Tracing Agentic Gen AI applications

This session will have a short presentation to introduce the need and concept of tracing for Gen AI apps and will include a demo/hands-on of how to setup and use popular tracing frameworks viz. LlamaTrace, LangSmith, LangFuse. The demo/hands-on will be in a Google Colab notebook so you can follow along live during the event.

Zoom link will be shared on the day of the event.

More info:
Frameworks like LangChain and LlamaIndex provide powerful abstractions for building Gen AI apps with their built-in support for interaction patterns like Chain-of-Thought, ReAct, Self-Ask-With-Search, etc. However, even moderately complex applications can get unwieldy to tweak for acceptable quality and performance without insights into the underlying behavior of these tools.

Gen AI tracing is designed to address this exact need. Tracing makes it possible to have at-a-glance and detailed insights into the functioning of a Gen AI application down to the internals of each unit-of-work like the use of prompt template, LLM, and tools.