Conversation Arena

Conversation

Client ID • ID

esc

Reset to Factory Settings?

This will restore all settings to their default values.

Unsaved Changes

You have unsaved changes in Settings. Would you like to save them before leaving?

Delete 0 Conversation(s)?

This will remove the selected conversations from the list.

Create Dataset

Add Conversations

Client Conversation Protocol Date Score Reviewed

Delete Profile

Are you sure you want to delete ? This cannot be undone.

Overwrite Profile

This will replace the saved settings in with your current configuration. This cannot be undone.

Are you sure?

This action cannot be undone.

Identity
Unique identifier (snake_case). Used in JSON output. Cannot be changed for built-in evaluators.
Human-readable name shown in the UI and reports.
Brief explanation shown below the evaluator name.
Generate a rubric draft
AI

Describe what you want to evaluate. The AI will draft a rubric and suggested output fields. Review and edit before saving.

This will overwrite the rubric and output fields in the editor.
Classification
Evaluators are grouped by category in the list.
Global evaluators run for every conversation. Protocol-specific evaluators only run for selected protocols.
Time Window Restrict evaluation to specific call portions
Evaluate only the first N seconds.
Evaluate only the last N seconds.
Advanced Options Meta evaluator setting
Meta evaluators appear at the root level of the JSON output (outside the "evaluators" object).
Define when each rating should be assigned. These definitions guide the AI model in scoring this evaluator. Be specific and provide clear examples.
Pass Score: 4
Fully met, no issues detected.
Minor Issue Score: 3
Small issues with minimal impact on the call.
Major Issue Score: 2
Significant issues affecting quality or outcomes.
Critical Score: 1
Severe issues that could cause harm or require immediate attention.
Unknown Score: null
When assessment is not possible due to audio quality or missing context.
Not Applicable Score: null
When the situation evaluated by this evaluator did not occur in the conversation.
Output fields define the JSON structure returned by the AI for this evaluator. The required fields (rating, evidence, rationale) cannot be removed. You can add custom fields to capture additional data.
Test this evaluator with sample conversation text. Paste a transcript snippet below and click "Run Test" to see how this evaluator would score it.
Enter a sample conversation to test. Use "LOLA:" and "Patient:" prefixes for speaker labels.

No test results yet

Enter sample text and click "Run Test" to see how this evaluator scores it


            

Create Profile

This profile will save a snapshot of your current configuration:
Tip: Drag categories to reorder them. The order determines how categories appear in the evaluators list and compiled prompt.

Edit Category

Unique identifier. Cannot be changed for built-in categories.
Human-readable name shown in the UI.
1

Define Search Criteria

Filter conversations to import from BigQuery

Enter Client IDs, Protocol IDs, or both. Protocol-only search queries all clients.
to
2

Select Conversations

Choose which conversations to import

3

Importing...

Downloading audio files