Skip to content

Call Forking for Voice AI

Description

Forks the live call audio stream to a voice AI pipeline — speech recognition, sentiment analysis, or real-time transcription — while the primary call continues uninterrupted. Enables real-time agent assist and post-call analytics without call quality impact.

Canonical use case

A contact centre streams the customer-side audio of every live call to a real-time transcription and sentiment engine, surfacing suggestions to the agent as the conversation unfolds.

Open Items

  • [ ] Canon alignment — populate canon_axiom_refs or confirm no existing axiom applies
  • [ ] Dependency assessment — set dependencies_assessed: true once SA has reviewed the full chain
  • [ ] effort_estimate — replace 0 with rough engineering days (order of magnitude)
  • [ ] public_description — write the public-facing description before publishing