Feature request to the LiveKit team - I think the new agent observability platform is fantastic, but it would be great if there was a way to tag issues as we identify them . Right now, the system makes it very hard to act on.

Hi, what sort of issues were you thinking about tagging? And what did you want the effect of that tag to be? If you could maybe walk through en example? I just want to make sure what I'm picturing in my head matches the request.

Here is the use case: Currently, we are able to look at the transcript traces and logs view along with the audio file to be able to go through the entire session and identify hiccups. When we do identify a particular hiccup (e.g. an STT accuracy issue in a particular scenario, an LLM error, or even the SFX not being played from our backend), I would like to tag those issues within the timeline of the call and then aggregate those issues through some statistical coding (like axial codes). This way, if we review say 50 sessions, we can pinpoint what the top 10 issues are and also identify these sessions that correlate to those 10 issues at a timestamp level so that we can snipe out the issues. Let me know if this is clear