Announcing Our Partnership with Genesys

Today, we’re excited to announce a new partnership with Genesys®, a global cloud leader in AI-Powered Experience Orchestration. Genesys has selected Scaled Cognition’s specialized customer experience (CX) large action models (LAMs) to power virtual agents on the Genesys Cloud™ platform, and invested in our company to accelerate joint innovation.

Our partnership reflects a core belief we’ve held from day one: that the customer experience requires specialized AI models built for reliable action, not just words. This collaboration will pair the industry-leading Genesys Cloud platform with a true AI model lab creating a clear choice for enterprises that refuse to compromise between reliability and innovation.

Together, Scaled Cognition and Genesys are extending the reliability and precision of the Genesys Cloud platform to deliver an agentic system designed for the complex CX needs of the world’s leading organizations.

Why This Partnership Matters

General-purpose AI models transformed how organizations think about automation and intelligence. When extended to the nuanced demands of CX, additional layers of reliability, control and accuracy become essential. Without this refinement, systems can struggle to maintain consistency and compliance across workflows. Scaled Cognition and Genesys are bridging this gap with a purpose-built approach designed for enterprise-grade performance.

As the world’s only AI model lab focused exclusively on CX, we create specialized LAMs along with development and evaluation tools, that allow customers to achieve:

• High automation rates: resolution of complex, end-to-end workflows with fewer handoffs and escalations.
• Hallucination-free outputs: deterministic control of outputs that help eliminate hallucinations.
• Strict policy adherence: reliable enforcement of enterprise rules through novel model design.
• Secure deployment: available through the Genesys Cloud platform to provide enterprises security they can trust.

Our partnership with Genesys will enable customers to rapidly create and deploy CX agents that deliver strong business results with our high performance LAMs on a platform trusted by enterprises globally.

‍

AI Determinism for Real-World CX

Our technology is built on a new class of large action models (LAM)—advanced neural models that extend the capabilities of today’s LLMs by modeling actions in addition to words as first-order objects. Our flagship model is the Agentic Pretrained Transformer®, or APT-1®. It is designed and pretrained for CX systems, which must speak fluently with humans, reason, plan and also act deterministically with respect to enterprise logic, enabling outcomes to be policy-aligned, repeatable, and free from hallucinations.

Our solutions also include high-leverage tools for developing and evaluating AI agents. Using the same rich RL simulation technology that backs APT-1 itself, enterprises can build, scale, and test agents more reliably than even before. Rather than tweaking prompts and hoping for the best, changes are made in a modular, declarative way, complete with regression testing over simulated rollouts. Together, these capabilities allow guaranteed compliance at scale, with outcomes that are reliable, transparent, and aligned to enterprise policies.

The gains are dramatic. APT-1 gets top scores on benchmarks such as tau-bench, a standard for evaluating agentic performance in real-world domains, while being small enough to self host within an enterprise VPC or on-prem. Moreover, our models are omni-channel and global, supporting both voice and text in a wide range of languages, out of the box.

‍

Looking Ahead

This partnership is just the beginning. Together with Genesys, we will continue to expand what’s possible with agentic AI from automating workflows to scaling policy-safe virtual agents across industries and regions.

In the months ahead, we’ll also be releasing exciting new capabilities that further boost automation rates and new paradigms for authoring agents. At Scaled Cognition, our mission has always been clear: to build specialized AI models you can rely on.

‍

‍

‍

‍