Voice and omni-channel agents · L2 · 15 min
A latency budget you label as a lesson
Learn in five minutes
- People notice waits in voice more than in chat.
- Lab 08 reports a stage p95 so an average cannot hide the slow leg.
- Any millisecond figure in that lab is a teaching fixture.
- This page states no vendor SLA.
- Cut a stage only after you have measured it.
Add the stage times you measured. The largest stage is the candidate to simplify. Do not invent a cloud region's number.
Visual
Add three toy stages
- Speech-to-text 400 toy ms.
- Model 250 toy ms.
- Text-to-speech 700 toy ms.
- The speech-out stage is the slow one in this fixture.
- Label the numbers as a fixture.
Quiz
Ecosystem
- SLA
- A service promise. This lab does not state one.
Sisters: Web Speech API, Turn-taking, Queues, CRM notes.
The retrieval index and the tool cap are shared. The channel changes the wait and the consent copy.
A later host could add a channel without rewriting the notes server.
Config
- Write the goal in one sentence.
- Name the input and the allowed output.
- Save the trace next to the result.
Troubleshooting
- The answer looks fluent and is wrong. Check whether a source passage actually contains the claim.
- The loop never stops. Set a step cap and stop when the tool result repeats.
Sources and go deeper
- Python 3 tutorial vendor last checked 2026-10
- MDN Web Docs, Web Speech API community last checked 2026-10
- Web Speech API community draft community last checked 2026-10
- Speech synthesis — Wikipedia wikipedia last checked 2026-10