Notifications
Clear all
Topic starter
17/07/2026 12:31 am
Hey everyone, new here! 👋 I come from a project management background, so I'm used to tracking tasks and resources, but this LLM observability world is new to me.
We set up tracing for our chat feature, and the volume is overwhelming. Every single user message creates a huge trace with embeddings, LLM calls, tool calls—it's a lot. Our storage costs are climbing, and finding real issues feels like searching for a needle in a haystack. How do you all balance detail with practicality? Is there a common threshold or sampling strategy that works? I'm worried we're logging everything but seeing nothing.