Command Center
Real-time production health for VoiceOps AI. Is the agent working? Are customers completing calls? Where is latency?
Active calls
247
+12 vs 5m ago
Task success (24h)
91.4%
-1.2pp vs yesterday
P95 end-to-end latency
1.84 s
+180ms after last deploy
Open incidents
2
18 failed calls · last hour
Call volume & failures
Last 24 hours · all production agents
Success rate
Hourly task completion
Latency by stage (P95)
Find where time is spent: carrier · SIP · ASR · LLM · TTS · workflow · API · DB
Agent health
Across vendors, languages, regions
| Agent | Vendor | Status | Success | Latency | Quality |
|---|---|---|---|---|---|
Collections Agent Collections · en-US · us-east-1 | LiveKit + Custom | healthy | 94.2% | 1.18 s | 91 |
Support Concierge Customer Support · en-US · us-west-2 | Vapi | degraded | 87.5% | 1.92 s | 82 |
Appointment Scheduler Scheduling · es-MX · us-east-1 | Retell AI | healthy | 96.1% | 980 ms | 94 |
Banking IVR Assist Banking · en-GB · eu-west-1 | Amazon Connect + Lex | critical | 71.4% | 2.65 s | 68 |
Outbound Survey Bot Surveys · en-US · us-central-1 | Twilio + Pipecat | healthy | 92.8% | 1.05 s | 89 |
Elevated LLM latency on Support Concierge
P95 LLM turn latency jumped from 900ms to 2.4s after deploy 2026-08-03. Tool-call retries correlated.
Banking IVR ASR confidence collapse (en-GB)
Post-deploy speech model swap reduced account-number recognition. Escalation rate +38%.
Telnyx SIP 503 burst — US East
Carrier-side 503s for 22 minutes. Failover to Twilio trunk engaged.