Skip to main content
PolyAI platform documentation PolyAI builds AI agents for real customer conversations — voice and chat. Agents answer questions, make bookings, resolve issues, and hand off to humans when needed. Most run on Raven, PolyAI’s proprietary LLM with sub-300ms latency across 24+ languages. Build, deploy, and monitor agents through Agent Studio, the ADK, or REST APIs.

Where do you want to start?

I'm exploring PolyAI

See the three surfaces and how they connect.

I'm setting up an agent

You have Studio access and want to configure knowledge, voice, and analytics through the UI.

I'm maintaining an agent

Make updates to a live agent – edit knowledge, voice, closures, and more.
You’re in simplified mode. Developer, API, and advanced docs are hidden. Click the Simplified — exit pill in the top-right corner (or visit the simplified-mode page) to switch back to full docs.

Get started

Quickstart

Build your first agent in minutes – no code required

PolyAcademy

Hands-on training for production agents

Build your agent

Set up your agent’s personality, knowledge, voice, and integrations.

Agent settings

Configure personality and rules

Knowledge

Create and manage your agent’s knowledge

Voice

Customize text-to-speech and audio

Chat

Deploy a webchat version of your agent

Integrations ecosystem

Connect your agent to telephony, CRM, payments, and knowledge platforms.

Telephony

Five9, NICE CXone, Twilio, Amazon Connect, Genesys, Dialpad, and more

CRM and ticketing

Salesforce, ServiceNow, Zendesk – sync data and create tickets during calls

All integrations

All pre-built connectors and managed services

Deploy and monitor

Track performance, review conversations, and manage your deployment pipeline.

Environments

Manage sandbox to production

Dashboards

Track performance metrics

Conversations

Review and diagnose calls

Resources

Maintain

Ongoing agent maintenance

Troubleshooting

Common issues and solutions

Release notes

Latest features and updates

Where do you want to start?

Build an agent with or without code

Low-code and no-code agent building in Agent Studio.

Maintain or update your agent

Edit knowledge, voice, closures, and more on a live Agent Studio agent.

Build like an engineer

Pull, edit, and push agent configuration as YAML and Python files with the ADK CLI.

Agent Studio APIs

REST APIs to update and configure agents programmatically.
Non-technical user? Try simplified mode – hides developer, API, and advanced docs so you only see what’s relevant for Agent Studio.
New here? See the platform overview for how Agent Studio, the ADK, and APIs fit together, or jump straight into the quickstart.

Build your agent

Set up your agent’s personality, knowledge, voice, and conversation flows.

Agent settings

Configure personality and rules

Knowledge

Managed topics, FAQs, and connected external sources

Voice

Customize text-to-speech and audio

Chat

Deploy a webchat version of your agent

Flows

Build multi-step conversations

Call handoffs

Transfer users to human agents

Integrations ecosystem

Connect your agent to telephony, CRM, payments, and other systems. See the full catalog.

Telephony

Five9, NICE CXone, Twilio, Amazon Connect, Genesys, Dialpad, and more

CRM and ticketing

Salesforce, ServiceNow, Zendesk – sync data and create tickets during calls

Managed services

Pre-built connectors for payments, reservations, and data

Custom and MCP

Build your own integrations through HTTP, Python, or Model Context Protocol

Deploy and monitor

Track performance, review conversations, and manage your deployment pipeline.

Environments

Manage sandbox to production

Dashboards

Track performance metrics

Conversations

Review and diagnose calls

Resources

Maintain

Ongoing agent maintenance

Troubleshooting

Common issues and solutions

Release notes

Latest features and updates
Last modified on April 23, 2026