The American Journal of Engineering and Technology
118
https://www.theamericanjournals.com/index.php/tajet
TYPE
Original Research
PAGE NO.
118-126
10.37547/tajet/Volume07Issue03-11
OPEN ACCESS
SUBMITED
23 January 2025
ACCEPTED
13 February 2024
PUBLISHED
12 March 2025
VOLUME
Vol.07 Issue03 2025
CITATION
Jivani Zubin. (2025). methods for integrating chatbots into customer
experience management systems. The American Journal of Engineering
and Technology, 7(03), 118
–
126.
https://doi.org/10.37547/tajet/Volume07Issue03-11
COPYRIGHT
© 2025 Original content from this work may be used under the terms
of the creative commons attributes 4.0 License.
Methods for Integrating
Chatbots Into Customer
Experience Management
Systems
Jivani Zubin
Engineering Manager, Meta New York, US
Abstract:
The article explores approaches to integrating
chatbots into customer interaction management
systems, as well as other digital platforms such as
educational, financial, and marketing environments.
The aim is to study architectural solutions and
algorithms that ensure the productive operation of
chatbots in terms of performance, scalability, and
flexibility in various user interaction scenarios.
The methodology is based on comparing centralized and
decentralized integration models. It examines data
transfer protocols such as REST API, GraphQL, and
WebSocket. Special attention is paid to natural language
processing algorithms, including transformers like BERT
and GPT, which can interpret queries, maintain context,
and quickly adapt to changes in communication
scenarios.
The article also discusses hybrid models combining
automation with human operators for handling non-
standard situations. Approaches focused on active
learning are examined, which improve chatbot
performance in real-time.
The results demonstrate that the use of chatbots in
customer interaction management systems and e-
commerce improves query processing, speeds up
responses, and enhances personalization. The
application of data analytics opens opportunities for
predicting customer behavior and generating proposals
tailored to user needs. Issues of data security,
encryption, authentication, and access control, which
are critical for regulatory compliance, are also
considered.
The conclusions highlight the necessity of a
comprehensive approach to chatbot design, selecting
flexible architectural solutions, adapting to business
processes, and implementing machine learning
The American Journal of Engineering and Technology
119
https://www.theamericanjournals.com/index.php/tajet
The American Journal of Engineering and Technology
algorithms. The proposed methods are expected to
benefit software developers, analysts, marketers, and
managers engaged in digital transformation.
Keywords:
chatbots, CRM systems, integration,
natural language processing, REST API, machine
learning, scalability, digital transformation, data
security, personalization.
Introduction:
Modern business digital transformation
processes demand improved customer service quality,
faster query processing, and enhanced personalization
of interactions. Amidst increasing globalization and
intensifying competition, companies face the need to
optimize these aspects. Chatbots serve as an effective
solution for automating customer communication.
These systems utilize algorithms that minimize time
spent on task resolution and analyze user behavior to
generate personalized recommendations.
The integration of chatbots is becoming increasingly
relevant due to the growing volume of data requiring
rapid processing across multiple interaction channels.
Organizations operating in e-commerce, financial
technology, and education are actively adopting such
systems to provide support, handle queries, and offer
recommendations. Implementing these solutions into
existing customer service management structures and
digital platforms demands significant preparation and
resolution of technical challenges.
Advancements in natural language processing
technologies, including neural networks, open up new
opportunities for chatbots. These systems can
understand the context of user queries, generate
precise responses, maintain a coherent dialogue, and
handle multiple interactions sequentially. The process
of designing chatbots involves selecting appropriate
architectural solutions.
The integration of chatbots into small, medium-sized
businesses, and micro-enterprises has drawn the
attention of researchers. Selamat M. A. and Windasari
N. A. [1] emphasize the importance of personalized
recommendations, intuitive interfaces, and a human-
centered approach in communication for attracting
and retaining customers. Bednyak S. G. et al. [2, p. 33]
explore the integration of chatbots with various
communication channels, which accelerates response
times, enhances information accuracy, and improves
the customer experience.
The implementation of artificial intelligence (AI) and
machine learning technologies into CRM systems
enhances client data management and automates
customer service. Li R. C. and Tee M. L. [3, pp 1092-
1096] analyze the impact of AI algorithms on increasing
the efficiency of customer engagement. In e-commerce,
chatbots contribute to the creation of personalized
customer service. Thatavarthi N. [6, pp. 1-5] studies the
use of Angular and Node.js technologies to accelerate
order processing. According to Santosh K. et al. [9, pp.
1135-1140], the application of GPT-4 in e-commerce
chatbots increases the accuracy of query processing up
to 95%, improving user interaction.
Chatbots also play a significant role in marketing
strategies. Khoa B. T. [7, pp. 19-32] highlights their role
in optimizing communications that influence buyer
decision-making in online commerce. Mashkouk R. et al.
[12, pp. 1-5] examine the automation of processes and
the integration of chatbots into production chains,
which increases operational efficiency.
In education, chatbots are becoming essential tools for
supporting students by improving access to educational
materials and enhancing interaction with learning
platforms. Snekha S. and Ayyanathan N. [5, pp. 58-62]
note that chatbots help students navigate academic
processes, access information about schedules, exams,
and consultations. Mohamed I. S., Abdelsalam M., and
Moawad I. F. [11, pp. 351-355] demonstrate how
ChatGPT enables natural communication and provides
recommendations to enhance the educational process.
In the financial sector, chatbots automate loan
application processing, reducing costs associated with
data analysis and improving the accuracy of
recommendations. Prathipa S. et al. [8, pp. 1-6] confirm
that using chatbots in microcredit services enhances
customer service and improves decision-making
accuracy.
Giri U. et al. [10, pp. 118-123] investigate the impact of
chatbot interface design on customer perception.
Factors such as communication tone, response speed,
and personalization are crucial for increasing user
satisfaction. Kaushal V. and Yadav R. [4] underline the
importance of omnichannel support and the ability of
chatbots to adapt to various communication channels,
thereby enhancing customer interactions in the B2B
sector.
In turn, when presenting statistics, the source [13] was
used, the information from which is posted on the
official website of the Tidio company.
The purpose of this study is to examine architectural
solutions and algorithms that ensure the efficient
operation of chatbots in terms of performance,
scalability, and flexibility across diverse user interaction
scenarios.
The American Journal of Engineering and Technology
120
https://www.theamericanjournals.com/index.php/tajet
The American Journal of Engineering and Technology
The methodology of this study is based on comparing
centralized and decentralized integration models.
RESULTS
Modern business platforms demand speed and
accuracy in processing queries. The complexity of
communication flows and the growing volume of data
necessitate the adoption of technologies capable of
processing information in real-time and adapting to
changing interaction conditions. In this context,
chatbots become integral components of digital
solutions for managing customer interactions.
It is projected that by 2028, the global chatbot market
will reach $15.5 billion, compared to $4.7 billion in 2020
(Fig. 1). With an annual growth rate of approximately
23%, this surge reflects the increasing demand for
efficient and cost-effective artificial intelligence
solutions
Fig. 1. Global forecast of chatbot market development until 2028 [13].
Small companies are rapidly adopting chatbots because
third-party chatbot developers can easily integrate
their solutions. In contrast, larger companies typically
employ a different approach by developing proprietary
solutions,
which
significantly
lengthens
the
development process. Figure 2 below illustrates the
readiness of various companies to implement chatbot
technologies.
A key trend for 2024 and beyond is the rise of AI-driven
chatbots, led by models such as GPT-4 and other large
language models (LLMs). GPT-4, developed by OpenAI,
represents a significant advancement over GPT-3,
offering more accurate and human-like language
processing capabilities [13].
The American Journal of Engineering and Technology
121
https://www.theamericanjournals.com/index.php/tajet
The American Journal of Engineering and Technology
Fig. 2. Willingness to implement chatbot technologies by different companies [13].
The process of integrating chatbots is based on either
centralized or decentralized architecture. In a
centralized model, interaction management is
conducted through a single point
—
a central server that
exchanges data with external systems via APIs. This
model simplifies management but introduces a single
point of failure. Decentralized architecture, which
utilizes microservices, ensures the independence of
each component, enabling flexible system scaling and
effective adaptation to changes [1, 2, 6, 9].
Several protocols are used to connect chatbots with
external systems, which are visually represented in
Figure 3.
Fig.3. Protocols of interaction of chatbots with external systems [1, 2, 6, 9]
P
ro
to
co
ls
f
o
r
th
e
in
ter
ac
tio
n
o
f
ch
atb
o
ts
w
it
h
ex
ter
n
al
sy
ste
m
s
The REST API is a standard data transfer
method that supports caching and provides
ease of integration.
GraphQL is a protocol for working with
more complex data structures that allows the
client to accurately form queries.
WebSocket is a protocol for two—way
communication with minimal latency, used
in real time, for example, to monitor
processes.
The American Journal of Engineering and Technology
122
https://www.theamericanjournals.com/index.php/tajet
The American Journal of Engineering and Technology
Chatbots leverage natural language processing (NLP)
models to perform semantic analysis of user queries.
Transformers like BERT and GPT enable the recognition
of contextual dependencies and provide accurate
answers to ambiguously phrased questions.
Classification and clustering algorithms help predict
user actions and adapt communication scenarios
accordingly. Convolutional Neural Networks (CNN) and
Recurrent Neural Networks (RNN) process textual and
voice-based requests with minimal distortions.
Long-term memory mechanisms based on LSTM
architecture and attention mechanisms enable
maintaining extended dialogues while preserving
context across multiple sessions. This feature is critical
for B2B systems, where queries are often complex,
requiring repeated clarifications.
The integration of chatbots into customer interaction
management systems is a multi-level process involving
the development of architecture, data processing
algorithms, and security mechanisms. Technically, this
process is grounded in microservices architecture
principles, allowing the creation of scalable and fault-
tolerant solutions. Within this structure, the system is
divided into several functional layers, each with clearly
defined tasks.
●
Presentation Layer: Processes user requests,
providing interaction interfaces via web applications,
messengers, and voice assistants.
●
Logical Layer: Manages dialogues, analyzes
user intents, and processes information.
●
Integration Layer: Ensures connectivity with
external platforms such as CRM, ERP, databases, and
analytics systems through API protocols and
intermediary services.
Data transfer between system components is
implemented through REST API, GraphQL, and gRPC.
REST API serves as the standard for most CRM
platforms, simplifying integration. However, for
working with more complex data structures, GraphQL
is preferred, as it allows for querying only the required
fields. The gRPC protocol is used in high-load systems
where data transmission speed is critical. For real-time
interaction, WebSocket connections are utilized. These
channels are ideal for handling events requiring
immediate reactions, such as updates on application
statuses or changes in user profiles.
At the core of query processing are NLP algorithms that
ensure understanding of text and speech. Initially, data
undergoes tokenization, stemming, and lemmatization.
This is followed by syntactic and semantic analysis to
extract key entities and comprehend user intent.
Modern NLP models, such as BERT and GPT, employ
attention
mechanisms
and
contextual
vector
representations,
facilitating
precise
query
understanding. Machine learning algorithms like RNNs
and transformers analyze data sequences, maintaining
context over multiple sessions [3, 7, 12].
Various approaches are employed to manage dialogue
states. Simple scenarios are implemented through rule-
based systems, which establish fixed transitions
between stages of interaction. For dynamic systems,
such as technical support, adaptive models are used
that adjust based on context. For instance, a chatbot
can automatically switch from autonomous query
handling to redirecting queries to a live operator if the
issue requires more in-depth resolution. Systems based
on neural networks with long-term memory retain
information about previous interactions, which is
essential for extended processes like contract
negotiations or deal management.
Depending on the structure of information, relational
databases such as PostgreSQL and MySQL, or NoSQL
solutions like MongoDB and Cassandra, are utilized.
Relational databases are suited for transactional data
storage, while NoSQL databases are ideal for
unstructured data such as conversation logs and
metadata. Graph databases, such as Neo4j, allow
modeling complex relationships between entities, such
as client-agent interactions or event chains in business
processes. For faster data access, caching systems like
Redis and Memcached are frequently employed.
To ensure scalability and reliability, cloud platforms and
containerization technologies are applied. Tools like
Docker and Kubernetes facilitate the deployment of
microservices in containers, enabling flexible responses
to changing workloads. Cloud platforms such as AWS,
Google Cloud, and Microsoft Azure offer tools for traffic
balancing, database management, and performance
monitoring. Load balancers distribute requests across
chatbot instances, while monitoring systems track
metrics like response time and the number of
unprocessed queries [5, 11].
Data security is a critical consideration as chatbots
handle sensitive information. Encryption using
algorithms like AES-256 protects data during
transmission and storage. Anonymization and
tokenization minimize the risks of personal data leaks.
Multi-factor authentication and intrusion detection
systems safeguard against attacks by identifying
suspicious activities. Audit logs and data transaction
The American Journal of Engineering and Technology
123
https://www.theamericanjournals.com/index.php/tajet
The American Journal of Engineering and Technology
records help reconstruct incidents and enforce
compliance with security policies.
The integration of chatbots with multichannel
platforms broadens the scope of client interactions
across various channels: messengers (WhatsApp,
Telegram), social networks (Facebook, Instagram), and
voice assistants (Google Assistant, Amazon Alexa).
Unified interfaces are developed to manage data and
dialogues regardless of the request channel.
Interactions with contact centers are facilitated via SIP
protocols and CPaaS platforms, which unify text and
voice channels into a single query processing system.
Active learning has become a significant area in chatbot
development, enabling adaptation to new scenarios
based on real-world data. This approach is particularly
relevant for markets with evolving conditions. The
integration of chatbots with predictive analytics
systems unlocks the potential for anticipating client
needs based on behavioral data and offering
personalized solutions [2, 4, 7].
For requests requiring human intervention, escalation
mechanisms are employed, transferring the query to a
live operator. Hybrid interaction models strike a
balance between automation and human engagement,
reducing response time and enhancing service quality.
To improve chatbot performance, monitoring systems
are implemented to track key metrics. These metrics
are detailed in Figure 3.
Fig.4. Metrics for evaluating the effectiveness of chatbots [4, 8, 10]
Logging systems and event log analysis tools assist in
promptly identifying failures, thereby minimizing the
time needed for adjustments.
Active learning methods ensure the continuous
improvement of algorithms based on new data and
feedback. This is especially crucial for systems
operating under rapidly changing market conditions.
Next, Table 1 outlines the advantages and
disadvantages of using chatbots in customer
interaction management systems.
Metrics for evaluating the
effectiveness of chatbots
Average response time
This is a parameter that
measures how much time
passes on average between
the user's request and the
chatbot's response. The
importance of this indicator
lies in the fact that response
time directly affects the user
experience: the faster a
chatbot responds, the more
likely the user is to remain
satisfied and continue
interacting. The metric is
The degree of user
engagement
This parameter shows how
actively users interact with the
chatbot. Engagement can be
measured in various ways.
The percentage of erroneous
requests
This parameter measures the
percentage of user requests
that the chatbot cannot
process correctly or that it
responds to with an error.
The metric is calculated as
the ratio of the number of
erroneous requests to the total
number of requests
The American Journal of Engineering and Technology
124
https://www.theamericanjournals.com/index.php/tajet
The American Journal of Engineering and Technology
Table 1. Advantages and disadvantages of using chatbots in customer interaction management systems
(compiled by the author).
Advantages
Disadvantages
Chatbots can automatically handle simple
queries, such as order status, schedules, or
general information, reducing the workload
for human operators.
Chatbots face challenges when dealing with
complex or unconventional queries that require
creativity and flexibility, necessitating human
intervention.
Chatbots operate 24/7, providing clients with
constant availability and instant responses to
inquiries at any time.
Chatbots do not always correctly interpret user
context or intent, which can result in inaccurate
responses.
Automation through chatbots reduces the need
for staff to handle standard queries, lowering
customer service costs.
Chatbots cannot fully replace human interaction,
affecting service quality, especially when
emotional support or complex consultations are
needed.
Chatbots can process multiple requests
simultaneously, increasing service speed and
reducing wait times.
Chatbots cannot adapt to new or evolving
situations
without
additional
training
or
programming, limiting their flexibility.
Instant responses to standard questions and the
convenience of using chatbots enhance overall
customer satisfaction.
Chatbots require ongoing updates and training to
accurately understand queries and maintain
relevant
answers,
necessitating
resource
investment.
Modern chatbots can integrate with CRM
systems to provide personalized service,
including
product
recommendations
or
purchase history.
Systems can encounter failures or errors due to
technical issues, inaccurate data, or poor
integration with other platforms.
Chatbots can perform routine tasks, allowing
human operators to focus on more complex
issues.
Chatbots lack the ability to perceive emotions or
the nuances of human interaction, which can
degrade service quality in stressful or emotionally
charged situations.
The integration of chatbots into customer interaction
management systems encompasses a wide range of
technologies. The use of modern data processing
methods, cloud platforms, and neural networks makes
these systems efficient, flexible, and adaptable to
evolving market conditions.
Recommendations for implementing chatbots in
customer interaction management systems
Based on the conducted research, a set of practical
recommendations has been formulated to guide the
integration of chatbots into customer interaction
management systems. It is crucial to define the
objectives the system aims to achieve. One primary
goal is the automation of routine operations, reducing
the workload on staff. Chatbots handle standard user
requests such as checking order status, answering
frequently
asked
questions,
and
scheduling
appointments. This reduces wait times, eases the
workload of employees, and, thanks to 24/7
The American Journal of Engineering and Technology
125
https://www.theamericanjournals.com/index.php/tajet
The American Journal of Engineering and Technology
availability, accelerates the service process.
Implementation requires selecting the right platform
and technologies to ensure seamless integration with
the CRM system. The chatbot must have access to
customer data to provide accurate and personalized
responses. Utilizing information about past purchases,
preferences, and interaction history enhances service
quality by offering tailored recommendations. To
achieve this, the chatbot should be equipped with
natural language processing (NLP) algorithms to ensure
accuracy in understanding user queries.
Configuring the chatbot requires a thorough approach.
In the initial phase, it should be set up to handle
standard queries effectively in daily operations.
Training should rely on real data collected from
customers. However, the launch is only the first step.
The system must be regularly updated and tested to
adapt to business changes. Additionally, collecting user
feedback is essential for promptly addressing
shortcomings.
Equally important is the creation of a scalable system.
The chatbot should be able to adapt to business growth
by integrating new features and connecting additional
services. For companies operating in international
markets, it is critical that the bot supports multilingual
capabilities, ensuring high-quality interactions with
users from different countries.
Testing and monitoring the system's performance
should become integral to its operation. The launch of
a chatbot is not the end of the process but the
beginning of ongoing effectiveness analysis and
interaction scenario checks. Analytical tools should be
used to monitor interactions, identify issues, and
address them in a timely manner.
Adopting an omnichannel approach allows the
integration of the chatbot with various customer
interaction channels, such as email, social media, and
phone calls. This enables employees to access relevant
communication data, ensuring a personalized approach
regardless of the channel used.
In conclusion, the implementation process should
follow a phased approach. It is recommended to start
with a pilot project involving a limited number of users,
enabling the system to be tested and refined before full
deployment. This approach helps identify and resolve
issues early. However, it is essential to remember that
chatbot
implementation
requires
continuous
improvement and adaptation. The bot must evolve,
analyze new data, and enhance its accuracy and
efficiency over time.
CONCLUSION
The
technologies
discussed
automate
service
processes, personalize customer interactions, and
improve satisfaction levels. The use of artificial
intelligence and natural language processing methods,
such as BERT and GPT models, enables the creation of
systems that maintain conversational context, adapt to
diverse scenarios, and provide rapid responses.
The analysis of architectural solutions for chatbot
integration highlighted that the choice between
centralized and decentralized models depends on the
requirements for system scalability and stability.
Centralized architectures simplify data and interaction
logic management, while decentralized solutions offer
greater flexibility and resilience to failures. Protocols
like REST API, GraphQL, and WebSocket optimize data
exchange between chatbots and CRM systems,
reducing latency and enhancing performance.
REFERENCES
Selamat M. A., Vindasari N. A. Chatbot for small and
medium-sized businesses: integrating the views of
customers and business owners //Technology in
society.
–
2021.
–
Vol.
66.
DOI:
.10.1016/j.techsoc.2021.101685
Bednyak S. G. et al. Using Chatbots //XXIV All-Russian
Student Scientific and Practical Conference of
Nizhnevartovsk State University. - 2022.
–
p. 33.
Lee R. S., Ti M. L. Development of a platform for
implementing an automated customer support service
in customer relationship management systems for
collaboration //IEEE 2021 International Conference on
Industrial Design and Engineering Management (IEEM).
–
IEEE, 2021.
–
pp. 1092-1096.
Kaushal V., Yadav R. Studying the successful
implementation of chatbots in business from the point
of view of interaction with B2B clients // Parallelism and
computing: practice and experience.
–
2023.
–
Vol. 35.
–
No. 1. DOI:10.1002/cpe.7450
Sneha S., Ayyanatan N. Educational CRM chatbot for a
learning management system //The international
educational magazine Shanlax.
–
2023.
–
Vol. 11.
–
No.
4.
–
pp. 58-62.
Tatavarti, N. (2023). Integration of chatbots controlled
by artificial intelligence with fully functional e-
commerce platforms to improve the quality of
customer service // Journal of Artificial Intelligence and
Cloud Computing.
–
2023.
–
Volume 2(4).
–
pp. 1-5.
Hoa B. T. The influence of chatbots on the relationship
between integrated marketing communications and
The American Journal of Engineering and Technology
126
https://www.theamericanjournals.com/index.php/tajet
The American Journal of Engineering and Technology
online purchasing behavior in the frontier market //The
Messenger Magazine.
–
2021.
–
Vol. 13.
–
No. 1.
–
pp.
19-32.
Pratipa S. et al. A credit management system using a
chatbot //The 2024 International Conference on
Communications, Computing and the Internet of Things
(IC3IoT).
–
IEEE, 2024.
–
pp. 1-6.
Santosh K. et al. Using GPT-4 capabilities to develop
context-sensitive personalized chatbot interfaces in e-
commerce
customer
support
systems
//10th
International Conference on Communication and Signal
Processing (ICCSP), 2024. - IEEE, 2024.
–
pp. 1135-1140.
Giri U. et al. Understanding the User Experience of
Chatbots for Customer Service: An Experimental Study
of Chatbot Interaction Design //The 2024 International
Conference on New Innovations and Advanced
Computing (INNOCOMP).
–
IEEE, 2024.
–
pp. 118-123.
Mohamed I. S., Abdelsalam M., Moawad I. F. Chatbot
support based on ChatGPT for customer relationship
management in educational institutions //6th
International Conference on Computing and Computer
Science (ICCI) 2024.
–
IEEE, 2024.
–
pp. 351-355.
Mashkouk R. et al. Improving production management
using intelligent chatbots //6th International Youth
Conference on Radio Electronics, Electrical Engineering
and Energy (REEPE), 2024.
–
IEEE, 2024.
–
pp. 1-5.
The future of chatbots: statistics of more than 80
chatbots for 2025. [Electronic resource] Access mode:
https://www.tidio.com/blog/chatbot-statistics / (date
of access: 12/20/2025).
