Skip to main content
Cartesia scores completed calls with built-in metrics and custom metrics assigned to the agent. Use built-in metrics for runtime health and custom metrics for outcomes specific to your application. Through the API, you can create custom metrics, add them to or remove them from an agent, and list or export results. You can also do this via the Playground UI as shown below.

Built-in metrics

Cartesia calculates these metrics for every call:

Custom metrics

Create an LLM-as-a-judge metric in the Playground, or with POST /agents/metrics, using a name and evaluation prompt. The prompt should define one concrete question and the expected output. Create metrics under Metrics in the sidebar. Before saving, the Playground can run the prompt on an agent’s five most recent calls, or on call IDs you enter. Metrics can’t be edited after they’re created. To change one, delete it and create a new one.
Metric names must be unique within your organization. They may contain lowercase letters, numbers, hyphens, underscores, and periods.

Assign a metric

Assign metrics from the agent page in the Playground, or with POST /agents/{agent_id}/metrics/{metric_id}. Cartesia evaluates calls that end after you assign the metric. When adding one in the Playground, you can check Run metric on the latest 5 calls to score recent calls once.
Add metric dialog in the Playground with the option to run the metric on the latest 5 calls

Add a metric from the agent's Metrics page.

See Results to retrieve and export metric results.