A voice agent is the AI that answers calls on your behalf. You can create one in under five minutes by picking a template, or build one from scratch with full control over its system prompt, voice, language, and connected knowledge.
Where to start
Open Dashboard → Automation → Voice and click Add voice agent. A modal opens with two paths:
- Pick a template — preconfigured agents for common use cases (receptionist, support, sales). The template sets the system prompt, voice, language, and begin message. You can change everything afterwards.
- Generate with AI — describe what the agent should do in a sentence or two, and Yaplet generates a tailored configuration for you.
- Start blank — empty agent, you fill every field.
Whichever path you pick, the agent then opens in the editor with three tabs.
General tab
Defines who the agent is and how it greets a caller.
| Field | What it does |
|---|---|
| Name | Internal label shown in the agent list and in inbox call threads. Pick something your team recognises, like "EN Receptionist · main toll-free". |
| System prompt | The agent's job description. Tell it who it represents, the tone to use, what it can and cannot do, and any hard rules. The clearer this is, the better the agent behaves on real calls. |
| Begin message | The greeting played at the start of every call. Keep it under two sentences. If your jurisdiction requires AI disclosure or recording consent, add that wording here — Yaplet does not auto-play one. |
| Interruptible greeting | When on, the caller can talk over the greeting and the agent will stop and listen. Off by default — the greeting plays in full before the agent starts listening. |
| Recording | When on, the entire call is recorded as an MP3 saved to the inbox thread. See Recordings and transcripts for retention details. |
Write the system prompt like you are onboarding a new hire — clear role, a few hard rules, and explicit examples of when to stick strictly to the knowledge base. The agent does best when you tell it what to do, not just what not to do.
Voice & language tab
| Field | What it does |
|---|---|
| Voice | One of 30 Gemini voices, grouped by family. You can preview every voice with your own begin message before saving. |
| Language | The primary language of the call. The agent replies in this language even if the caller switches mid-call. |
Saving this tab triggers greeting pre-generation — the begin message is converted to audio in the chosen voice and language and cached so the first second of a real call is instant. Changing the voice, language, or begin message regenerates the cached audio automatically.
See Pick a voice and language for the full voice catalog and language list.
Connections tab
| Field | What it does |
|---|---|
| Vex chatbot | The chatbot whose knowledge base, workflow tools, and context tools the voice agent inherits. Required — without a linked Vex, the agent has nothing to look up. |
| Phone numbers | The numbers that route calls to this agent. A number can only be attached to one agent at a time. Attaching a number here detaches it from the previous owner. |
See Connect knowledge to your voice agent for how the Vex link works in practice.
Pausing or deleting an agent
From the agent list, you can pause an agent to stop it from answering calls without losing its configuration. Paused agents do not pick up calls even if a number is routed to them — useful for planned downtime or holiday closures.
Deleting an agent detaches its phone numbers (so they can be reassigned) but does not delete past call transcripts or recordings from the inbox.
Editing the system prompt does not affect calls already in progress — the agent keeps the prompt it started with for the rest of that call. The new prompt takes effect on the next call.
What to do next
Once you have created an agent and attached a number, test it by calling the number yourself. The agent should pick up within a second and play the greeting in the right voice. See Connect knowledge to your voice agent for tuning the answers it gives.