Read a conversation

When prompt capture is onon, (seeConversations holds the transcripts.

The list

Every captured conversation with the developer, the tool, an excerpt of the first message, how many messages it has, and when it was last active. Filter by developer.

The transcript

Messages in order, each labelled by role — Developer, Assistant, or System prompt — with timestamps. A "truncated" badge means the content exceeded what is stored for a single message.

Seeing the system prompt is often the most instructive part: it shows what the tool told the model before your developer typed anything.

Redaction markers

Amber shield badges naming a kind of secret mark where redaction happened. The transcript contains [REDACTED:aws_access_key_id] rather than the key, because redaction runs before storage — the value was never written down.

Two different empty states

"Prompt capture underis Policy)off", with a link to Policy. Nothing is wrong; the feature is not on. "No conversations captured yet" — capture is on, nothing has come through.

Telling these apart is deliberate. An admin who sees a blank list usually assumes a bug.

What support can see

Nothing here. Sentilai support can see that a request whosehappened and what the policy decided; conversation content wasis storednot showsreadable aby small conversation icon on Activity. Click the row to read the exchange.us.

Conversations lists captured conversations per developer, newest first, and can be filtered to one person.

What you see

The messages as sent, with detected secrets already redacted. Content follows your retention period and disappears with it.

If capture is off — the default — this screen explains that and links you to Policy rather than showing an empty table.