In the past, mention AI to an executive, and they will likely picture a traditional rules-based chatbot or a standard text-completion copilot.The arrival of sophisticated, multi-modal AI frameworks and cognitive computing has shattered those old paradigms, shifting the conversation from simple automation to true operational autonomy.
In an insightful industry interview, David Russell, General Manager, Enterprise Digital (Corporate) at Nexon Asia Pacific, recent winners of the Genesys ANZ Partners of the Year and the APAC Innovation of the Year, sat down with industry CXFocus editor, Mark Atterby to discuss why the market is moving rapidly toward Agentic AI, and why this represents an entirely new proposition for operational efficiency and customer care.
Moving from answering questions to completing work
To understand the evolution of AI/CX technology, Russell breaks down the evolution of the market into three distinct categories:
- Chatbots: Rules-based tools that excel at simple question-and-answer deflection.
- Copilots: Inline AI assistants designed to help human agents do their immediate jobs faster.
- Agentic AI: Autonomous systems given the agency to understand a business objective, determine the required operational steps, interact with backend systems, and execute an entire workflow to deliver an end-to-end outcome.
“The fundamental difference is agency,” says Russell. “Agentic AI isn’t just generating an answer – it’s completing work on behalf of organisations and people. We are moving away from an automation focus toward an autonomy focus, asking how many business processes can be completed entirely from start to finish.”
Tackling Australia’s structural friction
This shift comes at a critical time for Australian enterprises, which are grappling with uniquely high labour costs and persistent skill shortages. Russell emphasises that the primary goal of Agentic AI isn’t cutting headcount, but rather removing friction and internal inefficiencies.
“Every minute an employee spends searching disparate systems, transferring customers, or manually updating records is incredibly expensive,” Russell notes. “Agentic AI delivers immediate relief by increasing employee capacity without increasing headcount. By removing routine friction, human agents are suddenly far more effective.”
Rather than decimating contact centres, this automation acts as a productivity multiplier. Taking the routine administrative load off the frontline has directly measurable outcomes – lower costs to serve, faster times to resolution, and ultimately higher customer retention because issues are handled consistently.
The rise of the enterprise orchestration layer
A common pitfall for many organisations expanding their AI footprint is the ‘siloed pilot’ trap. Companies frequently run isolated AI trials across separate departments—one in marketing, one in HR, and another in IT support—with zero visibility or data sharing between them.
To scale successfully, Russell urges businesses to transition from disconnected solution sets to an enterprise capability model built around a single orchestration layer.
“An orchestration layer allows every single agent, both AI-generated and human, to work within the exact same security, compliance, and customer experience framework,” Russell explains. This layer effectively acts as the connective tissue sitting across an enterprise, allowing an autonomous AI agent to make decisions and securely update entries across legacy ERPs, HR systems, CRMs, and billing platforms.
To build out this enterprise framework, Russell advises organisations to start by answering three foundational questions:
- What customer journeys create the most value? Where are the friction points that would benefit most from autonomy?
- What decisions should the AI actually be allowed to make? Defining the boundaries based on industry risk and security frameworks.
- What governance should apply consistently across every single agent?
Managing agent cognitive load
One of the most frequently overlooked outcomes of successful AI adoption is the dramatic shift in the nature of the human agent’s role.
When Agentic AI successfully absorbs high-volume, simple transactional queries (like tracking an order or resetting a password), human agents are left handling only the most complex, emotionally charged, and high-stakes tier-2 and tier-3 issues. Consequently, Average Handling Time (AHT) per human agent will likely rise.
“Agents are moving away from transactional processing and becoming pure problem solvers,” says Russell. To prevent staff burnout and empathy fatigue, organisations must fundamentally change how they support and measure their frontline teams:
- Real-time coaching: Moving co-pilots beyond simple text summaries toward delivering real-time supervisor coaching and next-best-action recommendations so agents feel supported mid-call.
- Redesigning KPIs: Shifting metrics away from speed alone. In an agentic world, it is the AI’s job to be fast. The human agent’s job is to solve highly complex scenarios in an empathetic, sophisticated way—and company reward structures must align with that quality of resolution.
Redefining success via business impact
As multinational LLM providers adjust token pricing and commercial models fluctuate, evaluating AI success requires moving past vanity metrics. “Saying customers interacted with a bot 50,000 times last month is not an indicator of success,” Russell warns.
Instead, organisations winning the market are tracking definitive business impacts: true containment-and-resolution rates, actual cost-to-serve reduction, and customer trust metrics (such as CSAT and NPS). Ultimately, the competitive edge belongs to companies that can seamlessly orchestrate their people, their autonomous AI, and their backend business processes into a single, unified operating model.