Hanna Koskiniemi 

Improving customer service operations through 
the implementation of artificial intelligence tools 

 

 

 

 

 

 

 
  

Vaasa 2025 

School of  Technology and Innovations  
Master’s thesis in Information Systems 

Master’s program in Information Systems Science 



2 

UNIVERSITY OF VAASA 
School of  Technology and Innovations 

Author: Hanna Koskiniemi 
Title of the thesis:  Improving customer service operations through the implementa-

tion of artificial intelligence tools 
Degree: Master Science in Economics and business administration 
Discipline: Master’s program in Information Systems Science  
Supervisor: Duong Dang 
Year: 2025 Pages: 85 

 
ABSTRACT: 
 
This master’s thesis explores the integration of artificial intelligence (AI) tools into customer ser-
vice processes to improve efficiency, customer satisfaction, and operational effectiveness in 
companies. As AI continues to revolutionize industries, its applications in customer service has 
gained huge attention, because those are able to automate repetitive tasks, enhance customer 
interactions, and provide valuable insights. Many experts are hopeful about the continued de-
velopment of AI. Despite these advantages, AI implementation faces challenges, including sys-
tem integration, employee resistance and ethical concerns related to data privacy and fairness. 
In any case, trends show that the use of AI has increased and will continue to increase in the 
future, bringing significant economic benefits to companies. The aim of this research is to exam-
ine how AI tools can be used the most effectively in customer service and how companies can 
successfully manage their implementation projects. Qualitative research is used as a research 
method in this thesis. The research is done with interviews and a literature review. 
 
Companies use various AI tools in their processes, such as chatbots, virtual assistants, and pre-
dictive analytics. In order for these tools to achieve the best possible results in a company, a 
company needs to know what kind of tools serve their needs best and how to successfully im-
plement them. The findings of this thesis show that AI improves response times and efficiency, 
but it needs careful planning, employee training, and alignment with business goals.  
 
The findings of this thesis show that while AI tools offer significant potential in automating rou-
tine tasks, improving response times, and personalizing customer interactions, their successful 
implementation requires careful consideration of factors such as organizational readiness, data 
quality, and employee training. Human factors, such as employee resistance, must be taken into 
account in the implementation processes of AI tools. This can be prevented by openness and 
transparency of projects. 
 
The thesis ends with practical recommendations for businesses looking to use AI in customer 
service. It highlights the importance of taking a strategic approach to integrating AI, which in-
cludes setting clear goals, choosing the right tools, and managing the implementation process 
effectively. A strategic and well-executed approach enables companies to maximize the benefits 
of AI while minimizing challenges, ultimately enhancing customer satisfaction and driving busi-
ness success. 
 
 
 

KEYWORDS: AI Implementation, AI tools, Artificial Intelligence, Automation, Customer Ser-
vice 



3 

VAASAN YLIOPISTO 
Tekniikan ja Innovaatiojohtamisen yksikkö 

Tekijä: Hanna Koskiniemi 
Tutkielman nimi:  Asiakaspalvelun parantaminen tekoälytyökalujen avulla 
Tutkinto: Kauppatieteiden maisteri 
Oppiaine: Tietojärjestelmätieteen maisteriohjelma  
Työn ohjaaja: Duong Dang 
Valmistumisvuosi: 2025 Sivumäärä: 85 

 
TIIVISTELMÄ: 
 
Tässä pro gradussa käsitellään tekoälyn (AI) käyttöönottoa yritysten asiakaspalveluprosesseissa 
tehokkuuden, asiakastyytyväisyyden ja yrityksen toimintojen parantamiseksi. Tekoälyn käyttö 
asiakaspalvelussa on saanut paljon huomiota, sillä sen avulla voidaan automatisoida toistuvia 
tehtäviä, parantaa vuorovaikutusta asiakkaiden kanssa ja tuottaa arvokasta tietoa. Vaikka teko-
älyn käytöllä on paljon etuja, sen käyttöönottoon liittyy myös haasteita, kuten järjestelmien yh-
teensopivuus, työntekijöiden vastustus sekä tietosuojaan ja oikeudenmukaisuuteen liittyvät ky-
symykset. Tekoälyn käyttö on joka tapauksessa yleistynyt ja sen odotetaan kasvavan entises-
tään, tuoden yrityksille merkittäviä taloudellisia hyötyjä. Tämän tutkielman tavoitteena on sel-
vittää, miten tekoälytyökaluja voidaan hyödyntää asiakaspalvelussa mahdollisimman tehok-
kaasti ja miten niiden käyttöönottoprojektit voidaan johtaa onnistuneesti. Tutkimus toteutettiin 
laadullisilla menetelmillä, ja aineisto koostuu haastatteluista sekä kirjallisuuskatsauksesta. 
 
Yritykset hyödyntävät erilaisia tekoälyratkaisuja, kuten chatbotteja, virtuaaliavustajia ja enna-
koivaa analytiikkaa. Jotta yritys saa tekoälytyökaluista kaiken hyödyn irti, yrityksen on tunnistet-
tava mitkä työkalut sopivat sen tarpeisiin ja miten ne kannattaa ottaa käyttöön. Tutkimuksen 
tulokset osoittavat, että tekoäly voi parantaa asiakaspalvelun nopeutta ja tehokkuutta, mutta 
sen onnistunut käyttöönotto vaatii huolellista suunnittelua, henkilöstön koulutusta ja yrityksen 
tavoitteiden huomioimista. 
 
Vaikka tekoäly voi tehostaa toistuvia tehtäviä, nopeuttaa vastausaikoja ja parantaa asiakaskoke-
musta, sen käyttöönotto edellyttää esimerkiksi yrityksen valmiuden, datan laadun ja työnteki-
jöiden asenteiden huomioimista. Erityisesti työntekijöiden suhtautuminen vaikuttaa merkittä-
västi siihen, miten onnistuneesti tekoäly saadaan osaksi toimintaa. Avoimuus ja läpinäkyvyys 
voivat auttaa tämän haasteen hallinnassa. 
 
Tutkimuksen lopussa on käytännön suosituksia yrityksille, miten tekoälyä voidaan hyödyntää 
yrityksen asiakaspalvelussa. Tekoälyn käyttöönotossa on tärkeää asettaa selkeät tavoitteet, va-
lita oikeat työkalut yrityksen tarpeisiin ja johtaa niiden käyttöönotto huolellisesti. Hyvin suunni-
teltu ja toteutettu prosessi auttaa maksimoimaan tekoälyn hyödyt, vähentämään haasteita ja 
parantamaan asiakastyytyväisyyttä sekä yrityksen menestystä. 
 
 
 

KEYWORDS: Asiakaspalvelu, Automatisointi, Tekoäly, Tekoälyn käyttöönotto, Tekoälytyöka-
lut 
 
 
 



4 

Contents 

1 Introduction 7 

1.1 Background 8 

1.2 Research problem 10 

1.3 Research questions and structure of the thesis 12 

2 Literature review 14 

2.1 Artifial intelligence in business 14 

2.2 Customer service processes and their challenges 22 

2.3 AI tools for customer service 24 

2.4 Key factors in successful AI implementation 34 

2.4.1 Technical factors 35 

2.4.2 Organizational factors 37 

2.4.3 Human factors and ethical considerations 41 

2.4.4 Leadership in AI implementation projects 45 

3 Research Methodology 48 

3.1 Research design 48 

3.2 Data collection 49 

3.3 Data analysis 52 

3.4 Ethical considerations 54 

4 Findings 55 

4.1 Interview results 55 

4.1.1 Theme 1. AI’s role in customer service 56 

4.1.2 Theme 2. Challenges in AI implementation 58 

4.1.3 Theme 3. Best practises for AI adoption 60 

4.2 Public attitudes towards AI 64 

5 Discussion 67 

5.1 Theoretical and practical implications 67 

5.2 Answering the research questions 70 

5.3 Limitations and future research 72 



5 

6 Conclusion 74 

References 75 

Appendices 83 

Appendix 1. Interview Questions 83 

Appendix 2. Acknowledgement of generative AI use 85 

  



6 

Pictures 
 
Picture 1 Security Policy and National Defence Survey 2021 65 

Picture 2 EVA Value and Attitude Survey Spring 2022 66 

 
 

Figures 
 
Figure 1. Thesis structure 13 

Figure 2 Visual representation of research process 52 

 
 

Tables 
 
Table 1. Comparison of AI tools for customer service 31 

Table 2. Types of business objectives 37 

Table 3. Steps in implementing AI tools 38 

Table 4 The key themes from the interviews 55 

Table 5 Findings summary table 62 

 
 

Abbreviations 
 
AI  Artificial Intelligence 
B2C  Business- to- Consumer 
NLP  Natural Language Processing 
ML  Machine Learning 
GDPR  General Data Protection Regulation 
CCPA  California Consumer Privacy Act 
IT  Information Technology 
EU  European Union 
 
  



7 

1 Introduction 

Artificial Intelligence (AI) has changed the operations of companies all over the world 

and this change continues. Artificial intelligence can analyze huge amounts of data, learn 

from this data, and make decisions based on it. This ability makes AI an important tool 

in business operations. AI increases efficiency, innovation and customer satisfaction. 

Along with other business operations, AI is particularly useful in customer service.  

 

Today, there is a lot of competition in the business world and companies need to respond 

to customer wishes and needs quickly and efficiently. Effective customer service in com-

panies plays a key role in increasing a company’s reputation and success, and companies 

are under the pressure to make it even faster and more easily accessible. AI tools, such 

as chatbots, virtual assistants, and analysis software, offer the companies the oppor-

tunity to respond to customer wishes and needs by automating routine tasks, improving 

response times, and providing insights that make customer experiences better.  

 

However, despite all these benefits, there may also be challenges when implementing 

an AI tool in a company. It is difficult for companies to choose right tools for their busi-

ness needs and these tools can be difficult to integrate into their workflows. Managing 

AI tools implementation project can also be challenging. The companies must have deep 

knowledge of how AI tools can be effectively implemented in customer service processes 

so that their positive impact is maximized.  

 

This thesis explores the use of AI tools in improving company’s customer service pro-

cesses, focusing on how these tools can be applied most effectively and how their im-

plementation projects can be managed successfully. The thesis will review the literature 

and scientific studies on this topic. The purpose of the thesis is to answer the research 

questions and, in addition, to produce practical suggestions on how companies can im-

prove their customer service processes with the help of AI tools.  

 



8 

1.1 Background 

Artificial intelligence is the simulation of human intelligence in machines that are pro-

grammed to think, learn, and make decisions. In business operations, AI has become a 

powerful tool which is driving innovation and efficiency. It is increasingly used to auto-

mate repetitive tasks, analyze extensive datasets, and support decision-making pro-

cesses, and these ways enhancing productivity and accuracy across industries (Perifanis 

& Kitsios, 2023, p. 2). 

 

The COVID-19 pandemic has done socio-technical, political-economic, and demographic 

changes faster, requiring businesses to manage market shifts and customer behaviour 

effectively. That is one reason why artificial intelligence has become important tool for 

businesses. AI enables organizations to improve operations and decision-making 

through automation and advanced analytics, helping them navigate an uncertain busi-

ness environment (Perifanis & Kitsios, 2023, p. 2). 

 

AI-tools can be utilized in various ways, such as AI-driven decision-making for loans and 

sales forecasting and automating manual processes. These technologies are particularly 

effective in situations where humans and AI work together to enhance productivity. De-

spite its potential, the implementation of AI requires careful planning to ensure strategic 

alignment and achieve effective outcomes. (Perifanis & Kitsios, 2023, p. 2). 

 

AI and analytics are game changers in addressing the challenges of modern business. 

However, adopting AI still faces obstacles, such as ensuring alignment between business 

users and analytics practitioners and establishing governance mechanism to manage AI’s 

implementation responsibly. AI governance, which includes tools and methods to quide 

AI development and application, is essential to ensure compliance with legal and ethical 

standards (Perifanis & Kitsios, 2023, p. 2). 

 

Many organizations are still in the experimental phase of AI adoption. They often begin 

with pilot programs and invest heavily in machine learning and other AI technologies. 



9 

These efforts aim to automate processes, improve efficiency, reduce costs, and achieve 

better market outcomes. However, to fully leverage AI’s potential, it must be integrated 

with other organizational systems and workflows. Despite its transformative capabilities, 

AI is not a standalone solution for competitive advantage. Businesses must combine AI 

with dynamic capabilities and adaptive strategies to unlock technological opportunities, 

drive innovation, and create lasting value (Perifanis & Kitsios, 2023, p. 2). 

 

AI has significantly impacted customer service, transforming the way organizations in-

teract with their clients. One of its primary jobs is automating responses through chat-

bots (Huang & Rust, 2018, p. 156). These tools, which are powered by natural language 

processing (NLP), can handle a wide range of customer inquiries, providing instant re-

sponses and allowing human workers to focus on more complex tasks (Perifanis & Kitsios, 

2023, p. 23). In addition to automation, AI enables personalization by leveraging ma-

chine learning algorithms to analyze customer preferences and tailor recommendations 

accordingly. For example, streaming platforms like Spotify and Netflix use AI to deliver 

personalized content suggestions. AI can also analyze customer emotions and provide 

proactive responses, enhancing customer satisfaction (Huang & Rust, 2018, p. 161). 

 

The increasing use of AI tools in customer service is result from customer expectations. 

Today’s consumers demand fast, personalized, and seamless interactions. AI meets these 

expectations by offering scalable and efficient solutions, helping organizations to stay 

competitive in a fast-changing business environment (Cao et al., 2024, p. 9). Technolog-

ical advancements, particularly in NLP and deep learning, have expanded AI’s capabilities. 

For example, AI-driven virtual assistants like Amazon’s Alexa (Amazon, n.d.) and Google 

Assistant (Google, n.d.) not only answer questions but they can also manage complex 

tasks like scheduling and shopping. This makes AI tools even better and more appealing 

to businesses (Cao et al., 2024, p. 5). 

 

AI tools address common challenges in customer service, such as long response times, 

inconsistent service quality, and high operational costs. By automating repetitive tasks, 



10 

organizations can allocate resources to more strategic targets. In addition to this, AI is 

revolutionizing social media marketing by enabling personalized customer interactions 

and image processing. With AI tools businesses can predict customer behaviour, use 

tools like Marketing Cloud to understand and shape it, and leverage AI-powered chatbots 

and virtual influencers to align quickly with customer needs. (Omeish et Al., 2024). 

 

Despite its benefits, the implementation of AI tools in customer service does not come 

without challenges. One significant challenge is integrating these tools into existing sys-

tems. The company’s infrastructure may lack the compatibility what is required for seam-

less AI implementation and that may increase expenses for the tool. Employee resistance 

is another common obstacle. As AI automates routine tasks, some employees may fear 

losing their jobs, which is making implementation project more difficult. To address 

these concerns businesses must invest in effective communication and training pro-

grams to help employees collaborate with AI systems. (Huang & Rust, 2018, p. 156). 

 

1.2 Research problem  

In today’s highly competitive world, effective customer service is essential for businesses 

which are seeking to retain customers and build loyalty. However, traditional customer 

service processes often struggle with inefficiencies, high operational costs, and the ina-

bility to provide consistent, personalized experiences across large customer bases (Mid-

lands Technical College, 2022). As a result, organizations are increasingly exploring the 

potential of artificial intelligence (AI) tools to address these challenges. Despite the 

growing use of AI in business context, there is a lack of comprehensive understanding 

about what kind of tools should they use and how these tools can be optimally inte-

grated into customer service processes and managed effectively.  

 

The research problem in this thesis is twofold: 

1. How can companies use AI tools in customer service operations effectively? 

2. What factors impact the implementation and usage of AI tools in customer ser-

vice operations? 



11 

 

While AI tools such as chatbots, virtual assistants, and predictive analytics have shown 

significant potential in automating and enhancing customer service, their success de-

pends on multiple factors, including organizational readiness, data quality, and user ac-

ceptance (Manyika & Bughin, 2018). Furthermore, as Fountaine et al. (2021, p. 18) define, 

failed implementation projects are often result from following things: 

1.  Companies lack a clear understanding of advanced analytics. 

2. Companies don’t assess feasibility, business value, and time horizons. 

3. Companies have no strategy beyond a few use cases. 

4. Companies don’t clearly define key roles. 

5. Companies lack experts who can bring business and analytics together. 

6. Companies isolate analytics from the business. 

7.  Companies squander time and money on enterprise-wide data cleaning instead 

of aligning data consolidation and cleanup with their most valuable use cases. 

8. Companies fully build out analytics platforms before identifying business cases. 

9. Companies neglect to quantify analytics’ bottom-line impact, lacking a perfor-

mance management framework with clear metrics for tracking each initiative. 

10. Companies fail to focus on ethical, social, and regulatory implications. 

 

The aim of this thesis is to explore the use and implementation of AI tools in customer 

service processes to improve their efficiency, scalability, and customer satisfaction. The 

specific parts in this thesis are identifying effective ways to utilize AI tools in customer 

service processes, exploring key success factors for AI implementation projects and find 

recommendations to improve the AI implementation projects. 

 

The research is very relevant today. According to a report by McKinsey & Company, AI 

adoption has grown significantly in customer service functions. According to this report, 

nearly half of responders in a 2018 McKinsey survey on AI adoption say their companies 

have embedded at least on AI capability in their business processes, and other 30% are 

piloting AI. Still, only 21% say their organizations have embedded AI in several parts of 



12 

business, and only 3% of large firms have integrated AI across their full enterprise work-

flows. This research was done of more than 400 use cases across 19 industries. (Manyika 

& Bughin, 2018). Also, Companies like Amazon, and Bank of America have already started 

leveraging AI to deliver better customer experiences (Amazon, n.d; Bank of America, 

n.d.). 

 

However, there are many things to consider when implementing AI tools, and the imple-

mentation does not always go smoothly. According to McKinsey & Company survey: AI 

adoption advances, but foundational barriers remain (2018), when asked about the big-

gest challenges to AI adoption, responders told that the most common barrier is strategy 

related. Companies don’t often have a clear AI strategy, appropriate talent or functional 

silos that constrain end-to-end AI solutions. Companies also often do not have leaders 

who demonstrate ownership of and commitment to AI. These statistics are showing that 

companies need systematic research about the best practices for AI adoption in cus-

tomer service.  

 

1.3 Research questions and structure of the thesis 

This thesis will explore the integration of artificial intelligence (AI) tools in customer ser-

vice processes. Thesis has two research questions: 

1. How can companies use AI tools in customer service operations effectively? 

This question involves identifying the optimal strategies for implementing AI 

tools to enhance customer service efficiency and effectiveness. By studying liter-

ature and investigating AI tools and their user experiences, the best tools and 

their implementation strategies in companies are identified. 

2. What factors impact the implementation and usage of AI tools in customer ser-

vice operations? 

This question focuses on the management aspects of integrating AI tools into 

customer service operations. It involves identifying the best practices for manag-

ing AI implementation projects. It also takes into account the challenges associ-

ated with implementing an AI tool, such as employee training and system 



13 

integration. The question also considers the ethical implications and data govern-

ance issues associated with the implementation of artificial intelligence in cus-

tomer service. 

 

 

Figure 1. Thesis structure 

 

This thesis is structured into six chapters. Figure 1 clarifies the structure of the thesis. 

The figure shows the structure of the thesis by main sections. Next to the main sections 

is a description of what the section contains. 

 

 



14 

2 Literature review 

The literature review in this thesis explores what AI is, where it comes from, and the 

factors that affect its use. The chapter looks at the history of AI, discusses trends and 

megatrends related to it, and explains how AI is used in business. It also examines the 

challenges in customer service and how AI can help solve them. Additionally, the chapter 

explores different AI tools and the key factors for successfully implementing AI in busi-

ness. 

 

2.1 Artifial intelligence in business 

Artificial intelligence is the ability of the machine to make decisions (Maisel et al., 2022, 

p. 4). It is an area of computer science that focuses on developing systems which are 

capable of performing task that typically require human intelligence. These tasks include 

problem-solving, decision-making, natural language processing, and visual perception. 

AI aims to copy cognitive processes such as learning and adaptation, enabling machines 

to analyze data, identify patterns, and make informed predictions or recommendations. 

For the thinking, computer needs natural language processing to communicate success-

fully in a human language, knowledge representation to store what it knows or hears, 

automated reasoning to answer questions and to draw new conclusions and machine 

learning to adapt to new circumstances and to detect and extrapolate patterns (Russell 

& Norvig, 2021, p. 19). In order for companies to make good decisions regarding AI tools, 

it is important to understand what these tools are and where they come from.  

 

The foundation of artificial intelligence (AI) was laid in 1943 by Warren McCulloch and 

Walter Pitts, who created a model of artificial neurons inspired the workings of the hu-

man brain. Their model proposed that neurons could be “on” or “off”, depending on its 

stimulus, and demonstrated that any computable function could be represented by a 

network of these neurons. They even theorized that these networks could learn. (Russell 

& Norvig, 2021, p. 35). 

 



15 

In 1950, Marvin Minsky and Dean Edmonds built the first neural network computer, 

called SNARC, using 3000 vacuum tubes. Around the same time, researchers like Chris-

topher Strachey and Arthur Samuel were creating some of the first AI programs, includ-

ing checkers-playing software. In addition to this, Alan Turing shaped the field with his 

article in 1950: “Computing Machinery and Intelligence”, where he wrote about Turing 

Test, machine learning, genetic algorithms, and reinforcement learning. Turing also 

warned about the risks of artificial intelligence to humanity (Russell & Norvig, 2021, p. 

35). 

 

In 1956 John McCarthy, Claude Shannon, Marvin Minsky, and Nathaniel Rochester orga-

nized a two-month AI workshop at Dartmouth College. This event brought together no-

table researchers like Allen Newell and Herbert Simon. The workshop didn’t lead to im-

mediate breakthroughs, but it built the groundwork of AI research (Russell & Norvig, 

2021, p. 36). 

 

Between 1966 and 1973 early AI systems succeeded in solving simple problems but they 

struggled with complex ones because of lack of task analysis and the limitations of prob-

lem-solving methods, which couldn’t handle the complexity of real-world scenarios. Ad-

ditionally, experiments in genetic programming produced little progress (Russell & 

Norvig, 2021, p. 39). 

 

In the 1980s, the back-propagation algorithm what was originally developed in the 1960s, 

was discovered again. Then happened a significant shift in AI research and applications. 

Since 1987, AI has become more flexible and since 2001 big data revolutionized AI by 

enabling the use of massive datasets, such as trillions of images, billions of hours of 

speech and video, and other large-scale information like genomic or social network data. 

These datasets have made AI capable of learning from big amounts of unlabelled data, 

which significantly improves tasks like image analysis. Research shows that increasing 

data size often leads to greater performance gains than refining algorithms. Big data has 

also played a big role in groundbreaking AI achievements, such as IBM Watson’s 2011 



16 

Jeopardy! victory, boosting AI’s commercial appeal and public interest (Russell & Norvig, 

2021, p. 41–44). 

 

Deep learning, a form of machine learning which uses a multiple layers of adjustable 

computing elements, got started in 2011, initially in speech and visual object recognition. 

A major breakthrough came in 2012 when a deep learning system performed better than 

previous methods in the ImageNet competition. Since then, deep learning has achieved 

remarkable results in areas like speech recognition, machine translation, medical diag-

nosis, and game playing. Its success brought interest in AI and trust that AI can deliver 

good results and perform well (Russell & Norvig, 2021, p. 45). 

 

By understanding the history of artificial intelligence, it is possible to identify a certain 

pattern in its development and, through those patterns, make predictions about how 

the development of artificial intelligence will continue in the future. As the history of AI 

has shown, each technological leap builds on the innovations, successes, and failures of 

previous eras. This historical perspective is crucial to ensure that in the future we design 

AI tools that align with societal needs and address future challenges (Sitra, n.d.). 

 

According to Sitra Megatrends 2023 (n.d.) the big change in the future includes well-

being challenges, the battle for democracy, economic foundations are cracking and com-

petition for digital power gears up. Technology is advancing rapidly, and new innovations 

are invented. Data collection and utilization are growing. In AI, key questions are in fu-

ture: how data can be collected and used and who holds the expertise, resources, algo-

rithms and patents. Environmental impact of new services is also one of the key ques-

tions. The European Union has taken steps to establish rules for AI and digital services in 

general, aiming to guide technological advancements toward more sustainable and so-

cially responsible directions (Sitra, n.d.). 

 

AI will undoubtedly remain at the centre of discussions in the coming years. At the same 

time, it is crucial to consider other technological developments and their connections 



17 

and wider impacts. Key areas include high-performance and quantum computing, fast 

wireless networks, and cybersecurity, which are essential for maximizing AI’s potential. 

Additionally, resilience has become increasingly important. With extreme weather 

events and cyberattacks rising, it is vital to prepare for scenarios where networks fail, or 

power is unavailable (Sitra, n.d.). 

 

Most experts are optimistic about ongoing progress of AI.  The median prediction sug-

gests that AI will reach human-level capabilities across a wide range of tasks within the 

next 50 to 100 years. In the next decade, AI is expected to contribute trillions of dollars 

annually to the global economy. However, some critics argue that achieving general AI 

could take centuries, and there are significant ethical concerns regarding fairness, equity, 

and potential harm caused by AI (Russell & Norvig, 2021, p. 1063). 

 

AI can be used for a variety of purposes. It can be divided into narrow AI which is de-

signed to perform specific tasks, and artificial general intelligence (AGI) which is still the-

oretical. Chatbots are a good example of narrow AI. The goal of AGI’s development is to 

create a thinking machine (Goertzel & Potapov, 2015, p. 3). 

 

In the business, AI applications have grown significantly. Companies are adopting AI to 

improve operational efficiency, reduce costs, and enhance customer experiences. AI is 

used for many different purposes, including automation of repetitive tasks, data analysis, 

decision-making, predictive analytics, and personalized customer service AI will augment 

human capabilities, freeing up workers to engage in more productive and higher-value 

tasks, and increase demand for jobs associated with AI (Manyika & Bughin, 2018). 

 

Artificial intelligence is based on several technologies, including machine learning (ML), 

natural language processing (NLP), and customer data analytics. These technologies are 

fundamental to understanding how AI can be effectively applied to business processes. 

In addition to these AI also includes deep neural networks, algorithms that learn and 



18 

adapt with AI, deep learning, image recognition, reason and decision automation and 

emotional intelligence (Bulusu & Abellera, 2021, p. 7–10). 

 

Machine Learning (ML) is a discipline focused on two interrelated questions: How can 

one construct computer systems that automatically improve through experience? And 

What are the fundamental statistical-computational-information-theoretic laws that 

govern all systems, including computers, humans, and organizations? Over the past two 

decades, machine learning has evolved from a theoretical concept into a widely used 

technology across various industries. It has become a very popular method for develop-

ing practical software in fields such as speech recognition, robot control, and customer 

service. In many cases, it is easier to train a machine by showing it examples of desired 

behaviour rather than programming all potential responses. The effect of machine learn-

ing has been felt across computer science and other industries, such as consumer ser-

vices, the diagnosis of faults in complex systems, and the control of logistics chains (Jor-

dan & Mitchell, 2015, p. 255). 

 

According to Jung (2022, p. 57), ML methods combine three main components:  

1. A set of data points that are characterized by features and labels 

2. A model or hypothesis space that consists of different hypotheses 

3. A loss function to measure the quality of a particular hypothesis 

 

Each of these three components have design choices for the representation of data, their 

features and labels, the model and loss function (Jung, 2022, p. 57). 

 

In practice ML can be used for example for helping financial companies, such as banks. 

It is possible to make an algorithm that can predict to whom the bank should give loans 

or credits. A bank always takes a risk when it lends money to a customer, so before lend-

ing money, the bank checks whether the customer has the ability to pay back their loan. 

Based on customer’s current income, many banks perform some kind of analysis that 

helps them decide whether the borrower will be a good customer for that bank or not. 



19 

This kind of analysis is manual and time consuming so automation could make it better. 

It is possible to develop an algorithm which will generate probability. This probability 

value indicates the chances of the case that borrowers can’t pay back their loans (Jalaj, 

2018, p. 2).   

 

Natural Language Processing (NLP) as a part of AI that enables communication between 

humans and computers. In customer service, NLP is important to applications, such as 

chatbots and virtual assistants. A chatbot interacts with customers in a natural language 

to respond to inquiries or perform transactions. Chatbots and virtual assistants are used 

because they are faster than humans and reduce costs by relieving human workers of 

repetitive tasks (Darlington Arinze, 2014, p. 33). 

 

NLP is widely used in customer service for sentimental analysis, where customer feed-

back, reviews or social media posts are analyzed to get public opinion or find out cus-

tomer satisfaction. While some of the consumer-generated content comes in the form 

of numerical values (e.g. reviews or rating scores), for example, the information posted 

on social media is in the form of text. This kind of text contains subjective evaluations, 

expressing so-called sentiment. There is a growing number of software companies that 

are offering sentiment analysis tools, but these tools have also some problems. While 

huge progress has been made in sentiment analysis due to advances in natural language 

processing, current approaches are still far from reaching human-like abilities. (Ge-

Stadnyk et al., 2017, p. 1–2). 

 

In today’s fast-evolving market, companies need to develop dynamic skills and tools to 

stay competitive. These capabilities give businesses the agility to quickly adapt and reor-

ganize their resources to seize new opportunities. The company’s IT capabilities help 

here when they are integrated with the company’s business strategies. However, it’s im-

portant for IT investments to be well-planned to ensure they don’t stop company’s ability 

to make changes quickly (Kitchens et al., 2018, p. 544). 

 



20 

Big data analytics has become a key area where IT can support competitive strategies. 

With the growing importance of big data for business survival, companies need to shift 

from using general analytics to specialized tools that target specific business needs. One 

major challenge is breaking down organizational silos to facilitate smooth data sharing 

across different departments. IT have a huge role in this, and that is crucial for driving 

innovation and maintaining a competitive edge (Kitchens et al., 2018, p. 544). 

 

The integration of data from various sources can be both expensive and complex. To 

effectively support big data analytics, companies must invest in robust data management 

and infrastructure. A focus on valuable strategies is essential to assess and integrate rel-

evant data, particularly for customer analytics, which can provide a competitive ad-

vantage. By building a solid IT infrastructure for data integration, businesses can maxim-

ize customer lifetime value and drive improvements in both customer acquisition and 

retention. For this to succeed, companies need to adopt a framework that blends busi-

ness strategy with IT strategy. This approach will help them create an infrastructure that 

not only supports advanced customer analytics but also generates lasting strategic value 

in the long run (Kitchens et al., 2018, p. 545). 

 

Understanding business ecosystems has become crucial for companies in today’s com-

petitive environment. To stay ahead, companies need to quickly identify competitive ac-

tions, trends, and investment opportunities. These things often need insights across in-

dustries and markets. While Structured data sources can assist in these analyses, they 

are typically fragmented and fail to provide a holistic view of the dynamic interactions 

within ecosystems. Additionally, some unstructured data, such as news articles, press 

releases, and industry reports, contains valuable insights but is difficult to analyze man-

ually due to the large amount of data (Basole et al., 2023, p. 1–2). 

 

The real challenge is making sense of the huge volumes of unstructured data. For exam-

ple, searching for information on strategic collaborations in artificial intelligence could 

bring up millions of documents, but going through all of them manually would be time-



21 

consuming and inefficient. Artificial intelligence (AI), especially through natural language 

processing (NLP) and advanced text mining methods, can help finding key concepts and 

terms from these documents. However, this is not enough. Understanding how these 

concepts relate to each other in the context of the broader ecosystem is crucial (Basole 

et al., 2023, p.1–2). 

 

Ecosystem intelligence involves human interpretation and judgement. While AI can help 

identify significant patterns and entities, decision-makers need a system that presents 

these findings in a clear and interactive visual format. This enables a deeper understand-

ing and quicker, more informed decision-making. So, AI tools for ecosystem intelligence 

should not just extract important data but also allow users to explore and understand 

how different pieces of data are connected, helping them find valuable insights (Basole 

et al., 2023, p.1–2). 

 

AI refers to technologies that mimic human cognitive functions, perform tasks similarly 

to humans, and have the ability to learn and improve over time. It is applied for example 

virtual agents, predictive analytics, recommendation systems, and speech analytics. In 

customer service, AI enhances customer experiences, automates processes, improves 

engagement, and gives new insights. AI has significantly changed customer service with 

three types of AI-enabled services: AI-supported services, where human agents use AI 

for decision-making or personalization, AI-augmented services, where AI interacts di-

rectly with customers or assists human agents, and AI-performed services, where AI 

completely replaces human agents in delivering services. Many studies have explored 

AI’s effects on customer behaviour and experiences, for example, how chatbot affect 

user compliance. Other studies focus on AI’s usability, service quality, and service types. 

AI transforms products and services, but it also changes business strategies and pro-

cesses. It enables companies to improve their operations and capabilities (Lan et al. 2024. 

p. 3). 

 



22 

2.2 Customer service processes and their challenges 

Exceptional customer service is increasingly recognised as a crucial contributor to busi-

ness success and sustainability in today’s highly competitive marketplace. Positive cus-

tomer experiences help boost revenues, reduce churn, minimise costs and build long-

lasting brand loyalty. With the shift towards service orientation, customer service has 

become a critical performance indicator and a key factor in determining a company’s 

overall performance and profit margins across various sectors. Satisfied customers are 

more likely to make repeat purchases and recommend the product or company to others, 

ultimately contributing to higher sales and revenue growth. Positive service climate and 

customer-centric culture lead also to increased employee satisfaction and commitment. 

Ensuring service quality is crucial for customer satisfaction and loyalty. This means that 

companies that prioritize addressing customer complaints and resolving issues quickly 

achieve higher customer satisfaction and loyalty. In contrast, poor complaint handling 

can lead to negative experiences and harm company’s reputation, directly affecting its 

performance (Buwah, 2024, p. 2119).  

 

For customers, the key factors that contribute to satisfaction in a service environment 

are connection, response time, and the quality of the response. Live chat customer ser-

vice systems help consumers make decisions by providing search assistance, guidance, 

and support. Customers often tend to overestimate how long they have been waiting for 

service, and they are particularly sensitive to delays when waiting online. Long wait 

times online, for example, create a negative service experience. On the other hand, 

when customers interact with a human-operated customer service system, they often 

have to endure queuing, connecting, and waiting for a response to solve their issues. AI-

powered customer service systems, however, can offer instant responses to customer 

problems at any time. This advantage has led to the adoption of AI customer service in 

industries. For instance, nearly all (99,78%) of Uniqlo’s customer service was AI-driven, 

and almost half (48%) of customers who used the AI service also made a purchase. Sim-

ilarly, in the banking sector, 6000 customers of China trust Inc. use AI-enabled customer 

service daily, and that is resolving 88% of their interactions (Xu et al., 2020, p. 2–3). 



23 

 

Artificial Intelligence (AI) has become a game-changer in improving customer service, 

helping companies tackle some of the biggest challenges they face in serving their cus-

tomers. As businesses recognize how important customer satisfaction is for success, AI 

has emerged as a key tool to ensure that service is faster, more personalized, and more 

efficient. One of the most common ways companies use AI in customer service is through 

AI-powered chatbots. These chatbots can handle a huge number of customer inquiries, 

work 24/7, and provide immediate responses. For example, DNB Bank has deployed AI 

chatbots to handle a large portion of customer interactions. By automating responses to 

frequent questions and simple requests, the bank has freed up human agents to focus 

on more complex issues. This not only shortens response times but also makes the whole 

process more efficient, and that benefits both customers and the bank. The use of chat-

bots in this way demonstrates how AI can affect customer service by making it quicker 

and more accessible for customers (Boost.ai, 2024). 

 

In addition to chatbots, AI has also improved voice-based customer service. In healthcare, 

for example, AI-powered voice chatbots are helping patients with tasks like booking ap-

pointments, getting pre-visit information, and accessing other healthcare services. These 

systems can understand voice commands, allowing patients to get help they need with-

out waiting for a human representative. This is particularly useful in healthcare, where 

phone calls and inquiries can quickly pile up. AI voice chatbots can handle these tasks 

efficiently and that reduces waiting times and makes services more accessible which ul-

timately enhances the patient experience (teneo.ai, 2024). 

 

AI is also effective in predictive customer analytics, especially in industries like banking. 

By analysing customer data, such as spending habits and transaction history, AI can pre-

dict what services or products a customer might need next. For example, banks use AI 

to recommend products like loans or investment opportunities based on a customer’s 

financial behaviour. This personalized approach increases the changes that customers 

will be interested the services what bank is offering. That improves loyalty and customer 



24 

satisfaction. Predictive analytics shows how AI can help businesses stay ahead of cus-

tomer needs and offer more tailored solutions that feel relevant to everyone (Abdul-

salam & Tajudeen, 2024, p. 32–34). 

Another application of AI in customer service is Agent Assist technology, which supports 

human agents during real-time customer interactions. Telecom companies, for example, 

have adopted AI systems that provide live suggestions to agents while they are talking 

to customers. These systems can recommend answers, retrieve customer histories, and 

offer other helpful insights, allowing agents to solve problems more quickly and accu-

rately. This technology helps agents resolve issues faster and more efficiently which is 

leading to better outcomes for customers. By enhancing the performance of human 

agents, Agent Assist tools improve overall customer service and satisfaction. Fo example, 

Vodafone implemented the TOBi chatbot to enhance customer support operations. TOBi 

helps customers with account management, technical support, and billing inquiries. TOBi 

handled over 2 million customer interactions within the first six months and it signifi-

cantly reduced waiting times and improved service efficiency (Uzoka et al., 2024, p. 

2501). 

 

In retail, AI-driven self-service options are increasingly popular. Companies like Uniqlo 

have integrated AI into their customer service system, and that is handling most interac-

tions through AI-powered tools. In Uniqlo AI system for example answer questions and 

recommend products. That gives customers fast and efficient service experience. This 

system has been incredibly successful. This shows that AI in self-service can help custom-

ers to get answers and also drive sales which is improving both customer experience and 

business results (Uniqlo, n.d.). 

 

2.3 AI tools for customer service 

The most used AI tools in customer service are chatbots, sentiment analysis, automated 

response systems, voice recognition AI, and predictive analytics. Each tool brings bene-

fits and challenges, and it is important that companies select the appropriate AI solutions 

that align with their specific needs (Uzoka et al., 2024, p. 2505–2506). 



25 

Recent studies support the effectiveness of AI tools in making customer service out-

comes better. AI implementation in customer service has improved both response speed 

and customer satisfaction, especially when it is used for repetitive tasks (Adam et al., 

2020, p. 429). AI technologies like predictive analytics and natural language prosessing 

enhance the capabilities of chatbots and other tools (Uzoka et al., 2024, p. 2506). AI tools 

and technologies improve customer service by increasing productivity, reducing re-

sponse times, automating tasks. These tools cand work 24/7 which also reduces costs. 

When businesses implement AI tools in a way that takes into account the protection of 

customer data, AI offers a big opportunity to improve customer service (Inavolu, 2024, 

p. 20). 

 

AI replaces human labour by taking over specific tasks, rather than entire jobs, starting 

with mechanical tasks and progressing to more complex, higher-level tasks such as ana-

lytical, intuitive, and empathetic tasks. Mechanical intelligence is the easiest for AI to 

replace, followed by analytical, intuitive, and empathetic intelligences. As AI develops, 

machines will take over more tasks that humans used to do. This will change the skills 

that workers need. For example, analytical skills may become less important, while skills 

like intuition and empathy will be more valuable, especially in jobs that involve market-

ing with people. In the future, AI might even be able to do tasks that need human em-

pathy (Huang & Rust, 2018, p. 155–157). 

 

Chatbots are perhaps the most recognized AI tool in customer service. These virtual as-

sistants are created to manage customer inquiries, offering quick answers to common 

questions and helping with basic troubleshooting. Some well-known chatbot platforms 

are Zendesk Chat, Drift, and Intercom. The key benefit of using chatbots is that they can 

provide 24/7 support, handling large numbers of customer interactions without needing 

a human agent. This is especially useful for businesses with customers in different time 

zones, because this ensures that service is available around the clock. Additionally, chat-

bots help lower operational costs and reduce human error by automating routine tasks. 

However, the technology has limitations. Chatbots primarily rely on pre-programmed 



26 

responses, which means they may struggle with queries that require human judgment 

(Lin, 2023, p. 3). 

 

Chatbots have gained popularity recently due to advancements in natural language pro-

cessing and machine learning, which allows them to respond in a more human-like man-

ner. A good example is ChatGPT, which offers an enhanced language model and user 

experience. Chatbots are also being used more in education, either as helpers to improve 

students’ understanding or as simulated students to support teachers (Lin, 2023, p. 3). 

 

Psychological research suggest that meaningful conversation often brings joy, which has 

contributed to the rise of social chatbots. For example, Microsoft’s Xiaoice is designed 

to offer communication, emotional support, and a sense of belonging, using empathy 

and personality to engage users better. Additionally, SeqGAN is a model for creating 

emotionally intelligent dialogues between humans and computers, but it hasn’t yet fully 

meet expectations in generating emotional responses (Lin, 2023, p. 3). 

 

Reinforcement Learning (RL) has also been applied to chatbots to improve response gen-

eration, dialogue management, and evaluation by maximizing rewards. While RL shows 

promise in making chatbots more natural and engaging, more research is needed for its 

use. For example, improving reward signals and training methods (Lin, 2023, p. 3). 

 

Sentiment analysis is emotion AI or opinion mining. It focuses understanding the senti-

ment behind unstructured data. For example, in the sentence “Liam likes apples”, Liam 

is the opinion holder, and his sentiment towards apples is positive. This emotional state 

is extracted from the text using techniques such as machine learning (ML), natural pro-

cessing (NLP), and lexicon-based methods. With the rise of the internet, the amount of 

data generated is growing rapidly. Data from 2020 onwards has surpassed all data gen-

erated before, and this trend is expected to continue. The amount of data being gener-

ated has grown significantly, mainly due to the rise of web applications and the world-

wide connectivity of the internet. The internet, which began in the 1970s for military use, 



27 

now has over 5,16 billion users. Social media platforms are reaching 4.62 billion users. 

This widespread use has led to a surge in people sharing opinions through blogs and 

social media posts. As a result, data generated online is a valuable source for various 

applications (Raghunathan & Kandasamy, 2017, p. 2). 

 

The growth of sentiment analysis is also closely tied to the increasing use of chatbots 

and virtual assistants. These tools use natural language processing (NLP) and sentiment 

analysis to interpret user queries and provide relevant responses. Their aim is to under-

stand the emotional state of the user. Over the past decade, sentiment analysis has ad-

vanced from traditional lexicon-based methods to more sophisticated machine learning 

and deep learning techniques, such as supervised machine learning, deep learning, and 

transfer learning. These improvements have made sentiment analysis more accurate and 

nuanced, and businesses get valuable insights into customer emotions and trends with 

them (Raghunathan & Kandasamy, 2017, p. 2). 

 

As Raghunathan and Kandasamy (2017, p. 2) present, sentiment analysis has various 

practical applications, which are helping businesses analyze large volumes of text data 

and gain insights for decision-making:  

 

1. Public opinion via social media. Before social media, organizations relied on me-

dia outlets to understand public opinion. Now, stakeholders have direct access to 

social media data, which allows them to collect and analyze opinions using tech-

nologies like Tweeby and TextBlob. This shift has made it easier to track public 

sentiment on various topics and detect fake news. This task would have been 

challenging using traditional methods like newspapers. 

2. Product analysis. As e-commerce has grown quickly, more customers are leaving 

reviews about products. This makes sentiment analysis important for under-

standing what customers think. By analyzing these reviews, sentiment analysis 

helps e-commerce sites and industries categorize opinions as positive, negative, 

or neutral, giving businesses useful feedback.  



28 

3. Stock market. The stock market reflects the state of the economy and is highly 

sensitive to real-world events. Sentiment analysis has been used to predict mar-

ket trends by analyzing public reactions to events like elections or disasters. Alt-

hough the stock market is volatile, sentiment analysis has shown potential in pre-

dicting market movements based on public sentiment. 

4. Enterprise management. As businesses face increasing competition, traditional 

management strategies are not enough for sustainable growth. Companies are 

using sentiment analysis to understand the emotional state of their employees 

and improve overall organizational efficiency. 

 

Sentiment analysis, or emotion AI, involves identifying, extracting, and quantifying the 

emotional state of subjective data. According to Raghunathan and Kandasamy (2017, p. 

2) it can be performed at three levels: 

 

1. Document level which determines if the overall sentiment of a document is pos-

itive, negative, or neutral. 

2. Sentence level which evaluates each sentence for sentiment and categorizes it as 

positive, negative, or neutral. 

3. Feature level which focuses on identifying the sentiment towards a specific fea-

ture or entity mentioned in the text. For example, in the sentence “The film had 

a good story”, “film” is the entity, and “story” is the feature. 

 

Sentiment analysis consist of two main stages: data pre-processing and classification. 

Pre-processing involves tokenization, which means converting words to their root forms 

through stemming or lemmatization and removing stop words. The second stage in-

volves applying a classification algorithm to the pre-processed data to categorize it ac-

cording to the sentiment detected (Raghunathan & Kandasamy, 2017, p. 2). 

 

Automated response systems manage common customer queries with different tools. 

The main advantage of automated response system is their ability to reduce response 



29 

times, which allows customers receive quick answers to their inquiries, even during off-

hours. A chatbot can handle routine tasks such as looking up order status or scheduling 

appointments. The most advanced bots use generative AI to personalize responses to 

customer questions. Companies can also automate email and social media autorespond-

ers. They can automatically acknowledge customer inquiries and provide initial infor-

mation or help. In addition to these, companies can use for example automated triaging, 

which means that human agent can categorize and route simple support requests to the 

right customer service agent for faster handling. Automated surveys and feedback re-

quests are one kind of tools to improve organization’s customer service. An automated 

system can send out a survey immediately after an interaction and gather critical insights 

from customers. AI can also give proactive support which is, for example notifying cus-

tomer of knowing issues like shipping delays or system outages (Service, 2025). 

 

According to Zendesk (2025), examples of automated customer service are: AI agents, 

IVR software, autoresponders, AI knowledge base, predictive analytics, ticketing systems, 

intelligent routing, automated notifications, workflow automation and automatic trans-

lation. Zendesk example of automated customer service gives a perspective on how an 

automated customer service situation works: 

 

Step 1.: Customer navigates to the support page on website. 

Step 2.: An AI agent greets customer and gathers information about problem. 

Step3.: The AI agent directs the customer to an article discussing the issue in company’s 

knowledge base. 

Step 4.: The article instructs customer to perform a certain task. If the task doesn’t work, 

the AI agent can offer a new solution or route the customer’s ticket to a human agent 

(Zendesk, 2025).  

 

There are different AI tools, and they all have their pros and cons. Table 1. Summarizes 

the most commonly used tools, their pros and cons, and their intended uses. One of the 

most common AI tools used today is chatbots. They offer real-time support by managing 



30 

various customer inquiries, such as answering frequently asked questions or helping 

with common issues. Popular platforms like Zendesk Chat, Drift, and Intercom allow busi-

nesses to engage with customers 24/7 in an efficient way. These bots ease the workload 

of human agents by taking care of repetitive queries. That way real human workers can 

focus on more complex and personalized issues (Lin, 2023, p. 3). 

 

Another important AI tool in Table 1. is sentiment analysis, which helps customers un-

derstand the emotional tone of customer interactions. Tools like MonkeyLearn, Lexalyt-

ics, and Aylien, use natural language processing and machine learning algorithms to an-

alyze large volumes of text data, including social media posts, emails, and customer re-

views. This tool is particularly beneficial for monitoring brand reputation and managing 

customer relations in real-time (Luo et al., 2021, p. 3). 

 

Automated response systems are another widely used AI solution. These systems allow 

businesses to automate replies to common inquiries which enables quick response times 

and reduces operational costs. Tools like Freshdesk, Zoho Desk, and HubSpot integrate 

automated responses with knowledge base systems, and this enables customers to re-

solve issues independently without waiting for human assistance. While automated re-

sponse systems improve efficiency, they can sometimes lack the personalized touch 

(Fokus Answering Service, 2024). 

 

Voice recognition AI is increasingly used tool in industries such as banking and healthcare. 

Voice recognition tools, such as Google Cloud Speech-to-Text and IBM Watson Speech-

to-Text, allow customers to interact with service platforms via voice command. This 

makes service more accessible, particularly for customers with disabilities or those seek-

ing a more hands-free experience.  While voice recognition AI can make services more 

accessible and faster, there are still some challenges with it, such as understanding dif-

ferent accents and dealing with background noise (Saka et. Al, 2023, p. 13).  

 



31 

Final tool in Table 1. Is predictive analytics. It is a powerful AI tool which is used to fore-

cast future customer behaviours and trends. By analyzing customer data, businesses can 

anticipate customer needs and personalize their service offerings. Tools like Salesforce 

Einstein and IBM Watson Analytics use machine learning models to predict outcomes, 

such as when a customer is likely needing support or what type of product they may 

purchase next.  However, there are also things to consider when using predictive analyt-

ics. Predictive analytics relies on customer data, so businesses must ensure they comply 

with regulations like GDPR and CCPA. Also, the accuracy of predictive analytics depends 

on the quality of the data used and incomplete or incorrect data can lead to inaccurate 

predictions and irrelevant support efforts (Nice, 2025). 

 

Table 1. Comparison of AI tools for customer service 

AI Tool Examples Benefits Drawbacks Best Use 

Case 

Customer 

Service 

Use 

Technical 

Use 

Chat-

bot 

Zendesk 

Chat, Drift, 

Intercom 

24/7 

availabil-

ity, han-

dles high 

volume 

of simple 

inquiries, 

cost-ef-

fective, 

reduces 

human 

error 

Limited by 

pre-pro-

grammed 

responses, 

struggles 

with com-

plex que-

ries, lack of 

human 

touch. 

Han-

dling 

FAQs, 

basic 

trouble-

shoot-

ing, ini-

tial cus-

tomer 

contact. 

Provides 

instant 

support, 

schedules 

meetings, 

routes 

complex 

issues to 

human 

agents. 

Requires in-

teraction 

with mes-

saging plat-

forms, reg-

ular up-

dates to im-

prove con-

versational 

flow. 

Senti-

ment 

Mon-

keyLearn, 

Improves 

customer 

Accuracy 

can be af-

fected by 

Analyz-

ing cus-

tomer 

Identifies 

customer 

mood and 

Analyzes 

large vol-

ymes of text 



32 

AI Tool Examples Benefits Drawbacks Best Use 

Case 

Customer 

Service 

Use 

Technical 

Use 

Analy-

sis 

Lexalytics, 

Aylien 

experi-

ence by 

under-

standing 

emo-

tional 

tone, 

helps 

proac-

tively ad-

dress is-

sues. 

poor data 

or algo-

rithm mis-

interpreta-

tion, strug-

gles with 

sarcasm. 

senti-

ment 

from so-

cial me-

dia, 

email, or 

chat in-

terac-

tions to 

bring 

satisfac-

tion. 

urgency, 

guides 

the next 

steps in 

service 

delivery. 

data using 

machine 

learning al-

gorithms to 

detect sen-

timent. 

auto-

mated    

re-

sponse 

system 

Freshdesk, 

Zoho Desk, 

HubSpot 

Auto-

mates re-

sponses 

to repeti-

tive que-

ries, re-

duces re-

sponse 

time, im-

proves ef-

ficiency. 

May lack 

personali-

zation, re-

quires on-

going up-

dates to re-

main effec-

tive. 

Manag-

ing com-

mon 

queries 

in email 

or chat, 

freeing 

up 

agents 

for com-

plex is-

sues. 

Responds 

with pre-

set solu-

tions, 

sends au-

tore-

sponses, 

escalates 

if needed. 

Incorpo-

rates 

predifined 

templates 

for com-

mon cus-

tomer que-

ries. 



33 

AI Tool Examples Benefits Drawbacks Best Use 

Case 

Customer 

Service 

Use 

Technical 

Use 

voice 

recog-

nition 

ai 

Google 

Cloud 

Speech-To-

Text, IBM 

Watson 

Speech-To-

Text 

Enables 

voice-

based 

customer 

service, 

increases 

accessi-

bility, 

speeds 

up ser-

vice. 

Requires 

accurate 

speech 

recogni-

tion, might 

struggle 

with ac-

cents or 

noisy envi-

ronments. 

Voice-

acti-

vated 

cus-

tomer 

service, 

accessi-

bility for 

im-

paired 

custom-

ers, call 

center 

automa-

tion. 

Converts 

customer 

speech 

into text, 

provides 

real-time 

responses 

via voice 

assis-

tants. 

Integrates 

with voice 

recognition 

software to 

transcribe 

and inter-

pret cus-

tomer con-

versations. 

Predic-

tive an-

alytics 

Salesforce 

Einstein, 

HubSpot 

AI, IBM Eat-

son Analyt-

ics 

Antici-

pates 

customer 

needs, 

personal-

izes ser-

vice, pre-

dicts out-

come, 

improves 

decision-

making. 

Requires 

large da-

tasets to 

function ef-

fectively, 

can be ex-

pensive to 

implement. 

Data usage 

and requla-

tions. 

Fore-

casting 

cus-

tomer 

needs, 

custom-

izing 

market-

ing strat-

egies, 

improv-

ing 

Analyzes 

data pat-

terns, 

forecasts 

service 

demand, 

adjusts 

customer 

service 

strategy. 

Uses ma-

chine learn-

ing models 

to analyze 

historical 

customer 

data and 

predict fu-

ture. 



34 

AI Tool Examples Benefits Drawbacks Best Use 

Case 

Customer 

Service 

Use 

Technical 

Use 

deci-

sion-

making 

in cus-

tomer 

service. 

 

2.4  Key factors in successful AI implementation 

Artificial Intelligence has become an important tool across various sectors. It is changing 

processes in industries like healthcare, finance, marketing, and customer service. How-

ever, its adoption is not always so easy. Understanding the factors that affect AI ac-

ceptance and obstacles and strategies to overcome them, is essential. AI acceptance de-

pends on both technical and human factors. One important factor is how useful people 

think AI systems are. Users are more likely to use AI if they believe it makes their work 

faster and more accurate compared to traditional methods. Also, it is important that 

these tools are easy to use. If AI systems are complicated or require a lot of training, 

users may be discouraged from using them. This is why it is important to design AI tools 

that are easy to use (Rane et al. 2024, p. 51). 

 

Trust in AI is another crucial factor. Users need confidence in the reliability, transparency, 

and fairness of AI systems. Trust is built when decision-making processes are clear and 

explainable, which is important to avoid worries about bias and fairness. Explainable AI 

(XAI) is one way to make AI more transparent, showing users how it makes decisions, 

which helps build trust (Rane et al. 2024, p. 51). 

 

Worries about job loss are also a big challenge, especially in industries where AI can re-

place regular tasks. To solve this, companies should focus on training workers for new 



35 

roles that AI can’t easily take over. There are also ethical and legal issues, like data privacy 

and who is responsible for AI decisions, that make it harder to adopt AI. The absence of 

clear rules makes these problems worse, so it’s important to have guidelines to ensure 

AI is used ethically. In addition to this, technical challenges, like making AI work with 

existing systems, can be difficult. The computing power and data needed to train AI mod-

els can be expensive, so investing in the right infrastructure is necessary (Rane et al. 2024, 

p. 51). 

 

To tackle these challenges, several strategies can be used. One effective method is to 

involve users in the design and development of AI systems. This helps make sure that AI 

tools meet their needs and expectations. This also builds trust by addressing concerns 

about usability. Education and awareness programs are also important to help users un-

derstand both the benefits and limitations of AI, which can reduce misunderstandings 

and fears. Clear ethical rules and guidelines are needed to create a trustworthy AI envi-

ronment, which is focusing on transparency, fairness and responsibility. Ongoing moni-

toring and checks of AI systems are also important to spot and fix problems, which helps 

build trust and encourages acceptance of AI technology (Rane et al. 2024, p. 51). 

 

2.4.1 Technical factors 

The technical requirements for using AI tools in customer service are usually the first 

thing companies have to worry about. Integrating AI into existing systems is a challenge 

for most businesses. Usually when business is looking for proper implementation of AI 

into its existing system, implementation will require help of AI solution providers. AI tools 

need a strong IT system that can handle complex algorithms and large amounts of data. 

This might create a storage issue for business. Data driven automation in business oper-

ations may also cause issues related to data security. Replacing outdated infrastructure 

with traditional legacy systems is still a major challenge for most organizations. Most 

Artificial Intelligence based solutions have a high level of computational speed. AI-based 

systems will be able to achieve more speed if business has a substantial infrastructure 

and high-end processors. The function and performance of business intelligence 



36 

operations heavily rely on AI algorithms. Enterprises planning to implement AI should 

have a clear idea of how AI-based solutions or technologies work and will be able to 

transform their outcomes (10xds, 2025). 

 

It is also important to have strong methods to find and fix bias in AI systems. AI can 

accidentally pick up biases from the data it’s trained on, which can lead to unfair results. 

To avoid this, AI systems should be checked regularly for bias. Using a variety of data 

when training the AI is also important to make sure it treats everyone fairly. In addition, 

AI should be built to spot bias and fix it right away, making sure that its decisions are fair 

for all. These steps are necessary to make sure AI is used in a responsible way (Inavolu, 

2024, p. 20). 

 

Choosing the right AI tool for a business is an important technical factor in ensuring suc-

cessful implementation. Because there are many options available, selecting the best 

tools can be challenging. The first step is to know the specific needs of the business. The 

budget and the technical skills of the team must also be taken into account. User-friendly 

tools may be necessary for teams with less technical expertise and more complex. Tools 

may be required for teams with advanced knowledge (T_HQ technology and business, 

2024). 

 

AI tools can be grouped into different categories based on their function. Automation 

tools help with repetitive tasks, data analysis tools assist in making data-driven decisions, 

and customer service tools, such as chatbots, enhance customer interactions. There are 

also tools for content creation, which help marketing teams with generating content, 

graphics, or videos (T_HQ technology and business, 2024). 

 

After narrowing down the options, the features of the tools should be evaluated. It is 

important to ensure that the tools provide the necessary functionalities, such as integra-

tion with existing systems or usability. It is also important to consider the support offered 

by the tool providers, like helpful guides and responsive customer service. Before 



37 

choosing a tool, it’s important to test it first. Many providers offer free trials, so the tool 

can be tested in the business setting to see if it meets expectations. Since AI tools often 

deal with sensitive data, it’s important to choose ones that follow data protection rules 

and have strong security features, like encryption and access controls (T_HQ technology 

and business, 2024). 

 

2.4.2 Organizational factors 

Artificial Intelligence is changing how businesses operate. Companies are investing heav-

ily in AI, but implementing AI without aligning it with business objectives often leads to 

wasted resources and limited impact. To get better result from AI, businesses need to 

connect their AI plans with their main goals. Business objectives are clear goals that a 

company works towards. They help guide decisions, shape plans, and decide how re-

sources are used. Common objectives include increasing sales, keeping customers, low-

ering costs, and improving products. For AI projects to be successful, they need to sup-

port these goals (RTS labs, 2024). The most common business objectives are presented 

in Table 2. 

 

Table 2. Types of business objectives 

Business objectives Explanations 

Strategic goals Focus on long-term aims like growing in 

new markets or building the brand (RTS 

labs, 2024). 

Operational goals Aim to make processes better, like im-

proving supply chain or product quality 

(RTS labs, 2024). 

Financial goals Focus on clear numbers, like increasing 

revenue or cutting costs (RTS labs, 2024). 



38 

Customer goals Aim to improve customer satisfaction, 

keep loyal customers, or attract new ones 

(RTS labs, 2024). 

 

These objectives are important to AI implementation because AI is a strong tool, but its 

value depends on how it is used. If a company uses AI without clear goals, the results 

may not be useful. For example, a business might us AI chatbots to improve customer 

service, but without clear goals, like speeding up response times or improving customer 

service ratings, the project may not meet expectations. Aligning AI with business goals 

requires thoughtful planning and collaboration. Table 3. presents the most important 

steps in implementing AI tools (RTS labs, 2024). 

 

Table 3. Steps in implementing AI tools 

Steps in implementing AI tools Explanations 

Identify business needs Answering the questions: Which pro-

cesses are inefficient or time-consuming? 

What areas could benefit from better in-

sights or predictions? Where are re-

sources being underutilized? (RTS labs, 

2024). 

Define clear AI objectives Clear objectives help guide AI projects 

and ensure they align with business goals. 

These objectives should explain what the 

AI project aims to achieve and how suc-

cess will be measured (RTS labs, 2024). 

Collaborate across departments AI projects usually need help from differ-

ent teams like operations, IT, marketing, 

and finance. Working together ensures 

that AI solutions are useful and solve real 

problems (RTS labs, 2024). 



39 

Steps in implementing AI tools Explanations 

Monitor and adjust AI systems constantly evolve and require 

continuous evaluation. Monitoring per-

formance metrics helps keep them 

aligned with business goals. If objectives 

or conditions change, the AI model or its 

settings should be adjusted (RTS labs, 

2024). 

 

Aligning AI initiatives with business goals is a challenging task to organizations. Although 

the benefits of artificial intelligence are well known, achieving these benefits is often 

difficult because of different obstacles, such as technical, organizational, and cultural is-

sues. To make sure AI projects are successful and produce clear results, these challenges 

need to be tackled step by step (RTS labs, 2024). 

 

One of the main challenges in aligning AI with business goals is not having clear objec-

tives. Companies often use AI because it seems like innovative tool or because compet-

itors are using it, rather than having a clear plan. Without clear goals, AI projects can 

become unfocused, and they may not deliver real value. For example, a retail company 

might invest in AI to improve speed, personalization, or service quality. This lack of focus 

can lead to wasted efforts and resources. To solve this problem, it’s important to clearly 

define what the AI project should achieve and make sure it fits with the overall business 

goals. Involving people from different departments can help set clear and useful objec-

tives. Using methods like OKRs (Objectives and Key Results) or SMART goals can also 

make sure the goals are clear and measurable (RTS labs, 2024). 

 

Data quality and access are also big challenges. AI systems depend on data, and if the 

data is poorly managed, it can make the system less effective. Common problems include 

data being in inconsistent formats, incomplete, or scattered across different depart-

ments. To fix these data problems, companies should focus on cleaning and organizing 



40 

their data. Centralized storage systems, such as data lakes or warehouses, can help col-

lect data from different departments in one place, making it simpler to manage and an-

alyze (RTS labs, 2024). 

 

Another big challenge is resistance to change. Employees might resist AI because they 

worry that it will take their jobs or disrupt their usual work. This can slow down the 

adoption of AI and make it less effective. To reduce this resistance, it’s important to in-

volve employees in the process from the start. Clear communication about why AI is 

being used and how it helps can ease their concerns. Offering training programs can also 

help employees learn to work with AI systems. It’s important to show how AI can support 

their work instead of replacing it (RTS labs, 2024). 

 

Resource limits make it harder to align AI projects with business goals. Many organiza-

tions, especially small and medium-sized businesses, have trouble getting the technical 

skills, equipment, and funding needed for AI projects. The costs of hiring experts, main-

taining AI models, and buying the right hardware and software can be too high. To solve 

these resource problems, businesses can look into ready-made AI solutions from cloud 

providers, which are usually cheaper and can grow with the company. Working with out-

side experts, vendors, or universities can also give businesses the skills and resources 

they need. Starting with smaller, high-impact projects can show the potential value of AI, 

making it easier to justify more investment. Additionally, government grants or incen-

tives may help cover some of the costs for adopting AI (RTS labs, 2024). 

 

Unrealistic expectations about what AI can do and how quickly it will work are also a 

common problem. AI is often seen as a quick solution to business problems, but in reality, 

AI models take time to develop, test, and improve. Results usually come gradually, not 

instantly, so expectations need to be adjusted. To fix this, companies should set realistic 

goals for AI projects and teach everyone involved that AI development takes time. Break-

ing projects into smaller steps and celebrating early successes can help keep things mov-

ing forward and manage expectations (RTS labs, 2024). 



41 

 

Successfully aligning AI with business goals needs careful planning, good leadership, and 

teamwork across departments. Leaders play an important role by making sure AI pro-

jects fit with the company’s bigger goals. Starting with small pilot projects helps test AI 

solutions and improve the approach before expanding. Focusing on clear ROI and train-

ing employees are also important for success. Working with the outside experts can help 

overcome resource challenges and speed up AI adoption. Regularly monitoring and im-

proving AI systems is important to make sure they keep adding value. Companies need 

to check how their AI models are performing, review them, and use feedback from em-

ployees and customers to make improvements. By doing this, companies can make sure 

their AI projects support business goals and deliver real results (RTS labs, 2024). 

 

2.4.3 Human factors and ethical considerations 

Human factors are often the most challenging element of AI implementation. The key 

human factor in AI adoption is employee mindset and acceptance. Even though genera-

tive AI tools like ChatGPT have developed quickly, many employees are still unsure and 

resistant to using AI. They worry about job security, the complexity of the technology, 

and the ethical issues around AI. This resistance is seen more in older workers, while 

younger workers are more familiar with AI tools. AI’s ability to learn on its own and make 

decisions without being programmed raises concerns about how transparent and ac-

countable it is (Golgeci et al. 2024, p. 1–3). 

 

When employees resist change, they experience several reactions, such as mistrust, ex-

istential questioning, and technological reflection. Mistrust comes from worries about 

AI’s reliability and possible dangers. Existential questioning is about doubts regarding the 

importance of human roles in a working place with AI. Technological reflection involves 

thinking about both the good and bad effects of AI. These reactions are connected, and 

organizations need to understand how to manage them (Golgeci et al. 2024, p. 5–10). 

 



42 

At the organizational level, things like making AI easier to use, combining AI with human 

work, and making AI seem trustworthy can help reduce resistance. Making AI accessible 

means making it simple and user-friendly, which lowers mistrust. Human-AI augmenta-

tion focuses on having AI work together with people to improve teamwork instead of 

replacing jobs, which helps with concerns about job roles. Lastly, AI legitimation makes 

sure employees see AI as useful and valuable by clearly explaining its purpose and ben-

efits. These actions help reduce resistance and make it easier for AI to fit into the work-

place (Golgeci et al. 2024, p. 5–10). 

 

Another human factor to consider is customer acceptance of AI-driven service. While 

many customers appreciate the convenience and speed of AI-based services, others may 

feel that AI lacks empathy and personal touch of human interactions. Companies must 

have a balance between AI and human service, so that customers receive personalized 

assistance when needed. Customers’ understanding of their own role and AI’s role, their 

motivation to use AI, and their ability to use AI devices make them more willing to adopt 

AI. However, privacy concerns make usually less likely to accept AI, especially when it 

comes to understanding their role. Trust in AI technology, on the other hand, makes cus-

tomers more willing to use AI when they feel capable of using it (Choi, 2023, p. 8–11). 

 

Customer acceptance of AI services depends on several factors, including the customer’s 

characteristics and the technology itself. Key factors for acceptance are role clarity, mo-

tivation, and ability. Trust in the technology, privacy concerns, and how “creepy” the AI 

feels also matter. AI adoption is also influenced by how well the technology fits with 

customers’ values and the risks they see in using it. It is important for customers to un-

derstand the role of AI in service interactions. Clear communication about what cus-

tomer and AI are each responsible for helps build trust and transparency. This clarity is 

necessary to avoid confusion and improve satisfaction, especially in important areas like 

self-driving cars or healthcare (Ostrom et al., 2019, p. 84–97). 

 



43 

Individual differences, like experience with technology, age, and culture, affect how peo-

ple accept AI. Younger people and those from regions like Asia are more willing to use 

AI. How well customers can use AI devices also affects their acceptance, as AI can either 

help or limit their abilities based on how it is designed. Although AI has many benefits, 

concerns about privacy, control, and social effects are still important. Privacy issues, es-

pecially with personalized services, may stop people from using AI, even though these 

services can be very useful. Also, AI replacing human interaction could make people feel 

isolated or harm social skills, especially in areas like eldercare or childhood development 

(Ostrom et al., 2019, p. 84–97). 

 

Positive effects of using AI include more personalized services, better abilities, time sav-

ings, and improved well-being. AI can create custom experiences, like personalized rec-

ommendations or learning that adapts to needs, which can make customers happier. AI 

also makes things more convenient, helping people save time and use services more 

easily. However, there are also negative effects, like service failures, loss of control, and 

bias in AI decisions. AI might make bad recommendations or be biased, which can harm 

consumers. Also, AI in service interactions can make people feel like they are losing pri-

vacy or control over their personal data. AI has many benefits in services, but customers 

will accept it if there is trust, clear communication, and privacy concerns are handled 

well. Companies need to manage these factors to make sure customers have good expe-

riences and avoid problems (Ostrom et al., 2019, p. 84–97). 

 

Ethics and governance have a critical role in the successful implementation of AI systems. 

To build trust with customers and meet legal requirements, organizations must focus on 

ethics in their AI strategy. Creating strong AI governance frameworks is essential. These 

frameworks give clear rules for using AI responsibly, making sure AI systems match the 

organization’s values and follow the rules. A good governance framework should cover 

key areas like accountability, transparency, and data security to help organizations use AI 

properly. Transparency is the most important ethical guideline. It focuses on how AI is 

used, being open about AI’s actions, and explaining AI’s decisions. Privacy guidelines 



44 

focus on protecting personal data, making sure organizations follow GDPR guidelines and 

anonymize data. Security guidelines are also important, and many organizations com-

bine privacy and security to keep AI safe and reduce risks (Balasubramaniam et al., 2023, 

p. 6–8). 

 

Fairness is important for both professional services and B2C organizations, aiming to re-

move bias and discrimination in AI systems. Accountability guidelines make humans re-

sponsible for overseeing AI decisions. These guidelines also include monitoring and cer-

tification. Reliability guidelines vary between organizations. Professional services focus 

on safety and the public sector emphasizes using reliable data. Ethical guidelines also 

highlight the importance of building trust. Transparency helps with privacy, security and 

fairness issues. Explainability makes AI’s decisions easier to understand for users and de-

velopers (Balasubramaniam et al., 2023, p. 6–8). 

 

The most important requirements related to transparency and explainability are under-

standability, traceability, and auditability. Understandability means clearly explaining 

how AI is used. Traceability focuses on tracking AI decisions and the data behind them. 

Auditability means that AI systems should be able to be checked for accountability. For 

these reasons transparency and explainability are important for building trust, fairness, 

and improving AI system quality (Balasubramaniam et al., 2023, p. 6–8). 

 

According to UNESCO (2024), there are ten approaches to AI ethics. 

1. Proportionality and Do No Harm. 

AI systems should only be used as much as needed to achieve a valid goal. Risk 

assessments should be done to avoid any harm that could come from their use. 

2. Safety and security. 

Unwanted safety risks as well as security risks should be avoided and addressed 

by AI actors. 

3. Right to Privacy and Data Protection. 



45 

Privacy should be protected throughout the entire AI process. Strong data pro-

tection measures should also be set up. 

4. Multi-stakeholder and Adaptive Governance & Collaboration. 

International law & national sovereignty must be respected when using data. 

Also, involving different groups of people is important for inclusive AI governance. 

5. Responsibility and Accountability. 

AI systems should be auditable and traceable. There should be checks, impact 

assessments, audits, and careful reviews to prevent conflicts with human rights 

and harm to the environment. 

6. Transparency and Explainability. 

The ethical use of AI systems depends on their transparency and explainability (T 

& E). The level of T & E should fit the situation, as it may conflict with other prin-

ciples like privacy, safety, and security. 

7. Human Oversight and Determination. 

Member States should ensure that AI systems do not displace ultimate human 

responsibility and accountability. 

8. Sustainability. 

AI technologies should be evaluated based on their impact on “sustainability”, 

which includes goals like those in the UN’s Sustainable Development Goals. 

9. Awareness & literacy. 

The public’s understanding of AI and data should be improved through open ed-

ucation, civic involvement, digital skills, AI ethics training, and media literacy. 

10. Fairness and Non-Discrimination. 

AI developers should support social justice, fairness, and equality, making sure AI 

benefits are available to everyone (UNESCO, 2024). 

 

2.4.4 Leadership in AI implementation projects 

Managing AI implementation projects requires strong leadership to guide the process 

from conception to deployment. Leaders must be capable of coordinating efforts across 

different departments, ensuring that technical, organizational, and human factors are 



46 

aligned. They must also be proactive in addressing challenges and adjusting strategies as 

needed. Global spending on AI reached $118 billion in 2022, but many companies report 

little success, with only a few AI projects being fully implemented. Even big tech compa-

nies like IBM and Amazon have trouble expanding AI. Despite the many benefits of AI, 

such as quicker communication, better predictions, and automation, many companies 

face big challenges when trying to use it (Ångström, 2023, p. 5–6). 

 

Successful AI implementation needs continuous learning and adjustment. Pilots and ex-

periments are important for starting AI projects and improving understanding within the 

organization. However, as AI is used in more complex situations, initial successes can 

become harder to maintain. As companies grow in their use of AI, they face challenges 

with data management, advanced algorithms, and a broader range of partners. To han-

dle these challenges, companies often choose simpler, more staple tools rather than new, 

complex solutions (Ångström, 2023, p. 13–19). 

 

A key part of adopting AI is having the right people. It’s not only about the technology, 

but also about building a team that can manage AI changes. To use AI successfully, com-

panies should invest in training and encourage teamwork to make sure AI adds value 

instead of becoming a problem. As companies become more AI-driven, they need to 

organize their data better, handle the increasing complexity of AI, and work with external 

partners. The challenge increases as AI affects customers and partners, requiring strong 

relationship management skills and attention to regulations (Ångström, 2023, p. 13–19). 

 

To get a clear picture of AI project management, it is a good idea to look at the operations 

of a real company, for example company called Consult. Consult is a North American AI 

consultancy, and it helps organizations leverage AI to solve complex business problems. 

Consult was founded in 2017, and it has made significant progress in managing AI pro-

jects, mainly in logistics and supply chain industries. The company offers advisory ser-

vices and custom software, such as machine learning models and optimization algo-

rithms to its clients. However, despite its success, Consult faces challenges in managing 



47 

AI projects, especially when balancing traditional project management, agile methods, 

and AI workflows (Vial et al., 2022, p. 671–675). 

 

Consult manages AI projects using three main strategies: traditional project manage-

ment, agile methods, and AI-specific workflows. The company organizes projects into 

five phases with regular checkpoints to review progress and feasibility. Agile practices 

help teams work in short cycles and involve customers early. However, the complexity of 

AI, especially with data science and machine learning, often causes uncertainties that 

make it harder to manage project goals and deadlines (Vial et al., 2022, p. 672–678). 

 

Conflicts happen between the different methods: traditional management, agile, and AI 

workflow. For example, traditional project management expects uncertainty to decrease 

over time, but AI projects often face more uncertainty as data issues and model issues 

come up. Agile practises focus on fast work cycles and delivering products, while the AI 

workflow uses small experiments that may not give quick results. These different ap-

proaches can clash, making it hard to manage customer expectations and align on pro-

ject goals (Vial et al., 2022, p. 677–682). 

 

To solve these conflicts, Consult uses several strategies. One is to “fire” customers if they 

are not ready for an AI project, based on their AI maturity and data quality. Another is to 

rethink the definition of “done” in AI projects, focusing on testing and experiments in-

stead of fixed deliverables. Consult also checks the value of extra data science work to 

make sure technical progress matches business goals. Lastly, the company pairs business 

consultants with data scientists to link technical skills with business knowledge (Vial et 

al., 2022, p. 683–686). 

 



48 

3 Research Methodology 

This study uses a qualitative research method to understand how artificial intelligence 

tools can improve customer service processes in companies. A qualitative method was 

chosen because it helps explore the experiences, views, and strategies that businesses 

use when they are adopting AI tools since it focuses on understanding how individuals 

or groups perceive and interpret social or human-related issues. Instead of following a 

strict framework, this approach allows for flexible research questions and procedures 

that evolve throughout the study. Data is usually collected in natural setting, such as 

interviews or observations, where participants share their experiences. Unlike 

quantative research, which focuses on numbers and statistics, qualitative research looks 

at human experiences and how organizations operate in a more detailed way. This ap-

proach is especially useful for studying complex and constantly changing topics where 

technical, organizational, and human factors all play a role (Creswell, 2023, p. 4).  

 

 

3.1  Research design 

Research designs are usually divided into three types: quantative, qualitative, and mixed 

methods. These types show whether the data collected are numbers (quantative), non-

numbers like words or images (qualitative), or mix of both. Quantative methods use 

numbers, which are often collected through tools like questionnaires and analyzed with 

statistics. Qualitative methods focus on non-numeric data, such as words or pictures, 

which are often collected through interviews or observations. Mixed methods combine 

both quantative and qualitative approaches (Saunders & Thornhill, 2023, p. 181–186). 

 

Quantative research is often linked to positivism and a deductive approach, where data 

is collected to test existing theories. It can also use an inductive approach to create new 

theories. Quantative designs usually involve fixed methods for collecting data and ana-

lyzing it with statistics to find results that can apply to larger groups. (Saunders & Thorn-

hill, 2023, p. 181–186). 



49 

The part of this study focused on examining statistics about what people thought about 

various AI tools. Two statistics were collected from the Finnish Data Archive Aila and 

analyzed using SPSS software. The statistics included multiple-choice questions, which 

respondents were asked to answer by selecting the most appropriate option. The answer 

options were assigned numerical values, and in this research these numerical tables 

were analyzed with SPSS and graphs were drawn to illustrate the answers and their dif-

ferences. These studies provided broader information from a larger group of people 

about how people view AI tools. These studies addressed questions related to people’s 

fear of losing their jobs when AI tools enter the company (EVAn Arvo- ja asennetutkimus 

kevät 2022) and people’s thoughts about the security of these tools (Turval-

lisuuspolitiikka- ja maanpuolustustutkimus 2021). 

 

This research is based on an approach that focuses on understanding social realities 

through the experiences of individuals and organizations. This is especially important 

when studying AI in customer service because the success of AI tools depends not only 

on their technical abilities but also on how businesses perceive, use, and integrate them 

into their daily operations. To get a good view of how AI is adopted, its benefits, and the 

challenges it brings, this thesis includes semi-structured interviews with key people in-

volved in the process (Creswell, 2023, p. 17). 

 

This thesis also includes elements of a case study approach. It looks at how different 

companies use AI in customer service. Case studies are useful because they provide real-

world insights by gathering detailed information. This method helps achieve the goal of 

understanding not just how AI tools work, but also how companies plan and manage 

their use to ensure successful results (Woodside, 2016). Some of the research findings 

were obtained from literature.  

 

3.2 Data collection 

In this thesis, data was collected from both primary and secondary sources to get a clear 

picture of how AI is used in customer service. The primary data came from semi-



50 

structured interviews, while the secondary data included a literature review and statis-

tical information from the Tietoarkisto Aila. Semi-structured interviews were used as the 

main way to collect data because they provide flexibility. This method is common in qual-

itative research since it helps researchers understand people’s experiences, opinions, 

and attitudes in depth (Ruslin et al., 2025, p. 10–11). A total of three participants were 

found for the interview. 

 

Participants for the interview were selected based on their knowledge and experience 

related to the research topic (Patton, 2014, p. 437). The interview questions focused on 

how AI is used in customer service, the challenges of implementing it, and the best ways 

to make it work successfully. The interviews were held online using Microsoft Teams. 

This made it easier for participants to join from any location. Each interview lasted about 

40 to 60 minutes, and they were recorded with the participants’ permission to ensure 

accurate interpretations. The transcripts were then studied using thematic analysis, 

which is a common method for finding key patterns and themes in the data. Thematic 

analysis is a way to find, examine, and describe patterns in qualitative data. It helps or-

ganize and explain the data in detail and can also be used to understand different parts 

of a research topic (Braun & Clarke, 2008, p. 79–81).  

 

Three interviews were conducted for this research. The names of the interviewees who 

are participating to this research have been coded as: Participant 1, Participant 2, and 

Participant 3. 

 

Participant 1 is an entrepreneur of a Finnish social media company that makes content 

for various social media channels, for example, TikTok, Instagram and YouTube, and gen-

erates revenue through commercial partnerships and the social media channels’ own 

earning opportunities, for example, YouTube’s partner program. The company employs 

one person, the founder. The company uses artificial intelligence to create and edit con-

tent and the company has also automated customer service by automating responses to 



51 

private messages on social media channels. To automate responses, the entrepreneur 

uses Instagram’s own response automation tool.  

 

Participant 2 is a nurse of a home care facility belonging to the Southern Finland Hospital 

District. Her workplace uses an automated medication dispenser in clients’ homes. The 

automated medicine dispenser is part of a larger shift toward digitalizing customer ser-

vice in healthcare, where automation helps support nursing staff and enhance patient-

centered care. The device alerts users when it’s time to take their medication and dis-

penses the correct pre-dosed amount medicine. Its primary goal is to reduce the need 

for frequent nurse visits while promoting independent living for customers in their own 

home. Unlike chatbots or virtual assistants, a medicine dispenser is a tangible physical 

automated device, which provides an interesting point of comparison for other entre-

preneur interviews. 

 

Participant 3 is a project manager at a medium-sized Finnish retail company. The com-

pany has implemented a chatbot to improve customer service. From time to time, espe-

cially during the high season, customer service receives a lot of inquiries about the same 

things, and the chatbot was implemented in the compa