Build an agent, then call it
Voice, model, prompt, tools, knowledge and telephony on one screen. Edit a draft, test it in the browser, publish when it sounds right.
One screen for the whole agent
Switch the engine and the model settings change with it. The test call is how every agent gets checked before it goes live.
You are the front desk for Acme Dental. Greet {{caller_name}} by name and look up {{user_number}} before you ask anything.
It is {{current_time_America/New_York}}. If the office is closed, offer to book a callback.
Use book_appointment to schedule. Never quote prices.
From draft to live calls
The loop every agent goes through, without a deploy pipeline.
- 01
Configure
Pick the engine, voice and model from catalogs filtered by provider, language and type. Write the prompt and attach tools.
- 02
Test in the browser
Push-to-talk or open mic, with test variables that stand in for real call data. Every tool call shows up as it happens.
- 03
Publish
Drafts never touch live calls. Publishing is one click, and the next call uses the new version.
- 04
Watch real calls
Transcripts, latency per segment and post-call analysis show what to change next.
Also in the builder
One place to configure
Voice, model, prompt, tools, knowledge and telephony live on the same screen.
- Voices by provider, accent and age
- Models by type
- STT by language
Draft and publish
Edit freely without affecting live agents, then publish with one click.
Dynamic prompts
Template variables are filled in when the call starts, from call metadata or earlier tool results.
- {{caller_name}}
- {{user_number}}
- {{current_time_…}}
Browser testing
Push-to-talk or open mic, with custom variables and every tool call visible.
Voice preview
Listen to a sample in the voice picker before choosing.
Retell migration
An import wizard maps existing Retell agents, prompts, tools, voices and variables.
See it with your agent
Want a walkthrough of the builder with your own use case? Tell us what the agent should do.