At a glance
1
active call per line
5–10s
typical persona initialization
Fail closed
when readiness expires
Operational limits
- Concurrency: one active call per line. A second caller receives busy. Multi-line scaling is the top capacity roadmap item.
- Answer latency: pickup within seconds; persona speaking roughly 5–10 seconds later. Video shows a holding screen.
- Readiness: workers continuously verify the media path. Evidence expires and must be renewed; until then the line declines calls.
- Idempotency: POST requests accept
Idempotency-Key. - Rate limits: generous project defaults; a limit returns 429 with
Retry-After. - Data: call records contain aggregates and lifecycle events. Transcripts are consent-gated and authenticated. Media is not recorded.