Since ChatGPT rose to prominence at the end 2022, CX technology vendors have been busy integrating the capabilities of Large Language Models (LLMs) and Gen AI into their software platforms. The aim has been to deliver more human-like, personalised, and efficient interactions at scale.
Over the years, chatbots have evolved through several distinct generations, moving from being essentially online Interactive Voice Response (IVR) systems to becoming smart, flexible virtual agents powered by sophisticated AI. This evolution represents a fundamental shift from rigid, scripted interactions to dynamic, conversational experiences.
Jarryd Tuyau, AI Lead – Asia Pacific, Genesys, comments, “Generative AI, combined with advanced agentic workflows and orchestration, has significantly closed the capability gap between artificial intelligence and humans. This new technology can now perform tasks that were previously exclusive to humans, such as the ability to think, plan, reason, remember, and generate content. By operating within customer-defined guardrails and rules, Gen AI dramatically reduces the rigidity that characterised earlier virtual assistants. As this rigidity decreases, the AI’s core capabilities increase, resulting in far more natural and human-like customer experiences.”
The evolution of human-like interactions
The most significant change brought by Gen AI is the dramatic reduction in the rigidity of virtual assistants. Previously, AI technology was limited by strictly defined rules and intents. Gen AI, by contrast, has closed the capability gap, allowing AI to perform tasks that were once exclusively human domain.
- Human-like cognition: The new generation of AI has the ability to think, plan, reason, remember, and generate responses. This capability, guided by customer-set guardrails, enables a much less restrictive and more natural experience.
- Enhanced capability: As the level of rigidity drops, the AI’s capability inversely increases, leading to interactions that feel far more human than their predecessors.
Dave Flanagan, CX Strategy and Innovation Lead from Nexon, notes this shift: “The ability to think, plan, reason, remember, generate within guardrails, boundaries, and rules that have been set by the customer, allows for a lot less of a rigid experience. Humans aren’t rigid in nature, in their thinking. Previously, the technology was. As that’s changed, and that level of rigidity has dropped, inversely the capability has increased, and that’s led to far more human experiences with that technology.”
Strategic implementation – From pilot to production
For large enterprises, the path to implementing Gen AI often begins with a measured, risk-adverse approach. Key to success is moving beyond a simple interest in the ‘feature’ and focusing on the core business strategy and goals.
Flanagan says, “Some of our major clients, such as Latitude Financial Services, are already heavily leveraging the latest generative AI capabilities, including AI scoring, auto-summarisation, and predictive wrap codes. This rapid adoption is due to a proven, risk-adverse framework we use: we move from pilot studies to validate the business case, and measure the outcomes, allowing us to safely transition into production. This is especially critical for clients operating in highly regulated industries.”
Focusing on the use case
The most crucial step is defining the use case and measuring its expected outcomes. Simply applying AI to a broken process or technology architecture won’t fix it. Tuyau says, “I believe that once organisations establish a foundational Experience Orchestration strategy—a clear methodology for deliberately managing customer interactions—the optimal use cases for AI become immediately apparent”.
“For large enterprise customers”, continues Tuyau, “this methodical approach is crucial. By fully understanding and planning customer journeys, they exercise the necessary muscle for journey design. Implementing sophisticated AI capabilities then becomes a natural next step, rather than a forced technology adoption. This process significantly amplifies the potential value and ensures the business case for AI implementation is robust and clearly defined”.
Organisations that invest time to validate their initial business case, measure the outcomes, and build confidence with internal security and data teams will find subsequent feature rollouts much faster. According to Tuyau, the first large AI project is the hardest, but the groundwork laid expedites all future deployments.
The return of bot personas and personalisation
The focus on defining a bot’s persona is returning, made easier by Gen AI. Previously a time-intensive process, defining a bot’s personality can now be done quickly using simple prompts and brand guidelines.
Flanagan reflects, “In my prior role at a brand agency, I dedicated significant effort to developing bot personas for major clients, including brands like Coca-Cola. In the early days of conversational AI, defining a bot’s voice aligned with brand guidelines was a time-intensive, specialised focus.This focus is now strongly returning—not because brands are dedicating more resources, but because the process has become dramatically simplified. Modern platforms allow users to define a sophisticated persona quickly, often by simply injecting existing brand or social media guidelines via a few prompts, resulting in a cohesive and well-suited personality”.
Furthermore, the wealth of data (interaction transcripts, sentiment analysis, etc.) within platforms allows for hyper-personalisation of the bot’s persona. Instead of a single, fixed persona, the AI can craft a specific interaction style tailored to the individual customer’s demographics or previous interaction history.
This level of personalisation ensures customers receive service that aligns with their expectations, acknowledging that different segments (e.g., different age demographics) may expect different styles of customer service.
The scope of AI agents and autonomous workflows
The capability of Gen AI to handle complex, multi-step tasks across multiple platforms is no longer the main constraint. Flanagan comments, “The scope of issues an autonomous agent can resolve before requiring human intervention is highly variable, depending on the complexity of the interaction. However, in my view, the AI capability now largely surpasses current business requirements and organisational maturity”.
The challenge is no longer technological. Modern AI is easily capable of handling complex, multi-step tasks across diverse platforms. “The bottleneck has shifted to the underlying operational infrastructure, specifically, whether the organisation’s systems are tightly integrated with seamless data and context sharing. This infrastructural readiness is the difficult piece that many organisations are still navigating”, says Flanagan.
A distinction must also be made between what the AI could do and what it should do. There remains a definitive place for human connection, particularly for high-risk or complex, emotional issues. The goal is to apply the technology strategically based on a ‘risk profile’ and use case analysis.
AI as a coworker: Agent-human collaboration
The collaboration between AI agents and human agents is still in its nascent stages, but the prevailing narrative is one of supervision and partnership. As Tuyau points out, “For the foreseeable future, autonomous agents will require human supervision and maintenance and must function collaboratively alongside human teams. While fully automated, end-to-end self-service AI agents are achievable in specific contexts, a crucial element of success remains the supervisory and management layer”.
He adds, “Just as human agents are managed by supervisors, we foresee a future where teams of bots are managed by other agents and supervisors. We are only at the beginning of establishing these operational models, and it’s unlikely that any business has fully mastered this structure yet”.
An emerging and critical trend, according to Tuyau, is the shift in internal perception. People adopt AI when they feel it is being done for them, not to them. Framing AI as a member of the team fosters acceptance of the AI agent, delivering benefits across the team, the business, and the customer experience. Striking this balance requires the right change management, a clear narrative, and a focused North Star to ensure AI is understood and welcomed for the value it provides.
The future
The speed of maturation in Gen AI is phenomenal, far outpacing the development seen in the previous decade. Businesses must discard preconceived notions from even 18 months ago and actively engage with the technology.
AI is not just optimising existing businesses; it is an enabler of entirely new business models built from the ground up. Small startups can leverage AI as a value multiplier, allowing them to scale quicker and experiment with new models without massive initial investment. Tuyau says, “We are rapidly approaching (and already seeing) a point where businesses are built from the ground up, using AI as the core enabler of the entire business model. This leverage, which allows a two-person company to achieve the output of twenty, dramatically accelerates growth, enables rapid testing of new models, and significantly reduces the need for massive initial investment. We are still only at the tip of the iceberg”.