Abstract
Background: The popularisation of artificial intelligence (AI) has accelerated its integration and use across various sectors for multiple purposes. Artificial intelligence technologies have also been applied to corporate operations to assist with handling customer inquiries. However, there remains a limited understanding of how these tools influence customer service capability and training, particularly within small- and medium-sized enterprises (SMEs) in developing contexts.
Objectives: Guided by the Implementation Science framework as a theoretical lens, this study investigates how AI support tools influence customer service capability, the factors affecting the diffusion, and establishes a framework to support the effective implementation and integration of AI tools to enhance customer service training.
Method: The study adopted a qualitative approach, employing the case study design. The data were collected using semi-structured interviews with participants from a purposive sample.
Results: The study revealed that AI-enabled chatbots improved information accessibility, offered personalised learning opportunities, facilitated self-paced and adaptive learning, provided consistency in responses, enhanced operational efficiency among customer service agents and contributed to teamwork and engagement. Findings suggest that the AI tool is most effective when used in conjunction with human facilitation. Challenges included the limited content depth of the responses and technical infrastructure constraints that questioned organisational readiness and strategic direction.
Conclusion: The findings highlight both the benefits and limitations of using AI-enabled tools in training environments. Despite this, human oversight and strategic infrastructure planning remain critical for successful adoption.
Contribution: This study presents the ABIRM framework developed to guide SMEs in the implementation and adoption of AI tools for training purposes.
Keywords: artificial intelligence; chatbots for training; customer service; workplace learning; implementation science framework.
Introduction
The success of a company heavily relies on the quality of customer service it offers, which is delivered by the customer service representatives. Ineffective handling of queries can cause customer dissatisfaction, diminish loyalty and harm the company’s reputation. Mbavai (2022:55) states that competitiveness in the industry is because of improved performance at every component level of the business. Essential practices should include efficient customer query handling, soliciting appropriate customer feedback and ultimately, providing effective customer service training.
Fowler, Stevenson and Wilson (2019:10) argue that effective training can ensure that representative agents provide quality customer service. The integration of technology into teaching and learning environments has undergone substantial change in recent years because of technological breakthroughs. One significant change is the growth of artificial intelligence (AI) and its creative uses in education and learning (Li & Miraj 2022). Artificial intelligence technologies have shown extraordinary skills, such as the ability to automate and simplify tasks, increasing productivity and producing better results. Examples of such technologies include ChatGPT and Notion AI, which have demonstrated how AI can be used to automate and streamline a variety of activities (Conroy 2023). Solutions driven by AI-technology enable digital transformation through automation, facilitation of seamless customer interactions (Ameen et al. 2021), as well as provide a platform that encourages continuous learning (Li & Miraj 2022). These features can revolutionise customer service practices and result in the overall enhancement of consumer service.
Artificial intelligence holds significant importance and benefits in its ability to automate and simplify processes, in addition to providing tailored offerings that collectively result in increased customer satisfaction (Jiang, Liu & Liu 2020:618). The ability to react to client enquiries swiftly and accurately, as well as efficiently satisfy their expectations, is a critical component of providing efficient customer service. By investigating the integration of AI technology into the commercial environment, specifically in customer service, companies can harness the potential of AI to eliminate the need for human facilitators and achieve improved customer service outcomes.
Problem statement
The problem driving this study is that we have a limited understanding of the factors influencing the adoption of AI technologies to enhance customer service within an organisation. High-quality customer service in the business environment is crucial for ensuring that customers receive the best service from the businesses they choose to support. There are ongoing issues related to the inability of customer service representatives to address queries effectively and a failure to meet customer expectations (Burton 2022:80). Technological advances have ensured the integration of technology into learning environments. Artificial intelligence has begun to produce innovative approaches to education and learning (Li & Miraj 2022).
Initial and continual training can be time consuming for employees, as small companies, non-corporate entities and start-ups may not have dedicated learning and development departments. Therefore, they require time from employees’ schedules to provide training to newcomers and conduct refresher training. Magnoler and Pacquola (2016:42) concur that this is attributed to the deprioritisation of effective training in small- and medium-sized enterprises (SMEs). As a result of traditional training techniques and the time required for instruction, employees struggle to remember the material they have learned. This is crucial as training ensures that employees can perform better at their jobs (Jaman & Hossain 2020:6).
Insufficient levels of product and customer service training are often offered because of time constraints imposed by traditional training methods. Additionally, there is no platform for employees to review previously taught material, necessitating more frequent refresher training. Training requires constant reassessment (Rosa 2023:2), thus affecting the amount of time facilitators need to spend retraining staff members as required. Furthermore, Syihabuddin et al. (2025:204) argue that knowledge retention necessitates structured documentation, as the lack thereof hinders the possibility of transferring and retaining knowledge.
Nassazi (2013:29) emphasises that effective employee training enhances productivity and boosts production. It is therefore necessary to investigate more effective and accessible training delivery techniques that are independent of in-person facilitators and do not require a significant amount of training-related man-hours. Thus, to improve the level of training provided, which will subsequently enhance customer service, companies must also investigate how AI may be used as a support model. It is crucial to examine the application of AI technology as a potential support model for current customer service training and evaluate the effectiveness of AI technology in reducing the need for human facilitator intervention. The overarching goal of the research is to better understand how AI can be effectively integrated into training programmes, ultimately enhancing the standards of customer service.
Research objectives
This research is undertaken with the following three objectives:
- To examine the utilisation of AI support tools to enhance customer service training.
- To investigate and understand factors influencing the adoption of AI technologies to improve customer service within the organisation.
- To develop a framework that assists in the implementation and adoption of AI tools to enhance customer service training.
Research questions
There are three questions this research aims to address:
- How can the utilisation of AI support tools improve customer service training?
- What are the factors influencing the diffusion of AI innovations to improve customer service within the business? and
- What elements will contribute to a framework that assists in the implementation and adoption of AI tools to enhance customer service training?
Literature review
Overview of artificial intelligence
Artificial intelligence has become increasingly popular in recent years because of the numerous benefits and ease it offers in users’ daily lives (Trabelsi 2024:142). Historically, AI was founded in the mid-20 century at the forefront of enabling machines to mimic cognitive functions that would otherwise be seen in human beings only (Elamin 2024:1152). Elamin (2024:1152) further references a paper written in 1950 by Alan Turing on ‘Computing and Intelligence’, which questioned the cognitive ability of machines in relation to human behaviour, acting as a catalyst to the innovation of AI. Alsaati (2024:1261) argues that AI is enabled by the use of large datasets to engineer machines in a way that autonomously allows them to analyse the data and make decisions based on their own experiences over time. Shrivastava et al. (2024:5501) concur that AI utilises data to automate repetitive learning and decision-making through algorithms that analyse data structures and frequencies in datasets, enabling self-instruction. Mondal and Lipizzi (2024) challenge the reliance on compromised quality data sets, threatening the integrity of outcomes produced because of biases. According to Cerezo-Martínez, Nicolás-Sánchez and Castro-Toledo (2024), this may pose significant challenges when using AI tools within training environments because of the responses that may be of a discriminatory nature as a result of biased training.
Current uses of artificial intelligence
Sangwan and Bansal (2024:4) reference several sectors where AI has been deemed beneficial, for example, healthcare, finance, the automotive industry, entertainment and education (Olorunsogo et al. 2024:482). Ali et al. (2024:17) concur this by demonstrating the essential use of AI in cancer treatments that range from early detection of cancer cells to the projected outcomes, mainly to minimise the errors related to diagnosis and treatment. Shrivastava et al. (2024:5504) discussed the use of AI for ailment detection in medical imaging, such as X-rays, computed tomography scans and magnetic resonance imaging, and further used for personalised treatment plans and the analysis of drug efficacy. The financial sector realises the benefits of the integration of AI with aspects such as fraud detection, autonomous credit scoring and trading using algorithms (Sangwan & Bansal 2024:4). Innovative AI solutions have equipped financial institutions with the tools needed to manage portfolios with the least amount of risk and navigate overall risk related to credit scoring and fraud (Wu 2024:18). Hidayat, Defitri and Hilman (2024:125) concur with the use of AI for managing risk through fraud detection and further detail the additional uses for predictive analysis in investments, automated tax planning using company data, predicting the financial performance of a company, the ability to automate routine processes using AI-based robotics process automation, market analysis using external data sources and providing personalised customer service using AI-enabled chatbots. Lad (2024:374) describes the key uses of AI in the automotive industry as the ability to achieve precision in custom manufacturing through optimisation of the tool path. Trabelsi (2024:149) vouches for the use of AI in manufacturing and production for control, self-correction of workflows and optimisation. Artificial intelligence video generation, made possible with text-to-video models such as Sora AI and King AI, has gained popularity in the entertainment sector (Tian 2024:52). Pandit and Kirdat (2024:2829) found that AI has been remarkable in the animation realm for virtual character design, character animation, voice synthesis and acting, generating backstories for characters and facilitating dialogue and interaction between animated characters. Furthermore, Alkan (2024:486) found that AI offers a personalised learning experience, at tailored speeds, which caters to the differing learning styles and levels seen in students. Tahir, Hassan and Shagoo (2024:1470) argue that AI is beneficial in the customised design and delivery of teaching content and can further be used in assessments to evaluate learner and group progress and provide valuable feedback.
Generative artificial intelligence in training environments
Generative AI exists within the domain of AI technologies, enabling the processing of Big Data using algorithms and creating human-like responses in the form of images, text, audio content and graphics (Sahu & Grover 2024). Al-Dahoud et al. (2024:405) highlight the most popular examples of generative AI built by OpenAI as ChatGPT, DALL-E and GPT-4. Digital communication has been transformed with the introduction of AI-enabled chatbots, revolutionising the way businesses communicate with their customer base (Nze 2024:35). According to Kurniawan et al. (2024:4), AI-enabled chatbots are beneficial in their ability to simulate human-like conversations using voice or text recognition. The offering is enabled by leveraging Machine Learning (ML) algorithms and Natural-Language Processing (NLP) models that facilitate the behaviour seen in AI chatbots (Ekechi et al. 2024:1266; Karyotaki, Drigas & Skianis 2024:44). Ekechi et al. (2024:1263) attribute the popularity of AI-enabled chatbots in customer support environments to their ability to act as sophisticated virtual assistants, thereby improving the support offered to customer bases and enhancing the overall customer experience.
The introduction of AI in learning environments has digitised and transformed the way content is delivered and consumed. According to a study conducted by Suntharalingam (2024:723), AI-aided learning resulted in improved academic achievements and knowledge retention, contributed positively to skills development seen in learners and tailored learning styles through algorithmic feedback mechanisms. Bhandare (2024:2) concurs that the practical integration of AI is beneficial for immediate evaluation because of the real-time feedback received during interaction, which can be used to improve performance and motivate learners. Rohde, Flindt and Rietz (2024:121) highlight adaptive learning systems as an AI benefit, offering personalised learning plans that address individual learner needs as opposed to the generalisation of a larger cohort. Furthermore, the nature of AI content delivery offers an enhanced user experience through flexibility, accessibility and practical demonstration, ensuing enriched engagement with the learning material (Rohde et al. 2024:122).
A study conducted by Huang and Lee (2024) found that the use of AI technology has seemingly provided better results than the traditional methods of content delivery and teaching. Ekechi et al. (2024:1262) visually present five grouped qualities and benefits of AI-enabled chatbots, depicted in Figure 1, which include: (1) properties of scalability and flexibility of AI-enabled chatbots, (2) cost-saving benefits because of reduced human involvement, (3) accuracy and quality of outputs, (4) advanced analytics and security enabled through algorithms and learning models and (5) productivity and efficiency achieved through automation.
 |
FIGURE 1: Schematic of chatbots’ quality and benefits. |
|
While AI-enabled chatbots are effective when the user experience interacting with the chatbot is positive, Rana and Jain (2024:332) caution that poorly designed chatbots that fail to meet users’ needs can drive them back to traditional support channels. Bhandare (2024:2) emphasises the importance of human experts in validating AI-generated material for accuracy and relevance, which results in achieving superior training programmes.
Challenges
A study conducted by Vij and Agarwal (2024:42) found that while the personalisation of learning content offered by AI tools had a positive outcome on academic performance, the lack of proper infrastructure, including hardware, software and effective training of facilitators, could counteract the benefits achieved. Eden, Chisom and Adeniyi (2024:10) emphasise infrastructure readiness as a concern and underline the importance of investing in resources and development, adhering to regulations and legislation and prioritising inclusive design and development to prevent bias. Challenges posed by algorithmic bias can result in inaccurate responses that may disadvantage minorities; thus, it is imperative that AI models are trained using balanced datasets (Elam 2024:88).
Artificial intelligence tools are largely dependent on the availability of data to function effectively, thus requiring the collection of sensitive student data such as academic reports, learner behaviour and personal information (Mimoudi 2024:277), posing data privacy risks. Tahir et al. (2024:1472) concur that privacy can be compromised because of the susceptibility and overexposure of children through facial recognition and recommender programmes. Furthermore, Mimoudi (2024:276) highlights the practical limitations educators face in adopting new complex technologies because of over-commitment with inadequate resources, which may hinder the adoption and integration of AI technologies into their stretched instructional responsibilities.
Implementation science
Implementation science (IS) is a framework focused on the promotion of systemic adoption into real-world settings through systematic and pedagogical evidence (Chu 2024:281) and describes the IS framework as effective in its means to understand how to integrate systems into practice and analyse best practices to encourage implementation and sustainability of the system in an organisation. The unique appeal of IS, compared to other models that assess systemic acceptance, is that it also focuses on evidence-based long-term implementation and successful adoption strategies for new technologies (Ramanadhan et al. 2024:47). Although predominantly used in health sector research (Shelton & Brownson 2024:174), the IS frameworks can be deemed suitable for use in other areas as well. The integration of IS use into different sectors is supported in a study conducted by Younas and Reynolds (2024:115), which highlights the effective bridging of AI and implementation strategies to support IS-leveraged outcomes. Figure 2 presents the types of outcomes in the application of IS research.
 |
FIGURE 2: Types of outcomes in implementation research. |
|
Chays-Amania, Schwingrouber and Colson (2024:8) argue the effectiveness of the model presented by Proctor et al. (2011:66) in determining the use of IS to implement solutions based on empirical evidence rather than conceptualised evidence. Furthermore, they support their argument using the eight implementation research outcomes, namely: (1) acceptability based on the satisfaction of the stakeholders, (2) adoption based on user assimilation of the innovation, (3) how the innovation appropriates to the need and context of the organisation, (4) the financial costs related to the implementation, (5) the feasibility of the innovation in terms of use in targeted settings, (6) fidelity that looks at the accuracy of implementation according to the planned specifications, (7) how well the innovation has penetrated the environment often determined by the uptake of the innovation and (8) assessing the longevity and sustainability of the innovation once in use (Chays-Amania et al. 2024:8). The dimensions of this model are considered helpful in the iterative refinement associated with AI technologies (Reddy 2024), leading to continuous improvement.
As a subset of the IS framework, the Consolidated Framework for Implementation Research defines five domains that influence implementation within an organisational setting, namely: (1) intervention characteristics, (2) inner settings, (3) outer settings, (4) individuals involved and (5) the implementation process (Damschroder et al. 2009; Daniels et al. 2022:936). Organisations can better understand the successful adoption and integration of new technologies through the structured approach offered by the mentioned domains.
Research methods and design
Qualitative research explores the behaviours and experiences of participants (Islam, Mazyed & Aldaihani 2021:2). The qualitative approach allows for researchers to gather and analyse data from the interactions with the participants through research techniques such as interviews, observations and focus groups (Alamri 2019:65). The case study design was selected in this study to contextualise a prototype of an AI tool, understanding its impact on improvement and performance in a real-world setting and how it can be applied for customer service training. The case study design was selected because of its usefulness when adopting a systems approach (Gammelgaard 2017:910) and its ability to grasp the realistic dynamics of a phenomenon using a selected sample of participants (Lavarda & Bellucci 2022:540). The data were collected from a SMEs organisation, Company X, allowing the researcher to gain insight into the behaviour of its customer service representatives. This single case study approach enabled a detailed exploration of the effectiveness of the AI chatbot in addressing the gaps identified in the organisation’s training. Purposive sampling was chosen to select participants based on their direct involvement with the chatbot prototype. Nyimbili and Nyimbili (2024:98) advocate that this sampling method involves selecting a sample from the population based on the researcher’s knowledge and experience related to the topic.
The study employed the experimental use of an AI chatbot prototype, followed by semi-structured interviews, acting as the primary data collection method. Semi-structured interviews allow for the combination of structured and unstructured interview formats to achieve flexibility in the conversation (Thille, Rotteau & Webster 2021:546). As concurred by Balushi (2018:728), semi-structured interviews enable depth to be reached with the participants so as to grasp the context from their responses fully.
An AI chatbot prototype was developed and deployed for use by the 10 participants selected for the study. While the study does not explicitly focus on the development of the AI chatbot prototype, the pilot prototype was tested with participants to aid in the framework development in this study. The prototype was developed on a platform called Poe, powered by GPT-3.5-Turbo as the base bot. The Poe platform incorporates a knowledge base feature that enables the upload of documents, including PDF and Microsoft PowerPoint training documents from Company X. The concurrent use of the generative base bot and established knowledge base ensures that responses are tailored to organisational-specific processes and information. The participants were required to use this chatbot for a week-long period in conjunction with their daily tasks and expectations. Each participant interacted and consulted with the AI chatbot at their discretion when assistance was required. Thereafter, each participant was interviewed to gather data on their experience following the use of the AI chatbot that was integrated into the systems they use every day. Data were collected post-assessment through semi-structured interviews, following the use of the prototype.
Ethical considerations
Ethical clearance to conduct this study was obtained from the Faculty of Informatics and Design Research Ethics Committee of the Cape Peninsula University of Technology (Ref. No. 216153972/2024/2).
Results
The study aimed to evaluate the effect of using an AI chatbot in the customer service training environment and assess the AI tool’s viability as an enhancement to human-facilitated training techniques in terms of product and business knowledge. An AI chatbot prototype was deployed in a chosen SME, followed by semi-structured interviews, which enabled the researcher to analyse the experiences, benefits and challenges encountered by the agents through the lens of the IS framework. The findings revealed the benefits of integrating an AI-enabled chatbot in the customer service environment of an SME. The improved accessibility to information contributed to enhanced productivity, increased confidence in addressing customer queries and overall efficiency in daily tasks. Furthermore, the chatbot supported personalised learning journeys while ensuring consistency in service delivery with standardised outputs.
The findings revealed the benefits realised from integrating and utilising the AI-enabled chatbot in the customer service environment of an SME. The improvement was noted in the accessibility of information, which resulted in enhanced productivity, increased confidence in addressing customer queries and overall efficiency in daily tasks. The chatbot allowed for personalised learning journeys while maintaining consistency across participants’ service delivery through standardised outputs.
The analysed data contributed to the development of the Assess, Build, Implement, Refine, Maintain (ABIRM) framework, which provides a practical guide for organisations to ensure the successful implementation of AI-enabled chatbots in training environments. Aligning with agile principles for effective implementation and scaling, the framework presents various elements for consideration in this regard. Furthermore, the framework highlights the importance of considering organisational readiness, how the new tool can impact existing processes and systems, the technical preparedness of the organisation to onboard an innovation and a phased yet structured rollout approach.
Table 1 shows the key themes identified during the data analysis and their relationship to the IS framework domains. The table depicts how the overarching IS framework acts as a guide to understanding how the key themes identified in this study can be categorised within the framework’s domains. The relationship between these categories illustrates the connection between the theoretical model and the empirical evidence.
| TABLE 1: Analysis of the relationship between the Implementation Science framework domains and the key themes identified in the study. |
Implementation science aims to promote the uptake of empirical findings that will lead to the improvement of people’s lives (Wiltsey Stirman & Beidas 2020:1034), aligning with the positive impacts seen in this study. Moreover, the framework provides a helpful view of how AI tools can enhance the way training is facilitated and enable improved performance among customer service representatives.
The ABIRM framework
The ABIRM framework was developed based on the data analysis from the study (see Figure 3). This framework introduces five scalable phases that present a practical guideline for SME organisations to adopt, implement and utilise an AI chatbot for customer service training, namely:
- Assess
- Build
- Implement
- Refine
- Maintain.
 |
FIGURE 3: The ABIRM framework for strategic artificial intelligence adoption in small- and medium-sized enterprise organisations. |
|
The ABIRM is intended to structurally guide organisations in implementing and adopting AI technologies in the customer service realm, ensuring the success, scalability and sustainability of the practical rollout within the organisation. Furthermore, it supplies organisations with strategic direction to ensure the effective integration and longevity of the AI tool.
Assess
The ABIRM framework encourages the initial assessment of the current state and appetite of the organisation for implementing and adopting AI technologies in its learning and development structure and policy. The as-is assessment requires reviewing existing training methods to identify current pain points and areas for improvement that can be easily addressed with the use of an AI-technology chatbot. In conjunction with the gaps identified in current training methods and modes of delivery, organisations must also assess their technological readiness to adopt innovations, such as AI tools. Supported by literature, Ogunyemi and Johnston (2012:5) concur that an organisation’s readiness to adopt new technologies is influenced by the state of its existing technological infrastructure, the availability of resources for scaling and the enthusiasm to adapt in order to support the new implementation in a live environment.
Configure
The assessment phase directly feeds into the to-be design, where organisations must consider the plan to build and configure the AI-technology chatbot in order to meet the business requirements arising from the first stage. Organisations must ensure that the build meets the requirements and is developed considering the actual business need and the workforce that will ultimately utilise the tool. The ABIRM framework recommends developing and configuring the tool in a way that supports adaptive learning and ensures personalisation of learning journeys to support autonomy among the workforce. Aggarwal et al. (2023:9) emphasise that the nature of AI chatbots allows for real-time feedback; thus organisations must design with this feature in mind. Real-time feedback mechanisms enable employees to assess their level of understanding and identify areas for personal improvement. Furthermore, by developing with the business need and workforce in mind, organisations can ensure that the tool meets requirements and is effective in the training environment, rather than just implementing a tool for the sake of keeping abreast with new technology.
Implement
The ABIRM framework offers recommendations for a phased implementation strategy, ensuring the gradual introduction of AI-technology chatbots to mitigate risks associated with large-scale rollouts. This is supported by a principle in the Agile Manifesto that advocates for working software over comprehensive documentation (Kakar 2023:21). The risks reduced by this approach include minimising disruptions to current workflows and processes, as well as the strain placed on technological resources. The phased implementation approach enables organisations to identify any challenges and detect any vulnerabilities that arise in the initial rollout stages, allowing for the re-evaluation of the rollout plan based on the outcomes. The implementation phase encourages employee support mechanisms, namely:
- ongoing engagement with the workforce,
- tool navigation training sessions and
- continued encouragement of tool use.
The study recommends facilitating an AI engagement forum during the initial rollout stages to provide a platform for employees to raise issues they encounter and make suggestions for improvement while also ensuring that employees feel heard and supported.
Refine
In the ABIRM framework, the refinement step enables organisations to build on the results from the initial adoption phase. Organisations can assess whether the AI tool has successfully addressed their needs and met the intended objectives of the implementation. The effective analysis of the data produced during the adoption and rollout phases enables organisations to address the challenges that employees face when using the tool. Furthermore, the data can assist in the effective prioritisation of fixes and bugs in accordance with the actual need and user experience. Aligning with the Agile Manifesto principles, organisations must respond to change rather than following a predetermined plan (Kakar 2023:21).
Maintain
Fostering a culture of continuous improvement ensures that organisations remain relevant (Rodgers & Antony 2023:645), thereby confirming that the technological investment remains effective in the long term. According to the ABIRM framework, the success of the AI chatbot in the organisation’s training realm depends on the tool’s dynamic ability to evolve over time, which relates to the maintenance and updates of the existing knowledge base aligned with organisational changes, as well as the regular maintenance of software and hardware components. Effective maintenance of the tool ensures that the organisation’s investment is protected, that the tool is relevant in addressing the challenges faced and that it conforms to emerging trends. Moreover, the tool’s relevance fosters a culture of continuous learning among the workforce, as it is always up-to-date with the most recent and relevant information.
Discussion
The study primarily focused on the adoption phases of the AI-enabled chatbot in the learning and development realm. It did not proceed beyond the initial stages of the prototype’s use. The study used a relatively small sample size, which limited the depth that could have been realised with a larger sample size. Furthermore, the study was focused on the context of SMEs, which may affect the generalisability of the findings.
This study was conducted with a pilot group in an SME organisation. It is recommended that comparative studies be conducted across various business sectors with larger organisational contexts. Further research should be conducted with a focus on the long-term impact of using AI chatbots as a complement to enhance current customer service training methods. Moreover, the effect of using an AI chatbot on employee satisfaction, knowledge retention and employee performance should be explored. Finally, the application of the ABIRM framework should be studied to explore its effectiveness in implementing and adopting AI technologies within the corporate sector, with the goal of enhancing customer service training.
Conclusion
This study argued for the transformation of customer service training to address challenges and enhance training using chatbots. The integration of an AI chatbot prototype into an SME revealed several themes that align with the IS framework. The effectiveness of AI chatbots in enhancing customer service training was motivated by several key themes, including improved accessibility to information, time savings, increased productivity, the ability to address queries, provision of support to customer service representatives and fostering a culture of collaboration and knowledge sharing among agents.
The findings revealed that integrating an AI-enabled chatbot contributed positively to enhanced customer service training in SMEs through improving accessibility to information, boosting productivity and agent confidence and supporting personalised learning by offering a platform for independent learning. The IS framework was used as a lens to analyse the utilisation of an AI-enabled chatbot across five domains: intervention characteristics, outer settings, inner settings, individuals involved and implementation processes. The analysis helped identify key factors for the effective introduction and sustainability of an AI-enabled chatbot in a real-world setting. Through the theoretical lens, the analysis provided a structured understanding of how the tool was received, highlighting key themes in relation to the framework components.
Furthermore, the study identified key factors influencing the diffusion of AI tools and presented the elements that contribute to a practical framework for implementing and adopting customer service training in SMEs. The aspects for consideration include organisational readiness, technical infrastructure reliability, the perceived usefulness of the tool and the relevance of the content presented in the chatbot’s outputs. These findings informed the development of the ABIRM framework, which is proposed as a practical guide for SMEs to implement and roll out AI technologies. The framework proposes a sequential approach, following agile principles to assess, build, implement, refine and maintain AI-enabled chatbots for customer service training. The integration of the IS framework to aid in the development of the ABIRM framework addressed an identified gap in the literature while providing practical and scalable guidelines for AI-technology adoption.
The study offers practical and theoretical contributions to the field of AI within workplace learning. It lays the groundwork for continued innovation and deeper exploration of technological innovations in corporate training environments.
Acknowledgements
This article is based on research originally conducted as part of Raeesah Mc Niel’s master’s thesis titled, ‘Enabling Customer Service Training with an AI Technology Chatbot’, submitted to the Faculty of Informatics and Design, Cape Peninsula University of Technology in 2025. The thesis was supervised by Errol Roland Francke. The manuscript has since been revised and adapted for journal publication. The original thesis is currently unpublished and was not publicly available online at the time of publishing this article.
Competing interests
The authors declare that they have no financial or personal relationships that may have inappropriately influenced them in writing this article.
CRediT authorship contribution
Raeesah Mc Niel: Conceptualisation, Formal analysis, Investigation, Methodology, Project administration, Visualisation, Writing – original draft. Errol R. Francke: Supervision, Writing – review & editing. All authors reviewed the article, contributed to the discussion of results, approved the final version for submission and publication and take responsibility for the integrity of its findings.
Funding information
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
Data availability
The data that support the findings of this study are not openly available because of human data involvement and are available from the corresponding author, Raeesah Mc Niel, upon reasonable request.
Disclaimer
The views and opinions expressed in this article are those of the authors and are the product of professional research. They do not necessarily reflect the official policy or position of any affiliated institution, funder, agency or that of the publisher. The authors are responsible for this article’s results, findings and content.
References
Aggarwal, A., Tam, C.C., Wu, D., Li, X. & Qiao, S., 2023, ‘Artificial intelligence-based chatbots for promoting health behavioral changes: Systematic review’, Journal of Medical Internet Research 25, 1–17. https://doi.org/10.2196/40789
Alamri, W.A., 2019, ‘Effectiveness of qualitative research methods: Inter views and diaries’, International Journal of English and Cultural Studies 2(1), 65–70. https://doi.org/10.11114/ijecs.v2i1.4302
Al-Dahoud, A., Fezari, M., Al-Dahoud, A., Aqel, D., Mimi, H. & Daoud, M.S.H., 2024, ‘Revolutionising space: The potential of artificial intelligence’, WSEAS Transactions on Computer Research 12, 404–414. https://doi.org/10.37394/232018.2024.12.40
Ali, A.H., Dheyab, S.A., Alamoodi, A.H., Magableh, A.A.A.R. & Gu, Y., 2024, ‘Leveraging AI and Big Data in low-resource healthcare settings’, Mesopotamian Journal of Big Data 2024, 11–22. https://doi.org/10.58496/mjbd/2024/002
Alkan, A., 2024, ‘Artificial intelligence: Its role and potential in education’, Journal of the Human and Social Sciences Researcher 13(1), 483–497. https://doi.org/10.15869/itobiad.1331201
Alsaati, A.A.A., 2024, ‘Examining the application of artificial intelligence in computers’, Edelweiss Applied Science and Technology 8(6), 1258–1264. https://doi.org/10.55214/25768484.v8i6.2235
Ameen, N., Tarhini, A., Reppel, A. & Anand, A., 2021, ‘Customer experiences in the age of artificial intelligence’, Computers in Human Behavior 114, 106548. https://doi.org/10.1016/j.chb.2020.106548
Balushi, K.A., 2018, ‘The use of online semi-structured interviews in interpretive research’, International Journal of Science and Research (IJSR) 7(4), 726–732. https://doi.org/10.21275/ART20181393
Bhandare, V., 2024, ‘The role of artificial intelligence in learning and development’, International Journal of Scientific Research in Engineering and Management 8(5), 1–5. https://doi.org/10.55041/ijsrem34968
Burton, S.L., 2022, ‘Artificial Intelligence (AI): The new look of customer service in a cybersecurity world’, Scientific Bulletin 27(2), 79–82. https://doi.org/10.2478/bsaft-2022-0010
Cerezo-Martínez, P., Nicolás-Sánchez, A. & Castro-Toledo, F.J., 2024, ‘Analysing the European institutional response to ethical and regulatory challenges of artificial intelligence in addressing discriminatory bias’, Frontiers in Artificial Intelligence 7, 1–14. https://doi.org/10.3389/frai.2024.1393259
Chays-Amania, A., Schwingrouber, J. & Colson, S., 2024, ‘Using implementation science to implement evidence-based practice: A discursive paper’, Journal of Advanced Nursing 81, 8050–8061. https://doi.org/10.1111/jan.16571
Chu, F., 2024, ‘Implementation science: Why should we care?’, Journal of the Medical Library Association 112(3), 281–285. https://doi.org/10.5195/jmla.2024.1919
Conroy, S., 2023, Notion AI vs ChatGPT, viewed 20 April 2023, from https://www.wepc.com/tips/chat-gpt-vs-notion-ai/.
Damschroder, L.J., Aron, D.C., Keith, R.E., Kirsh, S.R., Alexander, J.A. & Lowery, J.C., 2009, ‘Fostering implementation of health services research findings into practice: A consolidated framework for advancing implementation science’, Implementation Science 4(1), 50. https://doi.org/10.1186/1748-5908-4-50
Daniels, S.I., Cheng, H., Gray, C., Kim, B., Stave, C.D. & Midboe, A.M., 2022, ‘A scoping review of implementation of health-focused interventions in vulnerable populations’, Translational Behavioral Medicine 12(9), 935–944. https://doi.org/10.1093/tbm/ibac025
Eden, C.A., Chisom, O.N. & Adeniyi, I.S., 2024, ‘Integrating AI in education: Opportunities, challenges, and ethical considerations’, Magna Scientia Advanced Research and Reviews 10(2), 6–13. https://doi.org/10.30574/msarr.2024.10.2.0039
Ekechi, C.C., Chukwurah, E.G., Oyeniyi, L.D. & Okeke, C.D., 2024, ‘AI-infused chatbots for customer support: A cross-country evaluation of user satisfaction in the USA and the UK’, International Journal of Management & Entrepreneurship Research 6(4), 1259–1272. https://doi.org/10.51594/ijmer.v6i4.1057
Elam, K.M., 2024, ‘Exploring the challenges and future directions of Big Data and AI in education’, Journal of Artificial Intelligence General Science 5(1), 82–93. https://doi.org/10.60087/jaigs.v1i1.173
Elamin, M.O.I., 2024, ‘AI through the ages: Unlocking key opportunities and navigating challenges in the history and future of artificial intelligence’, International Journal of Religion 5(12), 1152–1166. https://doi.org/10.61707/tarzcj89
Fowler, D., Stevenson, R. & Wilson, A., 2019, ‘Customer service training and its effect on employee to employee relationships and work climate: A case study’, Journal of Marketing Management Research 23, 1–12.
Gammelgaard, B., 2017, ‘Editorial: The qualitative case study’, International Journal of Logistics Management 28(4), 910–913. https://doi.org/10.1108/IJLM-09-2017-0231
Hidayat, M., Defitri, S.Y. & Hilman, H., 2024, ‘The impact of artificial intelligence (AI) on financial management’, Management Studies and Business Journal (PRODUCTIVITY) 1(1), 123–129. https://doi.org/10.62207/s298rx18
Huang, H.-S. & Lee, B.-O., 2024, ‘The effectiveness of applying artificial intelligence in sick children’s communication’, Bioengineering 11(11), 1097. https://doi.org/10.3390/bioengineering11111097
Islam, M.A., Mazyed, F. & Aldaihani, F., 2021, ‘Justification for adopting qualitative research method, research approaches, sampling strategy, sample size, interview method, saturation, and data analysis’, Journal of International Business and Management 5(1), 1–11. https://doi.org/10.37227/jibm-2021-09-1494
Jaman, A.U. & Hossain, S., 2020, ‘Effective training methodology for BCS (Administration): A comparison of South & East Asian countries’, Bangladesh Journal of Administration and Management 33(2), 5–30. https://doi.org/10.56379/bjam.v33i2.2
Jiang, Y., Liu, J. & Liu, Y., 2020, ‘The impact of artificial intelligence chatbots on customer satisfaction and loyalty’, Journal of Business Research 116, 614–625.
Kakar, A., 2023, ‘A rhetorical analysis of the agile manifesto on its 20th anniversary’, Journal of the Southern Association for Information Systems 10(1), 20–29. https://doi.org/10.17705/3JSIS.00030
Karyotaki, M., Drigas, A. & Skianis, C., 2024, ‘Mobile/VR/Robotics/IoT-based chatbots and intelligent personal assistants for social inclusion’, International Journal of Interactive Mobile Technologies 18(8), 40–51. https://doi.org/10.3991/ijim.v18i08.46473
Kurniawan, M.H., Handiyani, H., Nuraini, T., Hariyati, R.T.S. & Sutrisno, S., 2024, ‘A systematic review of artificial intelligence-powered (AI-powered) chatbot intervention for managing chronic illness’, Annals of Medicine 56(1), 2302980. https://doi.org/10.1080/07853890.2024.2302980
Lad, S., 2024, ‘AI-driven automation in custom manufacturing: Enhancing precision and efficiency in automotive components production’, Journal of Artificial Intelligence General Science (JAIGS) 5(1), 373–380. https://doi.org/10.60087/jaigs.v5i1.206
Lavarda, R.B. & Bellucci, C.F., 2022, ‘Case study as a suitable method to research strategy as practice perspective’, Qualitative Report 27(2), 539–555. https://doi.org/10.46743/2160-3715/2022.4296
Li, Y. & Miraj, M.R., 2022, ‘Using artificial intelligence (AI) for education development’, Unpublished Doctoral thesis, Xinyu University.
Magnoler, P. & Pacquola, M., 2016, ‘Approaches to training in companies’, Research on Education and Media 8(2), 42–50. https://doi.org/10.1515/rem-2016-0017
Mbavai, F.R., 2022, ‘Effect of customer services on organisational performance in Tanesco: A case of Moshi District in Kilimanjaro Region’, Unpublished Doctoral thesis, Institute of Accountancy Arusha.
Mimoudi, A., 2024, ‘AI, personalised education, and challenges’, Proceedings of the International Conference on AI Research (ICAIR) 4(1), 271–280. https://doi.org/10.34190/icair.4.1.3133
Mondal, D. & Lipizzi, C., 2024, ‘Mitigating Large Language Model bias: Automated dataset augmentation and prejudice quantification’, Computers 13(6), 141. https://doi.org/10.3390/computers13060141
Nassazi, A., 2013, ‘Effects of training on employee performance: Evidence from Uganda’, Unpublished Doctoral thesis, Vaasan Ammattikorkeakoulu University of Applied Sciences.
Nyimbili, F. & Nyimbili, L., 2024, ‘Types of purposive sampling techniques with their examples and application in qualitative research studies’, British Journal of Multidisciplinary and Advanced Studies 5(1), 90–99. https://doi.org/10.37745/bjmas.2022.0419
Nze, S.U., 2024, ‘AI-powered chatbots’, Global Journal of Human Resource Management 12(6), 34–45. https://doi.org/10.37745/gjhrm.2013/vol12n63445
Ogunyemi, A.A. & Johnston, K.A., 2012, ‘Towards an organisational readiness framework for emerging technologies: An investigation of antecedents for South African organisations’ readiness for server virtualisation’, Electronic Journal of Information Systems in Developing Countries 53(1), 1–30. https://doi.org/10.1002/j.1681-4835.2012.tb00378.x
Olorunsogo, T.O., Anyanwu, A., Abrahams, T.O., Olorunsogo, T., Ehimuan, B. & Reis, O., 2024, ‘Emerging technologies in public health campaigns: Artificial intelligence and big data’, International Journal of Science and Research Archive 11(1), 478–487. https://doi.org/10.30574/ijsra.2024.11.1.0060
Pandit, A.R. & Kirdat, T.V., 2024, ‘Impact of AI in the animation industry’, International Journal for Research in Applied Science and Engineering Technology 12(III), 2828–2837. https://doi.org/10.22214/ijraset.2024.59501
Proctor, E., Silmere, H., Raghavan, R., Hovmand, P., Aarons, G., Bunger, A. et al., 2011, ‘Outcomes for implementation research: Conceptual distinctions, measurement challenges, and research agenda’, Administration and Policy in Mental Health and Mental Health Services Research 38(2), 65–76. https://doi.org/10.1007/s10488-010-0319-7
Ramanadhan, S., Alemán, R., Bradley, C.D., Cruz, J.L., Safaeinili, N., Simonds, V. et al., 2024, ‘Using participatory implementation science to advance health equity’, Annual Review of Public Health 45(1), 47–67. https://doi.org/10.1146/annurev-publhealth-060722-024251
Rana, J. & Jain, R., 2024, ‘Utility and acceptability of ai-enabled chatbots on the online customer journey in e-retailing’, International Journal of Computing and Digital Systems 15(1), 323–335. https://doi.org/10.12785/ijcds/150125
Reddy, S., 2024, ‘Generative AI in healthcare: An implementation science informed translational path on application, integration and governance’, Implementation Science 19(27), 1–15. https://doi.org/10.1186/s13012-024-01357-9
Rodgers, B. & Antony, J., 2023, ‘In pursuit of a culture of continuous improvement: Scotland’s national ambulance service’, International Journal of Public Administration 46(9), 636–646. https://doi.org/10.1080/01900692.2021.2009857
Rohde, N., Flindt, N. & Rietz, C., 2024, ‘A case for AI to increase benefits and solve challenges of e-learning based on the experiences of pedagogical experts’, Proceedings of the International Conference on Advanced Research in Education, Teaching, and Learning 1(1), 115–126. https://doi.org/10.33422/aretl.v1i1.209
Rosa, A., 2023, ‘Organisational training in startups: The incubators perspective in turbulent times’, International Journal of Business and Management 18(4), 1–12. https://doi.org/10.5539/ijbm.v18n4p1
Sahu, D.K. & Grover, V., 2024, ‘Exploring the potential of generative artificial intelligence’, International Journal of Scientific Research in Engineering and Management (IJSREM) 8(4), 1–5, viewed 10 May 2025, from https://ijsrem.com/download/enhancing-reality-exploring-the-potential-of-generative-artificial-intelligence.
Sangwan, A. & Bansal, R., 2024, ‘A comprehensive review of artificial intelligence’, International Journal for Multidisciplinary Research 6(5), 1–6. https://doi.org/10.36948/ijfmr.2024.v06i05.28932
Shelton, R.C. & Brownson, R.C., 2024, ‘Enhancing impact: A call to action for equitable implementation science’, Prevention Science 25(Suppl. 1), 174–189. https://doi.org/10.1007/s11121-023-01589-z
Shrivastava, A., Pandey, A., Singh, N., Srivastava, S., Srivastava, M. & Srivastava, A., 2024, ‘Artificial intelligence (AI): Evolution, methodologies, and applications’, International Journal for Research in Applied Science and Engineering Technology 12(IV), 5501–5505. https://doi.org/10.22214/ijraset.2024.61241
Suntharalingam, H., 2024, ‘Enhancing digital learning outcomes through the application of artificial intelligence: A comprehensive review’, International Journal of Innovative Science and Research Technology 9(4), 718–727. https://doi.org/10.38124/ijisrt/ijisrt24apr530
Syihabuddin, M., Suraj, N.A., Hyldgaard, S.E., Kainth, N., 2025, ‘Mini review on knowledge management in higher education challenges and strategies for effective implementation’, Socio-Economic and Humanistic Aspects for Township and Industry 3(1), 199–206. https://doi.org/10.59535/sehati.v3i1.466
Tahir, M., Hassan, F.D. & Shagoo, M.R., 2024, ‘Role of artificial intelligence in education: A conceptual review’, World Journal of Advanced Research and Reviews 22(1), 1469–1475. https://doi.org/10.30574/wjarr.2024.22.1.1217
Thille, P.H., Rotteau, L. & Webster, F., 2021, ‘More than words: Methods to elicit talk in interviews’, Family Practice 38(4), 545–547. https://doi.org/10.1093/fampra/cmab043
Tian, Y., 2024, ‘The application and practice of artificial intelligence in the entertainment field’, Applied and Computational Engineering 110(1), 50–54. https://doi.org/10.54254/2755-2721/110/2024MELB0098
Trabelsi, M.A., 2024, ‘The impact of artificial intelligence on economic development’, Journal of Electronic Business & Digital Economics 3(2), 142–155. https://doi.org/10.1108/jebde-10-2023-0022
Vij, A. & Agarwal, P., 2024, ‘Assessing the challenges and opportunities of artificial intelligence in Indian education’, International Journal for Global Academic & Scientific Research 3(1), 36–44. https://doi.org/10.55938/ijgasr.v3i1.71
Wiltsey Stirman, S. & Beidas, R.S., 2020, ‘Expanding the reach of psychological science through implementation science: Introduction to the special issue’, American Psychologist 75(8), 1033–1037. https://doi.org/10.1037/amp0000774
Wu, S., 2024, ‘The role of artificial intelligence in modern finance: Current applications and future prospects’, Applied and Computational Engineering 120, 17–22. https://doi.org/10.54254/2755-2721/2025.18825
Younas, A. & Reynolds, S.S., 2024, ‘Leveraging artificial intelligence for expediting implementation efforts’, Creative Nursing 30(2), 111–117. https://doi.org/10.1177/10784535241239059
|
Hide
Show all
Email this article (Login required)
