Monitoring & Activity
Each voice agent has an Activity tab that shows real-call history. Use it to verify the agent is doing the right thing, debug unexpected behaviour, and follow up on captured leads.
Activity tab
On the agent edit page → Activity tab. You see a list of recent calls (most recent first):
- Date + time
- Duration
- Outcome label (lead captured / callback scheduled / transferred / ticket created / etc.)
- End reason (caller_hangup / wallet_empty / agent_error / etc.)
Click any row to expand:
- Every tool the agent fired during the call, in chronological order
- Result message for each tool (e.g. “Saved Sayan as a lead”)
- Error indicator (red dot) for any tool that failed
- Per-tool duration in milliseconds
Post-call summary
Within ~30 seconds of hangup, the agent generates a 3-sentence summary of the call:
- Why the caller called
- What the agent did (which tools it used)
- Outcome / next step
The summary appears on the call row in the Activity tab. Useful for quickly catching up on calls without reading the full transcript.
Where captured leads land
The capture_lead tool writes the caller’s name + phone + intent
straight to your Contacts with the voice-ai-lead tag. Open the
Contacts page → filter by tag voice-ai-lead to see every lead
the AI captured.
Where scheduled callbacks land
The schedule_callback tool writes to a callback queue. A background worker fires a WhatsApp reminder to the caller at the scheduled time. You’ll see callbacks in your dashboard’s Activity feed and on the agent’s Activity tab.
Where bookings land
The create_booking tool (booking-template agents only) writes appointment slots that the agent then offers via list_available_slots. Future agent calls won’t double-book the same slot — already-booked windows get filtered out.
Where tickets land
The create_ticket tool writes a row to your Support inbox
with category voice_ai. Open Support → filter by category to see
voice-AI-created tickets.
Spend tracking
Per-call cost rows land in Settings → Track Expenses alongside
WhatsApp / email / RCS expenses. Each voice call shows up as one
row labeled Voice AI call - <duration>.
Live dashboard
Coming soon: a per-agent live dashboard showing:
- Calls today / week / month
- Avg duration trend
- Top tool fires
- Captured leads count
- Cost-to-revenue chart
For now, the Activity tab + Track Expenses cover the day-to-day visibility most customers need.
Common workflows
Following up on captured leads
- Contacts → filter by tag
voice-ai-lead - Sort by created date desc — newest leads first
- Open each contact → see the call’s tool fires + the original transcript timestamp on Activity
Auditing an agent’s behaviour
- Agent edit page → Activity tab
- Pick a call where the outcome looks wrong
- Expand the row → review every tool fire + the result message
- If the agent mishandled something, tweak the system prompt on the Persona tab so future calls behave differently
Reviewing wallet drain
- Settings → Track Expenses
- Filter category to “Voice AI”
- See every call’s cost + duration. Anomalies (very long calls, very high spend) point at agents talking too much — tighten the system prompt to keep responses 1-2 sentences.