Genesys has announced what it calls the industry’s first agentic virtual agent powered by large action models (LAMs), signalling a shift in enterprise customer experience (CX) from conversational automation to autonomous, outcome-driven execution.
Unveiled in Bengaluru, the new Genesys Cloud Agentic Virtual Agent is designed to move beyond scripted self-service by understanding customer intent, deciding next steps and executing complex actions across front- and back-office systems without human intervention. The company positions the launch as a response to persistently low self-service success rates, despite rising enterprise investment in automation.
According to Gartner, the average self-service success rate currently stands at just 22%, with nearly half of enterprise leaders ranking improvement in self-service outcomes among their top three priorities for 2026. Genesys argues that while large language model (LLM)-based bots have improved conversational quality, they struggle when asked to complete multi-step workflows or adapt to changing customer needs across systems.
From conversation to execution
The new agentic virtual agent addresses this limitation by combining LAMs with enterprise-grade orchestration, governance and guardrails. Rather than stopping at providing answers, the system can progress workflows across CRM, billing, service operations and other enterprise platforms, adjusting actions dynamically as conditions change.
Several organisations, including M&T Bank, Banco Pichincha, a Fortune 500 healthcare provider and a Fortune 50 North American retailer, are already piloting the technology to automate more complex customer requests.
Genesys said the capability is strengthened through its partnership with Scaled Cognition, whose APT-1 large action model has been integrated into the Genesys Cloud platform. Unlike LLMs, which are optimised for language generation, LAMs are designed for deterministic, action-grounded execution, reducing the risk of hallucinations and inconsistent outcomes.
Governance-first automation
A key differentiator, according to Genesys, is governance. Embedded controls, unified data access and transparent decision paths are intended to ensure that every autonomous action remains explainable, auditable and aligned with enterprise policies. Through Genesys Cloud AI Studio, organisations can define permissions, guardrails and behaviours, retaining oversight while scaling automation.
“Autonomy in customer experience only works when it’s built on trust, transparency and control,” said Olivier Jouve, chief product officer at Genesys. “By enabling AI to reason, plan and safely take action across systems, we’re giving organisations a responsible way to move beyond conversations and deliver outcomes customers can rely on.”
Genesys also plans to extend the platform with native support for open standards such as Agent-to-Agent (A2A) and Model Context Protocol (MCP), allowing secure collaboration between multiple AI agents and enterprise systems under a central governance model.
Industry response
Analysts see the move as part of a broader market transition. Hayley Sutherland, research manager for conversational AI at IDC, noted that scripted, turn-based self-service is increasingly inadequate for complex customer needs.
“Resolving complex requests requires AI that can plan and execute multi-step actions across systems, while remaining predictable and auditable,” she said, adding that Genesys’ focus on combining autonomy with orchestration and governance reflects where the market needs to head.
Enterprises echo this view. Trond Prestø, head of customer care at DNB, said agentic AI represents a natural next step in scaling self-service, provided it is implemented with strong controls and oversight.
Dan Roth, co-founder and chief executive of Scaled Cognition, was more direct: “In the enterprise, 80% accuracy is effectively useless for automation. Trustworthy automation is about reliability, not just intelligence. Combining large action models with Genesys’ governance framework delivers a more dependable AI stack for real-world operations.”
Genesys Cloud Agentic Virtual Agent is expected to be generally available globally in the first quarter of Genesys’ fiscal year 2027, running from 1 February to 30 April 2026.