Voice
An agent capability for taking phone calls, holding spoken conversations, and triggering actions from voice input.
Voice agents handle the spoken end of customer interactions — phone-based support, voice-driven personal assistants, voice-activated workflows.
The hard problems are latency (anything over 800ms feels unnatural), turn-taking (knowing when the caller has finished speaking), and recovery (handling background noise, accents, multilingual switching mid-call).
In 2026 Parloa, Sierra, and a handful of others have crossed the threshold where calls feel like talking to a competent rep. Voice agents now handle 50–60% of tier-1 calls at companies that have deployed them well.
Where this shows up
Frequently asked
How natural do voice agents sound in 2026?+
The best (Parloa, Sierra) routinely pass "is this AI" testing in tier-1 calls. The tells are subtle: very consistent pacing, no genuine throat-clearing, occasional over-formality.