Genesys recently announced what it calls the industry’s first agentic virtual agent built on large action models (LAMs). The company is framing the launch as a turning point away from LLM-based chatbots that answer questions, toward agentic AI and something that actually resolve customer requests end-to-end. That framing is correct. The more interesting question is what it means for the broader architecture of enterprise CX.

It’s important to understand the distinctions between LLMs and LAMs. LLMs are trained to generate plausible text. LAMs are agentic systems instructed to interpret the intent of conversations and take actions that should follow from them. The result is a model that selects and executes predefined steps (update account, trigger billing check, issue refund) rather than producing a response and hoping a downstream tool call succeeds.

Genesys combines deterministic controls for business logic with general language understanding making the “appropriate decision, not just a plausible suggested one.” The model is predicting the next token that should look like a tool call. For a contact center running at scale across regulated industries, that distinction is critical to ensuring the value proposition.

Where CXO and the AI Agent Control Plane Fit
Opus Research defines Conversational Experience Orchestration (CXO) as the operating model that unifies conversational AI, conversation intelligence, and agentic automation into a single governance loop. A CX control plane for AI agents tracks customer journey state across channels, enforces policy, manages identity and consent, and determines when AI acts versus when a human must step in. Evaluation and quality systems are equally critical ensuring success, safety, and compliance.

According to the announcement, the Genesys Cloud AI Studio functions as the governance and skill-definition layer and an event data platform provides real-time journey context. Planned future support for Agent-to-Agent (A2A) and Model Context Protocol (MCP) extends orchestration across the broader enterprise ecosystem without requiring Genesys to own every integration. The LAM executes (deterministically, auditably, within policy) which is precisely the division of labor the CXO control plane is designed to enforce.

Questions CX Leaders Should Be Asking
The Genesys announcement is a bellwether of dynamic changes occurring in contact center software, and it is directionally where the market is heading.

For CX leaders, the practical implication is that the architectural decisions you make in the next twelve months about where AI governance lives will be difficult to reverse. Before committing to any agentic platform, ask these questions:

• What actions can the agent actually take, and who governs that catalog?
• Where does customer journey context live, and can you export it?
• How do you ensure this institutional knowledge survives in a way that precludes vendor lock-in?
• What policy and business logic determines AI-to-human handoff?

The chatbot era optimized for better conversations. The emerging era of CXO and the AI agent control plane is optimizing for resolved outcomes. For CCaaS platforms, that shift redefines what the product is for.

Categories: Conversational Intelligence, Intelligent Assistants, Articles