With the advent of technological innovations, firms are improving their operations efficiency in terms of efficiency, quality and cost saving efforts to ensure that their processes run efficiently and interact globally [1]. AI refers to those technologies that make the machines think and learn how to solve problems that need human intelligence-commerce is the utilization of technology to buy and sell products and services on the internet. Firms can use websites, mobile applications and social media to communicate effectively with the customers. With the advent of this new technology era, there has been a transformation from conventional modes of commerce to the more convenient ones, expanding markets and changing consumer behavior [2]. The use of AI has been seen in the e-commerce industry for better customer experience, optimization and innovation. Some of the best e-commerce sites like Amazon, Alibaba and eBay have collaborated with AI in optimizing the supply chain, personalization and customer service.

This review aims to discuss the transformational influence of artificial intelligence (AI) in revolutionizing the e-commerce realm through the concept of personalization, enhanced customer experience, improved operations, and security. AI has already demonstrated its influence on e-commerce through the aspects of personalization, customer service automation, and fraud detection.

However, this review focuses on the development of personalization methods, customer experience, and improvement of operations in the e-commerce sector, considering new tendencies such as voice commerce and blockchain technology, as well as the ethical issues of the application of AI. This research will shed light on the existing situation in AI in the sphere of e-commerce. The scope of this review is examining customer-facing artificial intelligence (AI) applications and operational AI applications, which prioritizes peer-reviewed empirical studies, systematic reviews, and widely implemented industrial AI solutions.

The review was conducted through a structured search of peer-reviewed literature, primarily journal articles, conference proceedings, and systematic reviews, with a wide geographic coverage. The scope of the review comprises papers published mainly between 2018 and 2025. The databases used were Scopus, Web of Science, IEEE Xplore, Science Direct, Springer Link, Google Scholar, and ACM Digital Library. Search terms comprised combinations of artificial intelligence, e-commerce, machine learning, personalization, application systems, chatbots, and sentiment analysis.

The role of AI in the world of e-commerce has been significant as it has changed it for the better through personalization, customer service, operational efficiency, and security. Personalization driven by AI has allowed retailers to analyze user behavior like browsing history and purchase history and give them recommendations on what products would suit them. The personalized way of selling has increased engagement and sales [3]. AI in the field of customer service provides the use of chatbots and virtual assistants which handle all kinds of inquiries round the clock without any delay. The AI-driven chatbots allow communication between human users through natural language in the customer service domain [4]. From the operational point of view, AI helps in managing the supply chain and inventory effectively. Through the application of AI technology, e-commerce companies can achieve effective operations, decision making, and personalized services for their customers. Furthermore, the security in the field of e-commerce is also improved by using AI technology to detect fraudulent activities. New trends such as voice commerce using AI and smart assistants are changing the online shopping space. Voice commerce and smart assistants provide a smooth and convenient online shopping experience for consumers and, therefore, revolutionizing the e-commerce industry [5].

In general, AI technology has changed the online shopping space in a way that makes the online shopping process more personal, efficient, and cost-effective. For example, Amazon utilizes AI technology to personalize product recommendations for consumers, while Alibaba uses AI to make consumer demand predictions.

With the rapid development of e-commerce, AI-based personalization has turned into a disruptor of this field, transforming consumer engagement methods [6]. This innovative method makes use of algorithms and machine learning capabilities for the creation of personalized user experience in e-commerce. With the help of analysis of huge sets of data, including consumer behaviors, preferences, and previous interactions, as shown in Figure 1Figure 1, AI enables prediction of relevance of the content, product suggestions, and marketing. The modern customer is becoming more demanding when it comes to personalized interaction, which may include customized product recommendations or dynamic content recommendations or precise targeting of advertising campaigns [3].

The most important AI technologies, which are collaborative filtering and natural language processing, contribute to making the process of personalization possible. Collaborative filtering suggests recommendations based on the patterns in users’ behavior, while natural language processing increases the capability of recognizing user intents, making searches and recommendations even more accurate. By processing information from the users’ interaction, AI continuously learns about the preferences of each particular user and adjusts its recommendations accordingly [7].

Figure 1. AI applications in e-commerce [8].

One key feature of personalization using AI technology is the capacity of the technology to learn and become even better with time. Through continuous data collection from the interaction between the consumer and the e-commerce website, the accuracy of the models developed through AI becomes even better. This will not only enhance personalization but also improve customer engagement, satisfaction, and finally conversion. Continuous learning will ensure that the e-commerce platform is engaging for the customer as well as personalized which will help create a bond between the business and the customer [7].

The review indicates that AI-driven personalization is mature and is the most widely adopted application of AI in E-commerce. This system improves purchase intention, conversion rate, customer engagement and customer loyalty. However, to maintain consumer trust, data privacy, algorithmic bias, and transparency are challenges in the suggested algorithms.

Dynamic pricing refers to an advanced technique which involves changing prices in real time through consideration of several factors including demand, supply, competitor price levels, customer behavior, and inventory levels. This indicates that dynamic pricing is an essential application in e-commerce as it enables firms to respond efficiently to market uncertainty. Evidence from studies indicates that Artificial intelligence serves as one of the major means of supporting dynamic pricing strategies due to its ability to analyze large datasets and use machine learning techniques for identifying appropriate pricing levels [9]. As a result, e-commerce businesses can take full advantage of their operations and generate high revenues while increasing the volume of sales [10]. Nevertheless, challenges like customer perceptions of unfair pricing, lack of transparency in algorithmic decision, data quality requirements, and the risk of excessive price fluctuations that may negatively affect consumer trust. Therefore, while AI-driven dynamic pricing provides significant opportunities for improving e-commerce competitiveness, its effectiveness depends on balancing revenue objectives with fairness, transparency, and customer acceptance.

AI-powered customized marketing is widely applied in recommendation systems, targeted advertising, personalized email campaigns, dynamic pricing, and customer relationship management platforms. These applications allow businesses to move from traditional mass marketing toward individualized customer experiences based on behavioral, demographic, geographic, and contextual data.

One-to-one marketing (also known as customized marketing and customization) represents a highly developed form of service marketing [11]. One to one marketing focuses on differentiation of products because of providing a unique experience with the product for each and every individual customer [12]. This concept provides the foundation for AI-based personalization, where customer identification, segmentation, interaction, and customization are supported through advanced analytics and machine learning techniques.

Figure 2. Artificial intelligence in e-commerce market size [14].

Personalized Marketing is termed as utilization of data and technology by the organizations to provide personalized messages and offers to the customers [13]. Personalization can be in the form of recommendations, advertisements, email offers and dynamic pricing. Personalization uses demographic data, geographic data, behavioral data, contextual data, and more types of data. Figure 2Figure 2 illustrates the rapid growth of the AI market in e-commerce, reflecting the increasing adoption of AI technologies across the industry. This growth highlights the expanding use of AI-driven solutions to enhance customer experience and improve operational efficiency [14]. Nevertheless, AI-based personalization has challenged with, privacy risks, excessive customer tracking, algorithmic bias, lack of transparency, and lack of customer trust. Thus, AI-enabled customized marketing requires balancing the benefits with responsible data use and customer control over personal information.

Chatbots and virtual assistants are AI-based devices that enable simulations of human conversation and offer instant support to users, mainly in customer care services. Such intelligent devices, which are often embedded with NLP, can do numerous jobs such as answering frequently asked questions, guiding customers, making reservations, selecting products, and offering recommendations to customers [14]. In e-commerce, these technologies are commonly applied in customer support, product search assistance, order tracking, personalized recommendations, and automated sales interactions through websites, mobile applications, and messaging platforms. Empirical findings shows that Research further shows that conversational AI improves customer engagement by providing personalized and immediate assistance based on customer queries and preferences. Virtual assistants extend these capabilities by supporting business operations, information retrieval, and decision-making processes [15].

Sophisticated applications, for instance, CAVA from Lenovo, employ deep learning and voice/face recognition technologies for efficient data processing and improved user engagement [16]. These systems improve the shift from rule based chatbots to intelligent conversional agents capable of understanding context, learning from interactions, and providing more human-like responses.

AI-based systems are bringing a huge change in terms of not just customer service but also in terms of personalization, which helps businesses serve their customers faster, better, and more effectively without necessarily needing any human interference [17]. In e-commerce, AI-enabled customer support is widely applied through automated chat systems, virtual assistants, recommendation platforms, complaint management systems, and personalized communication tools. As far as customer service goes, AI-enabled customer support uses NLP and machine learning to answer queries, automate responses, and offer timely assistance, thus providing much higher service efficiency and satisfaction.

On the other hand, AI-based personalization is moving beyond the conventional recommendation engines based on only past purchases or browsing behaviors of users. Research by shows that next-gen systems integrate multiple data sources, including browsing data, purchase histories, customer interaction records, and social media activities to predict customer preference and deliver more relevant recommendations [6]. This way, through AI, businesses are able to identify and accommodate the needs of customers in a seamless manner to drive up their engagement and sales through highly relevant content. Figure 3Figure 3 illustrates AI-driven personalization techniques, highlighting how artificial intelligence enables personalized recommendations, customer segmentation, and tailored shopping experiences in e-commerce environments [16]. Such AI-driven personalization contributes to improved customer engagement, higher conversion potential, and stronger customer relationships by providing individually tailored experiences. However, the literature also identifies limitations associated with AI-driven customer support systems, including difficulties in handling complex or emotionally sensitive inquiries, dependence on accurate and representative training data, privacy concerns arising from customer data collection, and reduced customer satisfaction when human interaction is preferred. Therefore, effective implementation requires combining AI efficiency with appropriate human oversight and ethical data management.

Figure 3. AI-driven personalization techniques [16].

AI-based sentiment analysis enables e-commerce firms to analyze large-scale customer feedback from reviews, ratings, social media, and support interactions, transforming unstructured opinions into actionable insights for improving customer experience and business decisions [18]. Early studies show that machine learning-based sentiment analysis improved automated opinion classification; however, limitations remained in understanding contextual meanings, sarcasm, and domain-specific expressions [19]. Recent advances in deep learning and transformer-based models have improved contextual understanding and classification accuracy, particularly for complex and multilingual customer reviews. Research demonstrates that advanced models such as LSTM and transformer-based approaches (e.g., BERT) provide improved performance by capturing relationships between words and understanding contextual changes in customer opinions [20][21]. These approaches are increasingly applied in e-commerce for review analysis, customer satisfaction monitoring, and reputation management.

More advanced research in the field focuses on aspect-based sentiment analysis (ABSA), which identifies sentiments related to specific product or service attributes such as quality, speed of delivery, pricing, or customer support. ABSA was formulated by [22] as a benchmark dataset and evaluation framework proving the value of the approach in analyzing detailed information about customer experiences. The literature shows that ABSA provides more detailed insights than general sentiment classification because it enables firms to identify specific strengths and weaknesses of their products and services and prioritize improvement efforts. Sentiment analysis plays an important role in terms of managing reputation and retaining customers. Sentiment analysis allows the companies to detect negative trends in sentiment and to act to solve the problems in advance. In addition, combining sentiment analytics with recommender systems allows the companies to personalize recommendations with the help of emotions [23]. These applications demonstrate that sentiment analysis contributes to improved customer relationship management, faster response to market changes, and more personalized customer experiences.

Even though there are many benefits studies highlight several limitations affecting the adoption of AI based sentiment analysis. These include domain adaptation, sarcasm detection, multilingual support, and biased datasets. Therefore, future improvements in explainable AI, large language models, and context-aware systems are important for increasing the reliability and fairness of sentiment analysis applications.

AI based demand forecasting and inventory management form integral parts of supply chains in e-commerce business environments. Forecasting techniques utilizing AI such as recurrent neural network (RNN) models and hybrid models involving ARIMA-LSTM models have been found to make more accurate predictions about future demand for products compared to classical statistical techniques. For instance, a study that employed the use of ARIMA and LSTM models on sales information from e-commerce companies showed great improvement in forecasting accuracy, allowing for improved decision-making concerning inventory levels [24]. Improved forecasting accuracy contributes towards improved inventory replenishment strategies and inventory allocation, hence less overstock and stockouts [25].

The evidence further shows that AI-driven inventory management extends beyond demand prediction. Early research comparing traditional time-series models such as ARIMA with artificial neural networks demonstrated that neural networks outperform statistical approaches when dealing with nonlinear and highly variable datasets [26]. Neural networks are capable of learning hidden relationships within historical sales data without strict assumptions about data distribution. Further improvement in prediction accuracy was achieved through the adoption of long short-term memory (LSTM) networks that tackled the vanishing gradient issue that was associated with conventional recurrent neural network models. The purpose of using an LSTM model is that such models have the ability to maintain long-term dependencies in sequential data, making them very useful in time-series demand forecasting [20].

Research also indicates that reinforcement learning and predictive analytics improve inventory optimization by dynamically determining replenishment policies and supporting multi-echelon inventory management. Moreover, integrating external data sources such as weather conditions, macroeconomic indicators, promotions, and social media signals enables more responsive demand sensing and supply chain planning [27].

Despite these advantages, there are implementation challenges including poor data quality, limitations in interpretability of the model, high computational cost and infrastructure constraints particularly small and medium-sized enterprises (SMEs) Recent studies recommend that explainable AI and cloud-based e forecasting systems may help overcome these barriers and improve the accessibility of AI-driven forecasting solutions. In summary, AI-based forecasting and inventory management systems offer considerable advances in terms of accuracy, cost efficiency, and responsiveness of the process. The integration of deep learning, hybrid approaches, and reinforcement learning methods allow for better performance of e-commerce companies.

The literature shows that AI-driven automation in warehouse improves operational efficiency, and warehouse productivity through the integration of robotics, computer vision, and intelligent optimization systems. Studies of automation in the supply chain reveal that AI tools like computer vision item recognition and collaboration robotics considerably increase the efficiency of processes through adaptation to changing order volumes [28]. Moreover, the use of systematic review reveals that the combination of machine learning and warehouse management systems helps with layout optimization and labor costs reduction [28].

Several empirical findings further show that the traditional labor-based system inefficient. Because of that there is a need for an approach that would help make fulfillment centers more effective, cheaper, and scalable. Such an approach is represented by the integration of robotics and computer vision along with intelligent optimization algorithms. As a result, AI applications in automated warehouses in e-commerce are growing in terms of volume and order and makes them AI application very important. These tools include, machine learning algorithms, deployment of autonomous mobile robots (AMRs), CNN vision systems [29]. Machine learning algorithms are extensively used for picking route optimization, [30] note that picking order takes a considerable share of costs in a warehouse and its optimization can greatly decrease fulfillment time and labor costs. AMRs equipped with sensors and algorithms perform navigation in a variable environment through the utilization of real-time information [31]. CNN vision systems improve picking accuracy by detecting goods under different light conditions and packaging designs [29].

A further important area of applying AI technologies to automated warehouses includes the field of predictive maintenance. Predictive maintenance minimizes machine downtimes, minimizes maintenance expenses, and improves continuity. As stated by [32], data-based prediction systems are crucial elements of smart manufacturing and logistics ecosystems, thus providing a possibility for asset management practices to be proactive. The area of HRC is also of interest in contemporary fulfillment centers. Collaborative robots do not fully replace human operators but instead provide assistance to them in completing monotonous or physically difficult tasks and making the workplace safer and more productive. AI helps distribute task loads between human operators and robots in an optimal way.

However, the implementation of automated warehouses entails several drawbacks, including the significant amount of initial financial investments, security risks associated with cybernetics, workforce replacement concerns, and integration difficulties. Moreover, decision making through algorithms needs to be transparent and interpretable to ensure proper operation and compliance with safety standards.

The last-mile delivery optimization process happens to be one of the most complicated and expensive stages in the e-commerce chain of delivery. Because of that, there has been an increasing interest in the role of artificial intelligence (AI) in optimizing the last-mile delivery process [33]. Reviewing autonomous last-mile delivery technologies reveals how AI-powered decision algorithms contribute to route optimization and parcel distribution in dense urban areas [34]. It is also proven that using these technologies in conjunction with fleet management technologies contributes to better results in last-mile deliveries [35].

Last-mile delivery constitutes the final phase of the supply chain process, involving transportation of goods from distribution facilities to their final destinations, which are usually individual consumers. For e-commerce operations, this process constitutes the costliest and most operationally complicated part of the logistics system. Growth of online shopping activities together with customers’ demand for same day or next-day delivery services requires effective optimization strategies. Artificial Intelligence plays an important role in this respect in ensuring efficient route planning and delivery process [36]. Modern approaches to optimization using AI combine machine learning principles with heuristic/metaheuristic approaches. According to [37], nowadays data-driven decision-making systems are widely used in last-mile delivery processes.

Predicting delivery time and optimizing route order is best done by machine learning techniques. This entails analyzing data streams coming from GPS sensors, traffic data collectors, and customer scheduling software. The result is the use of intelligent systems, which cut fuel costs, reduce delays, and increase customer satisfaction by providing correct delivery time windows [36]. There have also been significant innovations regarding the adoption of delivery technologies in last mile delivery. In this respect, technologies like ground robots and drones are used for last mile deliveries. Such AI-driven delivery technologies are powered by computer vision, sensor fusion, and real-time routing algorithms. According to [38], drone delivery systems may help cut delivery time and the impact on the environment, especially in densely populated and remote areas.

AI also enables delivery consolidation and crow shipping methods, in which optimization algorithms allocate deliveries to drivers or transportation network systems. This strategy makes delivery more flexible and eliminates idle capacity in the vehicles. According to [37], algorithmic coordination is necessary for the achievement of balancing cost reduction and efficiency. Sustainability is another important factor in the last mile optimization process. Routing systems driven by AI help reduce the environmental impact through reduction in the distance covered and vehicle load. According to research, a good routing strategy is capable of reducing environmental effects greatly while keeping the delivery speed as it is [36]. Furthermore, integration of electric vehicles into the optimized AI fleet is also a factor favoring sustainable logistics.

Although there are a lot of benefits provided by the AI last mile delivery optimization, it also faces certain problems such as lack of data privacy, transparency of algorithms, limited infrastructure, and cybersecurity issues. In addition, the high capital needed to implement autonomous deliveries might be a barrier for small and medium-sized companies. However, AI based last mile delivery optimization increases efficiency, saves money, provides accurate deliveries and helps to reach sustainable objectives in e-commerce logistic chains. With the help of dynamic routing, predictive analysis, and autonomous deliveries AI makes last mile deliveries efficient competitive advantage.

The AI technologies examined in this review serve different functions. However, they are complementary rather than competing technologies. The purpose of recommendation models is to enhance customer personalization which predicts user preferences and improve product discovery. The NLP systems, including chatbots, virtual assistants, and sentiment analysis, focus on understanding and responding to customer interactions. Forecasting models such as LSTM and hybrid ARIMA-LSTM, on the other hand, facilitates demand prediction and inventory planning, while optimization algorithms support operational decision-making in warehouse automation, routing, and last-mile delivery. These AI technologies contribute to e-commerce performance by improving customer experience, operational efficiency, and supply chain responsiveness. However, their effectiveness depends on data quality, computational resources, model transparency, and integration with existing business processes, highlighting the need for balanced adoption of customer-facing and operational AI solutions.