February 10, 2026
Genesys Breaks New Ground with AI Agent that Completes Tasks Across Multiple Systems
Genesys has announced the launch of its Genesys Cloud Agentic Virtual Agent, the “industry’s first” virtual agent that can autonomously execute tasks across multiple enterprise systems to resolve customer requests end-to-end.
The new platform uses large action models from partner Scaled Cognition to understand customer goals, determine necessary steps, and execute complex actions across CRM systems, billing platforms, service operations tools, and other back-office software without human intervention. Several major organisations, including M&T Bank, Banco Pichincha, a global Fortune 500 healthcare company, and a Fortune 50 North American retailer, are already testing the technology.
The Self-Service Problem
According to Gartner research, the average self-service success rate sits at a dismal 22%. Most chatbots are “expensive failures”, with almost 90% customer interactions with them ending in frustration. Nearly half of business leaders say improving self-service outcomes ranks among their top three priorities for 2026, yet most are still throwing money at the same broken solutions.
While companies have rushed to adopt virtual agents built on large language models to improve conversational quality, these systems collapse when customers need actual problems solved. They can chat naturally and sound impressively human, but they hit a brick wall when asked to execute multistep workflows across different systems. Frustrated customers end up in the same place they always do, talking to a human agent and explaining their problem for the third time.
Enter the Action Model Revolution
The new Genesys solution takes a fundamentally different approach. It serves as a central orchestration layer that coordinates actions across disparate enterprise systems that would typically require manual handoffs, endless custom integrations, or customers navigating a maze of departments.
Scaled Cognition’s APT-1 large action model powers the Genesys Cloud platform, and unlike traditional language models that excel at conversation but fumble the execution, they are purpose-built for action-grounded work that spans multiple systems. This means they can reliably carry customer requests through to completion while avoiding the hallucinations that plague many AI-powered customer service systems.
For example, a customer needing a billing adjustment would normally navigate phone menus, explain their issue to a chatbot that can’t help, get transferred to an agent, and repeat everything while the agent logs into multiple systems.
With the Genesys agentic virtual agent, the customer solves the problem within one conversation. The agent automatically verifies the account in the CRM, processes the billing change, updates service records, and sends confirmation, all in a single interaction across multiple systems without transfers or repetition.
Trust, But Verify
Of course, giving AI agents the keys to your enterprise systems sounds either revolutionary or terrifying, depending on your perspective. Genesys has built in extensive safeguards to ensure it’s the former.
The system includes an embedded governance framework with guardrails, unified data access, and transparent decision paths to ensure every action remains explainable and aligned with company policies. Through Genesys Cloud AI Studio, organisations can design and configure their virtual agents by defining specific permissions and behaviours that match their compliance requirements across all connected systems.
Olivier Jouve, Chief Product Officer at Genesys, said: “Autonomy in customer experience only works when it’s built on trust, transparency and control. With our LAM-powered Agentic Virtual Agent, we’re enabling AI to reason, plan and safely take action across systems. This gives organisations a responsible way to move beyond conversations and deliver consistent outcomes customers can rely on.”
The platform includes action-level explainability and auditability features, meaning organisations can see exactly how decisions were made and executed across different systems. It also continuously learns to improve accuracy and containment rates while maintaining the oversight and compliance controls that keep legal and compliance teams sleeping soundly at night.
Genesys plans to expand the platform’s capabilities by adding support for open standards, including Agent-to-Agent and Model Context Protocol. This will enable the virtual agents to collaborate securely with other AI agents and enterprise systems, orchestrating multistep customer outcomes that span multiple platforms while maintaining centralised governance and policy enforcement.
