Skip to main content
Conversation Review is where you check what actually happened in a call or chat. This page shows transcripts, basic system decisions, and supporting metadata so you can confirm whether the agent behaved as expected. If you cannot explain what happened in a conversation using this page, the agent is not ready to move forward.

​
What Conversation Review is for (Level 1)

At this stage, use Conversation Review to answer simple but critical questions:
Did the agent understand the user? Did it answer from the right Knowledge Base topic? Did it escalate or hand off when expected? Did anything obviously go wrong?
You are not tuning behaviour yet. You are confirming that the basics work.

​
Starting in Conversations

Begin on the Conversations page. This table lists all interactions across Chat and Call. Focus on these columns:
  • Status: Live or Ended
  • Summary: High-level intent label
  • Start time: Helps align with test runs
  • Contact: Phone number or user identifier
  • Duration: Longer calls often signal confusion
  • PolyScore (if enabled): Quick quality indicator
Use Edit table to add helpful columns:
  • Environment
  • Variant
  • Handoff reason
  • Function call
This makes basic issues easier to spot without opening every conversation.

​
Filtering during testing

Use Filter conversations to narrow your view. Common Level 1 filters:
  • Environment = Sandbox
  • Start date = Today
  • Duration > 60 seconds
  • Status = Ended
Example: Filter to Sandbox + Duration > 90s to find conversations where the agent struggled.

​
Opening a conversation

Clicking a row opens Conversation Review. You will see three main areas:
  1. Transcript (centre)
  2. Conversation details (left)
  3. Diagnostic toggles (right)
At Level 1, focus primarily on the transcript and matched topics.

​
Reading the transcript

The transcript shows the conversation turn by turn. For each turn, check:
  • What the user said (ASR text)
  • How the agent responded
  • Which KB topic was matched (shown beneath the turn)
Example: User: “late checkout” Matched topic: late_checkout_policy
This is the expected outcome. If the wrong topic appears, the issue is almost always in KB naming or sample questions.

​
Matched Knowledge Base topics

Matched topics are your primary signal at this level. Use them to confirm:
  • The correct topic was selected
  • Similar topics are not competing
  • The topic name reflects the intent clearly
Example: The agent answers about billing, but the matched topic is general_help. This indicates a KB design problem, not a phrasing issue.

​
Conversation details (left panel)

The left panel confirms context. Check:
  • Environment (Sandbox vs Live)
  • Channel (Chat or Call)
  • Language and locale
  • Duration
  • Variant (if applicable)
This helps ensure you are reviewing the correct test and not a different version of the agent.

​
Basic diagnosis toggles

At Level 1, you only need a small subset of diagnosis tools.

​
Topic citations

Use this to confirm which KB topic was used for each response. If the agent answers correctly but cites the wrong topic, fix the KB before proceeding.

​
Function calls

Use this to confirm:
  • SMS was only sent after consent
  • Handoff occurred when expected
You do not need to inspect parameters in detail yet — only whether the action happened at all.

​
Audio playback (Calls only)

For call channels, use audio playback to:
  • Confirm ASR quality
  • Hear how long responses feel when spoken
  • Check whether users interrupt the agent
Split audio is especially useful for hearing the caller clearly.

​
Annotations

Annotations let you flag issues directly from the review page. At Level 1, use:
  • Wrong transcription → ASR problem
  • Missing topic → KB gap
Annotations help you build a clear list of fixes instead of relying on memory.

​
What “good” looks like at Level 1

A successful review session should allow you to say:
  • The agent selected the correct KB topic
  • The response made sense for the question
  • Escalation behaviour was reasonable
  • Any problems are clearly attributable to KB or ASR
If you cannot clearly state what to fix, stop and simplify before adding complexity.

​
Final check before Level 2

Before moving on:
  • You can explain each response in plain terms
  • You can identify whether an issue is KB or ASR
  • You are confident the agent behaves consistently in Sandbox
Level 2 will focus on control, variation, and optimisation. Conversation Review is the foundation that makes those steps possible.