Voice Interface

Reading is a full-attention activity. If the only way to know what an autonomous system is doing is to look at a screen, it can only work while you are watching it — which defeats the point of it being autonomous.

• The spoken block is a status headline, never a narration of the written answer — the screen already has the detail.
• Deterministically truncated, so a long answer never becomes a long monologue.
• A local speech engine is the default; a premium cloud voice is opt-in rather than the standard path.
• Speech is suppressed entirely when nobody could hear it, so tokens are not spent on an empty room.

Hands-free status is what lets the system keep working while its owner is doing something else — which is the whole design goal.