Skip to content
Pondara Pondara docs

Review a dialogue

The dialogue review shows the full hybrid dialogue of an engagement: every turn the learner spoke, every turn the AI actor spoke in role, the reflection answers, and the side notes the AI added during extraction. You will land here from a verdict drawer (“open full dialogue”), from the assignments list, or from the learner detail page.

Before you begin

  • You are signed in as a trainer.
  • You have an engagement to read — typically because you are about to render a verdict, or because you want to coach the learner afterwards.

What you see

The page opens on the transcript:

  • Learner turns — what your learner said, in their own words.
  • Actor turns — what the AI said in role as the customer, the patient, the supervisor, the difficult colleague. The AI never breaks role here.
  • Reflection answers — the learner’s responses to the reflection prompts at the end of the engagement.
  • Side notes — short observations the AI made during signal extraction, with the dimension they refer to. These are not visible to the learner; they are working notes for you.

The right rail summarises the engagement: the scenario, the field, the dimensions exercised, the AI signals already extracted, and the status of each (pending verdict, your verdict if rendered).

Reading for verdict

The verdict drawer shows a snippet — usually a single turn or a short window. It is a window, not the whole record. When the snippet’s nuance feels off or you remember more context than the snippet shows, click “open full dialogue” and read around the snippet. Then return to the verdict drawer and render. See render a verdict.

Reading for coaching

When you read for coaching rather than for verdict, look for:

  • Moments of decision. Where did the learner shift direction? What did they say or hear right before the shift?
  • Moments of stall. Where did the dialogue grind? Was it hesitation, a missing fact, a difficult emotion?
  • Language patterns. What did they default to under pressure — questions, statements, deflection, naming the elephant?
  • The reflection. How does the learner’s own reading of the engagement compare with what the transcript actually shows?

The dialogue review is the richest material you have for a coaching conversation. The radar tells you what moved; the dialogue tells you how and why.

When something goes wrong

  • A turn is missing. Pondara saves on every turn; missing turns are unusual. Refresh; if still missing, capture the engagement id and open a ticket.
  • The actor sounds out of character. Note it for the field expert who authored the scenario — this is the signal that tunes actor briefs. The AI is not infallible in role.
  • Side notes contradict the AI signal. That is a useful signal to surface in your verdict — adjust or reject accordingly, and the disagreement loop carries it back to the model.