configure-metrics

To set up evaluation metrics for AI agents, you can use guided metric selection to attach built-in or custom criteria directly to agent workflows, automating performance monitoring across all future runs.

Yes, you can create custom LLM-based metrics for monitoring using optimized prompt templates, alongside audio-based and pause-detection criteria, to evaluate specific agent behaviors and performance.

Standardizing AI agent performance monitoring involves attaching selected metrics as defaults to your agents, ensuring consistent evaluation logic is automatically applied across all subsequent interactions and runs.

Yes, audio-based evaluation criteria are fully supported, allowing you to configure specific metrics to monitor and assess voice interactions and audio responses from your AI agents.

You can add pause-detection metrics to a customer support agent by creating custom criteria with optimized templates, which automatically integrates the evaluation logic to measure conversational pauses.

Configuring evaluation criteria solves the inconsistency and manual overhead of performance tracking, ensuring your AI agents are measured accurately and systematically during operational workflows.