Meeting Events and Transcript

During every meeting, CoAgentor builds a real-time event log — a record of every moment your agent responded, what triggered it, and what it said. This article explains what gets recorded and how to interpret it.

→ View your meetings →

What gets recorded#

Trigger events#

Every time a trigger rule fires, a meeting event is created with:

• Trigger label — the name of the rule that fired (e.g. "Competitor mentioned")
• Trigger timestamp — when during the meeting the event occurred
• Transcript context — the portion of the transcript that caused the trigger to fire
• Agent response — the full text of the agent's response
• Confidence score — how confident the AI was that the trigger condition was met (0–1)
• Audio — for Live agents, a playback of the spoken response is stored and accessible in the event log

What is NOT recorded#

CoAgentor is designed with strong privacy defaults:

• Raw meeting audio is never stored. The audio pipeline processes the audio stream for transcription only — no recording is kept.
• Full transcripts are not exported. Only the transcript segments that triggered an event are stored as context. The full conversation is not retained after the meeting ends.
• Agent audio for Silent agents is never generated. No ElevenLabs call is made for silent-mode responses.

The only audio stored is the pre-generated agent speech audio for Live agent responses — and only for the triggered events themselves.

The confidence score#

Each trigger event includes a confidence score between 0 and 1, representing how strongly the AI matched the transcript against your trigger rule description.

| Score | Interpretation | |---|---| | 0.9–1.0 | Very strong match — trigger condition clearly present | | 0.7–0.9 | Good match — probable trigger condition | | 0.5–0.7 | Moderate match — may be a borderline case | | < 0.5 | Weak match — possible false positive; consider refining the rule |

If you're seeing many low-confidence triggers, narrow your rule description. If you're seeing none, broaden it.

Viewing a live meeting#

While a meeting is running, open the meeting card to see events appearing in real time. The event log updates as triggers fire — you don't need to refresh.

Viewing a completed meeting#

After a meeting ends, the full event log is available on the meeting detail page. From the Meetings dashboard, click any past meeting to open it.

The detail page shows:

• Meeting metadata (date, duration, platform, agent used, source)
• A chronological event timeline
• Each event expandable to show full transcript context and response
• Agent audio playback (if applicable)

Meeting duration and hours#

Meeting duration is tracked from the moment the bot joins to the moment it leaves. This duration counts against your monthly meeting hours.

If a meeting is marked Failed (the bot never successfully joined), no hours are counted.

→ How hours are counted →

Submitting feedback on a meeting#

After a meeting, you can submit feedback on specific events directly from the meeting detail page. Feedback types include:

• Wrong trigger — this trigger shouldn't have fired
• Missed trigger — a trigger should have fired but didn't
• Praise — the agent's response was excellent
• Bug — something went technically wrong
• Suggestion — a general improvement idea

Feedback is sent to the CoAgentor team and helps improve trigger evaluation accuracy.