PolyAI

PolyAI, founded in 2017 by Cambridge machine learning researchers, builds dialog agents for enterprise customer conversations. Its Agentic Dialog Platform lets teams design agents in Dialog Studio, run them on its own dialog models, and deploy across voice, SMS, WhatsApp, web chat, email and mobile channels.

What It Is

PolyAI is an enterprise conversational AI platform for voice and text customer service agents. Dialog Studio provides flows, tools and brand voice in one place, from no-code to full code. Wren is the dialog agent assistant that helps design, build, test and continuously optimize agent experiences. Documented use cases include account management, authentication, call routing, billing and payments, booking and reservations, order management, troubleshooting and FAQ.

Models and Voice Stack

PolyAI states that Dialog-RSN-1 is its flagship audio-native model, reasoning directly over raw call audio for turn-taking, recognition and response, and that it is now available on AWS SageMaker. Dialog-Raven-3.5 is a text-based model for webchat, SMS, messaging and multilingual use. The company says it provides its own ASR and TTS engines, barge-in handling and low latency, so developers write the business logic.

Developer Workflow

The Agent Development Kit (ADK) is an SDK and CLI for building Agent Studio-compatible agents from a local IDE. Agent resources are versioned files in Git, and the poly push command fits CI/CD pipelines. Developers can write custom functions that call backend APIs in real time, run automated test suites from the CLI, and push third-party data into the flow at runtime through hooks. A Platform API, an MCP integration and a Model API are also listed.

Security and Deployment

PolyAI lists SOC 2+ Type 2, ISO 27001, PCI DSS and GDPR certifications, SIP/TLS with SRTP, mTLS, SSO/RBAC, regional data residency, PII redaction, guardrails and human-in-the-loop controls. Integrations listed include Microsoft Teams, Slack, Zoom, ServiceNow, Zendesk, Twilio, NICE and Dropbox.