Observability
- Metrics: webhook delivery rate, latency per endpoint, call statuses, escalation ratio, template success rate.
- Structured logs with correlation (requestId/callId).
Retries
- Webhooks: exponential backoff on 5xx/timeouts.
- Voice campaigns: configurable retry policy (when applicable).
Procedures
- EHR/HIS outage: switch to “informational” mode and retry later.
- WhatsApp issues: pause automation and handle manually.
Backup and retention
- Differentiated retention policies for answers and transcripts.
- Encrypted exports upon request.