Agent Evaluation
The Stratix Premium Agent Evaluation surface — Traces, Judges, and Trace Evaluations.
Last updated
Was this helpful?
⌘Ctrlk
The Stratix Premium Agent Evaluation surface — Traces, Judges, and Trace Evaluations.
Available in Stratix Premium. This surface is part of the logged-in workspace at stratix.layerlens.ai. Stratix Public users can browse the catalog but cannot use this feature.
The Agent Evaluation section in the Premium left rail bundles three surfaces:
Traces — upload, browse, and inspect agent traces
Judges — build and manage LLM-as-a-judge graders
Trace evaluations — score traces with scorers and judges
These three surfaces compose the full agentic-evaluation workflow.
A trace is a record of an AI call (or chain of calls): inputs, outputs, every tool call, every span, latencies, costs, errors. Stratix's trace pipeline ingests trace JSON and stores it for inspection and evaluation.
Inspect individual traces. See Traces.
Build LLM judges for subjective dimensions. See Judges.
Run a scoring config (scorers + judges) over a trace set. See Trace evaluations.
A complete agentic evaluation:
Capture trace set — upload representative traces (or stream them from your application)
Define criteria — natural-language assertions, deterministic rules, LLM judges
Run trace evaluation — Stratix grades every trace against every criterion
Read verdict + root-cause — failure tied back to the trace, span, and decision
Last updated
Was this helpful?
Was this helpful?