Voice Agent
Run tests

Understand a test suite

See how one requested test becomes a group of logical definitions and physical runs.

A test suite is the group created when you select Run Test. It is not a separate mode or button in the current product.

Logical definitions and physical runs

A logical definition describes the metric-driven test you request. A physical run is one concrete call or simulated interaction needed to evaluate it.

The product may create multiple physical runs because:

  • Several selected driving metrics need distinct definitions.
  • Audio Robustness creates clean and noisy variants.
  • Language Bias creates Accent A and Accent B variants.

Compatible metrics can share the same physical run, so the relationship is not always one metric to one call.

Review the planned scope

Before submission, check Test runs needed. The configured default maximum is 40 physical runs per suite, though deployments can set a different limit.

Results are progressive

The suite and its run statuses are available before all scoring finishes. Open the suite from Results to follow progress, then use Recording breakdown to inspect individual runs as they become available. If a suite remains at Starting, use the refresh control to request its latest state.

For variant-specific results, the suite can show Robustness Breakdown and Bias Breakdown when those options were included.

Learn how robustness and bias expand a suite

On this page