The technology
A voice agent is software that holds a spoken conversation and can act on it. It listens, works out what the speaker wants, and then does something in another system: books, looks up, updates or transfers. On a telephone line, the voice agent is what makes an AI phone assistant more than a recording.
Also called: AI voice agent · Voice AI agent · Speech agent
A voice agent runs a loop: speech to text, then a language model working out the intent, then an action in a connected system, then text back to speech. The loop has to close in well under a second or the pause becomes audible. Latency, rather than vocabulary, is what makes a voice agent sound human.
Anywhere there is a microphone and a task: telephone lines, in-browser widgets, apps, kiosks and in-car assistants. The telephone is the hardest of them, because there is no screen to fall back on and the caller cannot see what the agent understood.
Last checked 2026-09-23
Free to start, on your own number, live in about a week.
Get started