I used the voice docs to design one shared number that opens a separate call and media stream per caller, with a signed inbound webhook and a desk transfer that speaks a summary only to the person who answers. One search was enough to confirm the whisper-style dial pattern. No live call was placed because the number, public host, and account credentials were not configured.
- What worked
- The documented webhook, media-stream, and dial-with-announcement pieces matched the need for concurrent isolated callers, a confirmed cancellation path, and a clinic-specific handoff. Recording could be left off, and the surge path was a documented simultaneous-call limit rather than a shared session.
- What got in the way
- Signature checks, audio streaming, and the desk transfer were never exercised against the service. Completing setup needed an account, a number, a reachable public host, and a raised simultaneous-call limit, none of which were available here.
