Create Workflow
This is where you actually build an agent. It’s organized into tabs across the top.
Configuration tab
Set the workflow’s identity and voice:
| Field | What it does |
|---|---|
| Name / Description | Internal labels shown on the workflow’s card. |
| Primary language | The language the agent speaks and listens in. |
| Call timezone | Used to resolve any relative time references during the call. |
| Voice | Pick the voice your agent speaks with, from the voices available on your account. |
| Advanced (behind “Show advanced”) | Speech-to-text/text-to-speech/language-model selection, and workflow parameters like barge-in behavior and silence timeout. |
The exact list of voices and AI models available depends on your account and is shown live in the picker — it isn’t a fixed list, so check the in-app selector for what’s currently available to you.
The other tabs
Each of the remaining tabs has its own dedicated page, since getting them right is most of the work of building a good agent:
| Tab | What it’s for | Page |
|---|---|---|
| Builder | The plain-language system prompt (max 10,000 characters), Prompt Library, Generate with AI | Writing the Prompt, Pronunciation |
| Properties | What data the agent captures from the call | Properties |
| Integration | What runs automatically after the call — Email, Webhook, Telegram, Push, Result Generation, plus attached integrations | Follow-up |
| IO (Outbound Workflows only) | Mapping contact-list columns to input variables, and defining output values per contact | Outbound Workflow |
Once your agent is built, continue to Deploying to put it on a phone number or web widget, then Test before you rely on it for real customers.