The Gist
- Access expanded. PolyAI's Agentic Dialog Platform is now open to any builder.
- Enterprise-grade tech. Models handle high-complexity, mission-critical conversations at scale.
- Faster automation. CX, product and developer teams gain rapid access to advanced dialog agents.
PolyAI on May 18 opened its Agentic Dialog Platform to the public, making enterprise-grade conversational AI available to any team with an email address. The platform, previously limited to large enterprise customers, is free for the first two months.
According to the company, builders can create production-ready dialog agents in under 10 minutes using the same infrastructure that powers deployments for Marriott, Foot Locker, PG&E, Caesars Entertainment, UniCredit and FedEx. PolyAI said the platform supports 75 languages across 25 countries, with its largest deployments performing work equivalent to more than 1,000 full-time employees per enterprise.
Table of Contents
- PolyAI CTO: Why Agentic Dialog Must Be Built Into the Model Itself
- PolyAI Platform Capabilities
- What PolyAI Is Opening Up
- The Model Underneath
- Enterprise Deployments Cited
- What It Means for CX Teams
- Recent PolyAI Developments
- Agentic AI in the Contact Center
- PolyAI Background
PolyAI CTO: Why Agentic Dialog Must Be Built Into the Model Itself
Most models treat dialog as an afterthought, with prompts layered on top after the fact. With Raven, agent behavior is in the weights, not bolted on through prompts that drift under pressure.
- Shawn Wen, co-founder and CTO
[PolyAI]
PolyAI Platform Capabilities
The open platform introduces several tools aimed at accelerating agent development and deployment.
| Capability | Description |
|---|---|
| Agentic Dialog Platform | Publicly accessible enterprise-grade conversational AI infrastructure |
| Poly Agent Builder | Natural-language input generates production-ready agents in minutes |
| Agent Development Kit | API keys, CLI, IDE and Git support for developer workflows |
| Raven | Proprietary dialog model trained on 1B+ enterprise conversations |
| Multi-Model Support | Deploy Raven or integrate GPT-5, Claude, Gemini and others |
What PolyAI Is Opening Up
The platform — previously available only through enterprise engagements — is now free for the first two months for any team with an idea and an email address, according to the company. PolyAI says the technology currently operates across 75 languages and 25 countries.
The launch includes three components. The first is Poly Agent Builder, a no-code tool designed to let CX, operations and product teams build agents by describing business needs in natural language. PolyAI says the tool auto-configures the knowledge base, conversation tracks and guardrails in minutes.
The second is an Agent Development Kit (ADK) — a developer-facing offering with self-serve API keys, native integrations and CLI support. Developers can build in their own IDE, version with Git and deploy from a terminal.
The third component is a shareable testing environment that allows teams to validate agent behavior across channels before deployment.
The Model Underneath
PolyAI's platform runs on Raven, a proprietary dialog model the company says was trained on more than one billion enterprise conversations. Unlike general-purpose large language models adapted for voice after the fact, PolyAI says Raven was built for dialog from the ground up — with agent behavior embedded in the model weights rather than layered on through prompting.
Wen added: "We built the best default for dialog, then made the platform open to every model an enterprise builder might need."
Builders can use Raven as the default or bring in third-party models including GPT-5, Claude, and Gemini, according to the company.
Enterprise Deployments Cited
PolyAI points to a range of existing enterprise deployments as proof of platform scale. The company says it powers FedEx operations across more than 20 countries, helped UniCredit improve Net Promoter Score by 14 points, and serves more than 3,000 restaurant locations — including Fogo de Chão, which the company says achieves a 95% guest satisfaction score using the platform. Marriott, PG&E, Foot Locker, and Caesars Entertainment are also listed as customers.
PolyAI says its largest deployments perform the equivalent work of more than 1,000 full-time employees per enterprise.
"Everyone wants to move at the speed of thought right now: writing code, shipping products, using AI to do in minutes what used to take days," said Nikola Mrkšić, Co-founder and CEO of PolyAI. "But dialog with complexity, the kind where an agent must participate in a real, high-stakes conversation and actually deliver a resolution, is a problem that hasn't been cracked at scale. That's exactly what we've spent a decade solving with the world's best brands, and today we're opening it up to every builder."
What It Means for CX Teams
The move positions PolyAI in a market increasingly crowded with agentic AI vendors targeting the contact center — but the company is betting that its dialog-specific foundation model differentiates it from general-purpose platforms that have added conversational AI capabilities as an extension of existing chat infrastructure. Opening the platform to self-serve builders also suggests PolyAI is looking to expand its reach beyond large enterprise contracts into the broader developer and mid-market segment.
Recent PolyAI Developments
London-based PolyAI has built momentum over the past 18 months, anchored by an $86 million Series D closed in December 2025 and co-led by Georgian, Hedosophia and Khosla Ventures, with participation from NVentures (NVIDIA), Citi Ventures, Zendesk Ventures and Point72 Ventures. The round pushed total funding above $200 million, with proceeds directed toward its Agent Studio platform and go-to-market expansion.
External validation has followed. Deloitte ranked PolyAI the #1 fastest-growing AI company in its UK Technology Fast 50, and a Forrester TEI study quantified a 391% ROI for customers, including $10.3 million in three-year labor savings and a 50% reduction in call abandonment.
Geographic expansion is accelerating in parallel. This month, PolyAI announced a Toronto technology hub alongside senior hires drawn from Cohere, Spotify, Google and Shopify. The company now reports more than 200 enterprise customers and thousands of live deployments.
Agentic AI in the Contact Center
Agentic AI platforms are pushing contact centers beyond scripted chatbots toward systems that reason, plan and act autonomously on complex customer issues.
From Chatbots to Autonomous Agents
Unlike rule-based tools, agentic AI operates across multiple data streams — support tickets, purchase history, product usage and real-time system status — to resolve issues end-to-end. As a CMSWire analysis outlined, natural language enables meaningful task completion rather than brittle prompt responses.
Key enterprise applications include:
- Autonomous handling of complex inquiries end-to-end
- Intelligent call routing based on customer intent
- Real-time agent coaching during live interactions
- Automated account updates and order processing
- Predictive issue identification and proactive outreach
ROI & Performance Benchmarks
The business case is increasingly quantifiable:
- Autonomous AI systems reported 28% faster resolution and 19% higher first-contact resolution
- Generative AI agents drove a 14% increase in resolutions per hour and 9% shorter handling time
- Omnichannel deployments cut operational costs by up to 30%
Gartner predicted that by 2029, AI will independently handle 80% of routine contact center inquiries. Enterprise conversational AI is projected to grow 192% from $17.05 billion in 2025 to $49.8 billion by 2031.
PolyAI Background
PolyAI, founded in 2017, offers AI chatbots for customer service leaders at mid-to-large enterprises managing high-volume contact centers. The company provides voice-first, omnichannel virtual agents across voice, chat and SMS, with analytics, monitoring and enterprise controls for policy and escalation.
Have a tip to share with our editorial team? Drop us a line: