Docs / Voice agent implementation status
PARTIAL — SIMULATION / DRY-RUN ONLY
⏱️ 2 min read · 324 words
Voice agent implementation status
The current repository provides a governed voice-agent control plane, not a deployed autonomous calling product.
| Capability | Current state | Evidence in implementation |
|---|---|---|
| Provider profile | IMPLEMENTED / DRY-RUN | <code>hospitalVoice</code> provider configuration |
| Caller identity | IMPLEMENTED | governed caller profiles |
| Voice selection | IMPLEMENTED | voice configuration with customer-owned values |
| Campaigns | IMPLEMENTED / DRY-RUN | campaign + EHR contract reference |
| Audience manifest | IMPLEMENTED | case-scoped EHR manifest model |
| Approved action | IMPLEMENTED | approved-action request + decision model |
| Call simulation | IMPLEMENTED | deterministic simulation scenarios |
| Human escalation | IMPLEMENTED | escalation/exception handling |
| Kill switch | IMPLEMENTED | global voice-agent simulation kill switch |
| Live telephony provider | NOT IMPLEMENTED / DISABLED | no live provider adapter in current repository |
| Live recording | DISABLED | simulation policy rejects recording |
| Live transcript storage | DISABLED | simulation policy rejects transcript storage |
| Autonomous EHR mutation | NOT PERMITTED | write-back remains governed/disabled |
Why the boundary is explicit
The product documentation deliberately avoids marketing a simulation as a live calling agent. A future live-telephony release would require a real provider adapter, webhook/signature handling, recording/transcript policy, consent controls, retry/idempotency behavior, and separate production validation.