Part A · Feel it
Chapter 3 of 3 in this part
What changed: from menus to conversation
GoalYou can tell a voice agent apart from a phone menu, a chatbot and an answering service.
For twenty years, “automating the phone” meant menus: press 1, press 2. Callers learned to press 0 and wait for a person. What changed is that a machine can now understand free speech, do the task, and reply within a normal conversational pause.
Four ways to answer a phone
| Phone menu | Chatbot | Answering service | Voice agent | |
|---|---|---|---|---|
| Works on the phone | Yes | No, text only | Yes | Yes |
| Understands free speech | No, keys or keywords | In writing only | Yes | Yes |
| Does the task (books, updates) | Simple tasks | Sometimes | Takes a message | Yes |
| Day and night | Yes | Yes | At extra cost | Yes |
| The caller’s language | “Press 2 for…” | Often | Rarely | Yes, it switches by itself |
The one-second rule
In conversation, people reply to each other after a gap of about 200 ms, in every language studied (Stivers et al., PNAS, 2009). A silence of about one second is the most we tolerate before it feels like a problem (Jefferson, 1989). A voice agent has to reply inside that window, or the conversation breaks.
What the caller notices
- They can say it in their own words: no menu to learn.
- They can interrupt, and the agent stops to listen.
- They get an answer or a booking, not a promise of a callback.
- They are told they are talking to an AI, and still reach a person when they ask.