Artificial intelligence is often framed as a disruptive force that threatens to eliminate frontline employees. Headlines about automation and job loss dominate public debate, and in customer service, the question is usually posed in strong terms: Will intelligent technologies replace human workers?
This study suggests this is not the most useful question. In many service settings, intelligent technologies do not simply take over human work. They change the conditions under which human service agents do their work, sometimes for the better, sometimes creating new pressures, and always in ways that depend heavily on how organizations choose to implement them.
In the article Augmenting human service agents with intelligent technologies, published in the Journal of Service Management, Waelbers and colleagues reviewed 99 empirical studies on the use of intelligent technologies in service interactions. The review shows that technologies such as artificial intelligence, machine learning, natural language processing (NLP), and emotion recognition are increasingly used to support service agents before, during, and after customer interactions, not to replace them.
A shift from automation to augmentation
For more than a decade, organizations invested in automation technologies, chatbots, self-service portals, and interactive voice systems, primarily designed to reduce labor costs and increase throughput. Yet service firms continue to rely heavily on human employees, and consumers often prefer or expect human contact, exhibiting a well-documented aversion to algorithm-provided services in complex or emotionally charged situations.
This has shifted attention toward augmentation: using technology not to replace human strengths such as empathy and contextual judgment, but to complement and extend them. The distinction matters enormously. Automation removes tasks from humans. Augmentation gives humans better tools to perform them.
Six ways technology is reshaping frontline service work
The study’s thematic analysis identified six main ways in which intelligent technologies augment, rather than substitute for, the work of service agents.
- Before the conversation begins. Machine learning systems now analyze historical demand patterns to forecast call volumes and allocate agents accordingly. Routing technologies match customers with agents whose expertise fits the problem, while NLP models classify incoming requests in real time. These developments improve operational efficiency without removing human interaction from the service process.
- During the conversation. Speech recognition converts spoken dialogue into structured data. Topic modeling identifies recurring issues across thousands of interactions. Intent recognition detects underlying customer goals in real time. Together, these tools allow service agents to respond more accurately and efficiently, not by overriding their judgment, but by surfacing relevant information faster than human memory alone could.
- Supporting employee well-being. Customer service is emotionally demanding work with high turnover. Some systems now detect stress patterns in interactions and alert supervisors when timely support may be needed. Others monitor workload distribution and identify skill gaps for targeted training. However, technology that monitors employees can also pressure them; the same systems that flag overload can make workers feel surveilled. Implementation choices are therefore not neutral.
- Service agent monitoring. Digital evaluation tools now assess interaction quality, procedural compliance, and behavioral patterns across large volumes of service encounters. These systems improve consistency and enable organizations to identify coaching opportunities far more efficiently than manual supervision. Yet pervasive monitoring can reduce perceptions of autonomy and increase stress. The evidence is clear: when monitoring is framed as developmental support, it can enhance motivation. When framed as surveillance, it tends to undermine trust.
- Understanding the customer’s emotions. Sentiment detection and emotion recognition tools analyze tone, vocabulary, and speech patterns to identify affective states in real time. Multimodal systems combine audio and visual signals for more accurate assessments. These tools help service agents recognize frustration or confusion early, enabling more empathic and precise responses. At the same time, emotion detection introduces new expectations: agents must now interpret machine-generated insights while maintaining authenticity, a balance that requires professional judgment, not just technical compliance.
- Collaborative service systems. Increasingly, service work is organized around active collaboration between human agents and intelligent technologies. Chatbots handle routine requests and hand off complex cases to humans. Recommendation systems surface relevant solutions. Service robots assist with physical tasks in retail and hospitality settings. Mixed-reality interfaces support decision-making in advisory contexts. In these configurations, technology and human expertise co-create value; the question is not who does the work, but where human judgment adds the most value.
Intelligent technologies create value when they give agents better information, reduce avoidable effort, and strengthen professional judgment. They become problematic when organizations treat technological output as a substitute for human understanding.
— Bea Waelbers
Balancing technological augmentation and service customization
A key contribution of the review is a conceptual framework mapping the six themes along two independent dimensions: technological augmentation and service customization. Technological augmentation refers to the degree to which intelligent systems actively participate in service interactions. Service customization reflects the extent to which interactions adapt to individual customers’ needs rather than following uniform scripts.
These two dimensions are independent. High augmentation does not automatically produce personalized service, and deep customization can be achieved through human skill alone with minimal technology. The outcome depends not primarily on which technology is deployed, but on how it is integrated into professional practice, and whether the design supports or constrains human judgment.
This framework helps explain why some organizations experience improved service quality following digital transformation, while others encounter employee resistance or customer dissatisfaction. The outcome depends not only on technological capability but also on how systems are integrated into professional practice.
The critical role of organizational design
Augmentation is not a property of technology. It is a property of implementation. The same tool can support or constrain employees depending on how it is designed, framed, and embedded in work routines.
Monitoring technologies illustrate this most clearly. Digital performance systems can improve consistency and enable targeted coaching, or generate constant pressure, erode autonomy, and increase turnover. The difference is about whether feedback is used developmentally or punitively, and whether employees have visibility into the data.
Getting this right also requires attention to ethics. Successful implementation depends on organizational trust, transparency, and role clarity. Employees need to understand how systems support their work rather than replace it. As personalization deepens, questions of fairness, bias, and privacy become harder to ignore. These are not technical problems. They are organizational ones, and they deserve the same attention as the technology itself.
The practical implication for managers is straightforward. The relevant question is not which tasks can be automated, but how intelligent technologies can be embedded in service systems in ways that genuinely support both service quality and the people delivering it.
Looking ahead
Service work is not standing still. Hybrid configurations, where conversational agents handle routine requests and human agents step in for tasks that require judgment, empathy, or complexity, are becoming the norm rather than the exception. But the main conclusion is not that AI is good or bad for service agents. The more important point is that augmentation depends on design and implementation. Intelligent technologies create value by giving agents better information, reducing avoidable effort, and strengthening professional judgment. They become problematic when organizations treat technological output as a substitute for human understanding. The future of service work will not be decided by technology alone. It will depend on whether organizations use these technologies to replace human judgment or to help service agents use them better.
Reference
Waelbers, B., Henkel, A. P., & Bromuri, S. (2026). Augmenting human service agents with intelligent technologies. Journal of Service Management, 37(6), 26–49. https://doi.org/10.1108/JOSM-01-2025-0035
Coauthors
Dr. Alexander P. Henkel is an assistant professor at the Open Universiteit, Heerlen, The Netherlands. His research sits at the intersection of consumer psychology, organizational behavior, and service interactions, with a particular interest in how artificial intelligence and digital technologies reshape customer experiences and frontline work. His work has appeared in leading marketing and service journals, including the Journal of Consumer Research, Journal of Consumer Psychology, Journal of Service Research, and Journal of Service Management. ;
Prof. Dr. Stefano Bromuri is a Full Professor of Software Engineering and Artificial Intelligence at the Open Universiteit in Heerlen, the Netherlands. His research spans deep learning, signal processing, and neurosymbolic reasoning, with applications ranging from healthcare decision support to AI-driven business process optimization.
