Use Jev as your detector judge

TraceRoot detectors use a judge model to screen your agent's traces, flag the ones that went wrong, and alert you with a root-cause investigation.

Detectors can now be powered by Jev, TypeSafe AI's decision model, using your own API key. Jev evaluates each trace and names the most likely problem instead of returning a plain yes or no.

Animated setup walkthrough for using Jev as a detector judge.

Add TypeSafe AI in Workspace Settings, then select jev-1.13.0 as the detector model in your detector's settings.

Start tracing your agent and use Jev as your judge in TraceRoot today.