data Infrastructure