The customer support landscape is transitioning from basic automation into the era of digital labour. Within modern service ecosystems, agentic AI marks a significant advancement over previous generations of support tools by moving beyond static responses toward autonomous problem-solving.
Digital labour encompasses productive activities facilitated by digital infrastructures, where work is coordinated.It can also describe a hybrid workforce where AI “agents” work alongside human employees to increase scale and speed. Bernard Slowey, SVP of Digital Customer Success at Salesforce, comments, “In customer service, this means AI agents that don’t just answer questions or deflect tickets, but actively resolve issues using real-time context from customer data, product history, and prior interactions. They operate with clear objectives and guardrails, working alongside human teams rather than replacing them”.
“This is fundamentally different from earlier generations of AI, which were largely rule-based or retrieval-focused. Traditional chatbots could route or surface content; agentic systems can reason, decide, and act”.
Unlike rule-based systems of the past, these agents work with clear objectives and guardrails, functioning as partners to human teams rather than mere replacements. “The impact of this shift is measurable and growing, with Australian service leaders projecting that AI will resolve more than half of all customer cases by 2027. This transition is less about cost-cutting and more about meeting permanently heightened customer expectations for speed, consistency, and scale”, says Slowey.
Intelligence in action – moving beyond knowledge articles
The shift toward more empathetic and intelligent support model is driven by two primary pillars. “The first is trusted, unified data, where the AI operates with full customer context including account history and unstructured documentation. Without this foundation, AI remains reactive” says Slowely”.
The second pillar is intentional design focused on the service experience itself. “By recognising situational context—such as whether a user is reporting a critical outage or seeking general guidance—the system can adapt its tone and timing. This focus ensures that proactive service makes customers feel understood and supported even within a fully automated interaction”.
The architecture of trust and human synergy
When AI agents act autonomously, trust becomes a non-negotiable requirement. Slowely advises that systems should be built upon dedicated security layers that ensure data protection through masking, secure retrieval, and zero-retention controls. He says, “Organisations maintain complete sovereignty by setting specific guardrails, such as ensuring high-stakes financial transactions are always routed to a human. Furthermore, comprehensive audit trails provide transparency into every action the AI takes, ensuring accountability. To mitigate risks, designers implement explicit boundaries to prevent hallucinations and provide clear escalation paths, ensuring that autonomy never results in a lack of oversight”.
Effective human oversight is integrated into the workflow as a mechanism for continuous improvement rather than a bottleneck. When an escalation occurs, the agent transfers the full context of the interaction to a human representative, eliminating the need for the customer to repeat themselves.
“On the backend, every human intervention becomes a learning signal. We analyse where customers choose to escalate, what triggered it, and how a human resolved the issue. Those insights are fed back into Agentforce to refine behavior, close knowledge gaps, and improve future performance”, says Slowey.
Empowering the workforce
The transition to agentic AI has revealed surprising insights into human behavior and workforce development. Experience shows that customers often feel more comfortable asking technical or “simple” questions to an AI that they might hesitate to ask a person, building a unique psychological dynamic of confidence. However, these successes also highlight that an AI is only as effective as the data it can access. For instance, if support content is siloed across different platforms, the AI requires integrated data solutions to provide accurate answers. Solving these connectivity issues is essential for maintaining the system’s efficacy.
The global explosion of data has far surpassed human processing capabilities, with daily creation expected to reach 463 exabytes this year. AI agents are uniquely equipped to handle this massive scale, consuming vast amounts of information to generate real-time insights and continuously refining their decision-making to drive superior outcomes.
“Digital labour has emerged as the essential solution to these dual pressures. While automation has historically improved speed and efficiency, the modern convergence of data availability and urgent demand has reached a tipping point that allows AI agents to power a sophisticated, autonomous digital workforce”, Slowey comments.
Breakthroughs in machine learning, natural language processing, and automation have finally transitioned AI from a theoretical tool into a practical reality, enabling agents to engage in complex reasoning and independent action that bridges the gap between massive data sets and an understaffed labor market.
Ultimately, agentic AI elevates the roles of both customers and support professionals. By absorbing millions of repetitive, high-volume interactions, the technology allows human agents to transition into more complex, judgment-based roles. At Salesforce, this has enabled support engineers to move into strategic positions, such as helping other organisations implement AI or transitioning into customer-facing advisory roles. The result is a more skilled and engaged workforce where human empathy and deep product knowledge are prioritised, creating a service model that scales sustainably without leading to employee burnout.