Agent Studio

Agent Studio is PolyAI’s visual interface for building, deploying, and monitoring voice agents. Configure knowledge, voice, integrations, and conversation flows. Most agents run on Raven, PolyAI’s proprietary LLM, though other models are also supported.

If you are a developer looking to extend your agent with Python functions and API integrations, see Extend with code instead.

What you can do

Create agents

Set up new agents with a name, language, voice, and greeting in minutes

Add knowledge

Build your agent’s knowledge with topics, FAQs, and connected data sources

Design conversations

Create multi-step flows for bookings, forms, and structured tasks

Configure voice

Choose from a library of voices and fine-tune speech settings

Deploy and monitor

Promote agents through environments and track performance with dashboards

Connect integrations

Link to telephony, CRM, and third-party services without writing code

Get started

Build your first agent

Quickstart guide – create an account, build an agent, add knowledge, test, and deploy.

Learn the platform

PolyAcademy – structured training from basic setup to advanced configuration.

Go deeper

Explore the sections below for full configuration options.

Navigate the interface

Agent Studio has five navigation sections in the left sidebar.

Some sections are permission-gated or feature-flagged. If a section is missing, check your role and permissions.

Analytics

Review conversation data and track agent performance.

Sub-section	What it does
Smart analyst	Ask questions about your agent’s performance in natural language
Conversations	Browse call transcripts, listen to recordings, and review individual interactions
Agent analysis	View performance metrics, trends, and aggregate insights

Build

Create and configure your agent’s behavior, knowledge, and conversation logic.

Sub-section	What it does
Agent	Set your agent’s personality, greeting, model, and behavioral rules
Knowledge	Add managed topics and connect external data sources
Flows	Design multi-step conversation workflows
Tools	Write Python functions and manage code-level integrations
Variant management	Configure multi-site or multi-brand variations
Call handoffs	Set up transfer destinations and routing rules
SMS	Configure text messaging actions
Real time configuration	Manage runtime configuration values per environment
Test suite	Run automated tests against your agent

Channels

Voice and chat channel settings. Voice

Sub-section	What it does
Agent Voice	Choose and customize your agent’s voice, including TTS settings
Response control	Manage stop keywords and pronunciation overrides
Audio management	Configure hold music, IVR prompts, and other audio
Speech recognition	Tune how your agent interprets spoken input
Voice configuration	Advanced voice channel settings

Chat

Sub-section	What it does
Chat configuration	Configure chat-specific agent behavior
Widget	Set up and customize the web chat widget

Configure

Manage project settings, phone numbers, integrations, and operational tooling.

Sub-section	What it does
General	Project-wide settings and preferences
Numbers	Buy, manage, and configure phone numbers
APIs	Manage API connections and configurations
Integrations	Connect third-party services (CRM, telephony, etc.)
Metrics	Define custom metrics for tracking
Dashboards	Manage analytics dashboard configurations
CSAT	Configure customer satisfaction surveys

Deployments

Promote your agent through environments and track version history.

Sub-section	What it does
Environments	Move your agent from Sandbox to Pre-release to Live
Project history	View past versions, compare changes, and roll back if needed

Build your agent

Agent settings

Configure your agent’s personality, greeting, model, and behavioral rules

Knowledge

Create managed topics, connect external data sources, and set up actions

Flows

Multi-step conversation workflows

Voice

Text-to-speech, voice selection, and audio settings

Variant management

Multi-site and multi-brand configurations

Integrations

Connect telephony providers, CRMs, and third-party services

Next steps

Quickstart

Build your first agent in minutes