LLM Tracing with Langfuse: Debug and Observe Complex AI Pipelines Locally

In this video, we dive into LLM tracing and observability using Langfuse, one of the most popular tools for understanding what happens inside your LLM-powered applications. You’ll learn how to run Langfuse locally using Docker and use it to trace simple LLM calls, post-processing logic, and multi-step pipelines involving multiple LLM invocations. We cover how tracing works for: Single LLM API calls LLM calls followed by custom Python logic Multi-step pipelines with multiple LLM calls and intermediate outputs You’ll also explore the Langfuse UI to inspect traces, token usage, latency, cos…