Get started
Introduction
Build voice agents that people reach by phone or from your website.
What is Waylo Voice?
Waylo Voice is a platform for voice agents. You set up an agent once (what it knows, how it sounds, what it can do), and Waylo runs every call: listening, thinking and speaking in real time.
Voice agents can:
- Answer calls on a phone number.
- Talk to visitors from a button on your website.
- Call your APIs and search your documents during a call.
- Leave a transcript of every call.
How a call works
Every turn of the conversation goes through three steps:
Transcriber
Speech-to-text
Turns the caller's voice into text.
Model
Language model
Writes the reply, and can call your tools and search your documents.
Voice
Text-to-speech
Says the reply in the agent's voice.
You choose the transcriber, model and voice per agent. Use Waylo's provider accounts, or add your own provider keys.
What you work with
| Agents | The instructions, voice and models. Edit a draft, publish it, roll back to an earlier version. |
| Tools | Your APIs, called by the agent during a call. |
| Documents | Files the agent searches to answer questions. |
| Channels | How people reach an agent: a phone number, or the Web SDK on your site. |
| Sessions | Every call, with its transcript. |
| Webhooks | Call events sent to your systems. |
You set all of this up in the Waylo dashboard, from your servers with the API, or from an AI agent through the MCP server.