May 17
Loading video player...
In this video, we walk through how to add tracing and observability to an LLM-powered application using LangSmith. AI agents don't fail with clean error messages, they wander, and standard logs rarely tell you why. We show how to close that observability gap by configuring LangSmith from scratch, instrumenting a simple OpenAI-based app, and using the LangSmith dashboard to inspect traces and monitor production health. You'll learn how to: - Understand the observability gap unique to LLM applications - Configure LangSmith using environment variables in a .env file - Get automatic tracing on an OpenAI client with no extra instrumentation code - Run a traced application and inspect each step in the LangSmith UI - Use the Monitoring dashboard to track latency, success rates, and token costs - Catch runaway agents before they burn through credits in production Timestamps: 0:00 - The observability gap in LLM applications 0:35 - How LangSmith closes the gap 1:06 - Environment setup and tracing variables 1:22 - Walking through the main script 1:54 - Installing dependencies and running the app 2:16 - Viewing the trace in LangSmith 2:20 - The Monitoring dashboard and production health This video is for developers and ML engineers building agentic systems or LLM-powered applications who want reliable tracing and observability without writing custom instrumentation. Clyep produces technical videos for complex software products, including product demos, developer tutorials, release videos, and technical explainers. Learn more: https://clyep.io/ If you found this useful, subscribe for more technical walkthroughs and explainers.