Transcript Notes
What this chapter covers.
Telethryve 2.0 turns voice into a real command surface.
After the phone is paired, the user can open the voice card and speak naturally instead of typing every instruction. The desktop bridge receives the live audio stream, buffers it on the Mac, waits for a natural pause, and hands the turn to the local speech stack for transcription.
From there, the transcript enters the normal Telethryve work loop: queue the request, continue an active run, or start a new Codex task. Replies can come back as text, direct desktop audio, or a short lived audio package for the live paired phone.
The important part is the shape of the workflow. Voice is not a novelty layer.
It becomes a hands free way to steer real Codex work from the mobile app while the workstation does the execution.