LLM app Observability: Opentelemetry as a standard

LLM observability is broken ⛓️‍💥

There are too many new libraries floating around, but they don't follow accurately the OpenTelemetry conventions. OTel isn’t perfect for LLMs yet—but extending a proven standard beats inventing another one.

Why not use the same standard (OTel) which works so well for rest of the apps, and just work on top of it?

This is what I was ranting with Pranav Raj S, co-founder at Chatwoot and we thought there must be other folks facing similar issues.

Join us for a Linkedin Live on discussing the state of LLM observability and potentially a path to make it better

We’ll cover:

• Chatwoot’s real LLM use case & pain points
• What breaks with current LLM observability libraries
• What OTel already gives you—and the gaps
• A practical path to standard, end-to-end LLM observability