Evals
Use evals as quality gates before changes reach production—datasets, suites, and comparisons tied to the same runs and traces as policy enforcement (what was allowed) vs. output quality (what was good enough).
How Quantlix fits together
End-to-end path for your first run
DeploymentPolicyRequestModelTrace / logsQuality
Start measuring quality
Create a dataset of inputs (and optional expected outputs), attach evaluators with thresholds, then run a suite on a workflow or deployment. Use comparisons to spot drift between releases.
Eval workspace
Jump to a section