The impact of artificial intelligence on business models: a bibliometric-systematic literature review

A significant number of companies have embraced the use of artificial intelligence (AI) in their operations. Many are incorporating it to enhance their products or services ^[1]. Moreover, it is utilized in the processes of value creation, value proposition and value capture. AI possesses the potential to fundamentally transform the business models (BM) that have been relevant until recently. AI-generated content, such as that produced by Generative Pre-trained Transformer (GPT) technology, is becoming increasingly popular and is causing a paradigm shift in content creation and knowledge representation (Wang et al., 2023). Moreover, AI technologies act as a catalyst for BMI, allowing companies to generate disruptive innovations and potentially transform global competition (Lee et al., 2019). To fully benefit from emerging disruptive technologies, businesses must transform their model concepts (Åström et al., 2022). Therefore, AI is often implemented in corporate digital transformation projects to support businesses and help with guidance on data gathering and analysis (Brock and Von Wangenheim, 2019). With growing interest in AI’s principles, policies and incentives (Fosso Wamba et al., 2021), understanding the implications of AI in BM is paramount.

The term AI was coined in 1955 by John McCarthy as “the science and engineering of making intelligent machines” (Manning, 2020). Since its inception, AI structures have undergone a process of evolution, becoming increasingly sophisticated and adaptable. Whereas an early method of symbolic AI relied on human-created rules for decision-making, modern approaches of machine learning and deep learning have enabled algorithms to learn from data and achieve goals and tasks through flexible adaptation with a fast-learning approach (Haenlein and Kaplan, 2019). Speculative future endeavors aim for broader AI capabilities beyond current technological limits (Parliamentary Research Services and Boucher, 2020).

The concept of the BM has been understood heterogeneously due to a lack of standardization in the past (Wirtz et al., 2016). BM can represent specific attributes of real firms, serve as cognitive or linguistic frameworks for understanding and communicating a business’s structure or function as formal conceptual representations of how a business operates (Massa et al., 2017). A widely accepted components of BM are value creation, which defines how firms create value using resources and capabilities, value proposition, which shows how products and services are delivered to customers and value capture, which defines how firms get revenues and ensure sustainable performance (Clauss, 2017; Foss and Saebi, 2017; Teece, 2010).

The number of studies on the interplay between AI and BM has increased in recent years (e.g. Åström et al., 2022; Foss and Saebi, 2017; Jorzik et al., 2024). The rise of computing power and the advancements in big data technologies have enabled AI to spread out considerably and become a popular research subject (Duan et al., 2019). This is evident in Wamba-Taguimdje et al. (2020), who highlight that the organizational benefits of AI, such as the potential of AI transformation projects to improve organizational performance and processes, ultimately increase overall business value. Huang and Rust (2021) corroborate further by showing that AI can enhance customer engagement by simplifying routine tasks, personalizing data-rich services to individual preferences and fostering deeper relationships with customers. Burström et al. (2021) note that introducing AI interrupts the traditional value-creation processes and influences the BM. Furthermore, Åström et al. (2022) proposed a process framework to illustrate the main activities that companies should perform regarding value creation and value capture for AI BMI and commercialization. Consequently, the ways in which AI is influencing and contributing to various aspects of BM are becoming the growing focus of research.

Given the surge in publications, a literature review is essential to assess and analyze the scholarly impact, trends and patterns (Sauer and Seuring, 2020). In this regard, a study that merits mention is the contribution of Di Vaio et al. (2020), which offers a systematic literature review centered on the role of AI in constructing sustainable BM through its influence on production and consumption patterns. Based on 73 articles from 1990 to 2019, the findings showed that AI can represent the vehicle to meet the Sustainable Development Goals (SDGs). Notwithstanding the valuable contributions of Di Vaio et al. (2020), a comprehensive synthesis of the current state of knowledge on the AI’s impact on BM remains elusive, particularly considering the rapid growth of this topic in the last four years. In addition, since the release of publicly available AI programs, such as OpenAI’s ChatGPT in November 2022 (OpenAI, 2020), Meta’s LLaMA in February 2023 (Meta, 2023) and Google’s Gemini in December 2023 (Google, 2023), it is reasonable to anticipate that a significant volume of novel research has been conducted. This requires a thorough understanding of current advancements. Furthermore, the extant literature does not elucidate the interconnections and overarching tendencies in the field of AI’s impact on BM, leaving key aspects uncertain.

This study aims to provide a thematic exploration and mapping of the developments in the field of AI and BM and to extend the existing knowledge base by developing a framework for AI adoption in BM and suggesting a future research agenda. It is guided by the following research questions: (1) What is the current state of AI and BM research, and what established themes are addressed? (2) Where is the research on AI and BM heading, and what emerging issues need to be addressed in future research?

To address these questions, the study conducts a Bibliometric-Systematic Literature Review’ (B-SLR) of 87 journal articles. It performs a co-citation, co-occurrence and timeline view analyses to reveal the intellectual structure, main thematic clusters and the recent developments in the field of AI and BM. Based on these analyses, it proposes a framework for AI adoption in BM and outlines promising directions for future research.

This study makes a number of relevant contributions to the literature. First, it contributes to the current discussion on AI’s impact on BM by synthesizing and extending the existing knowledge base in the field. Here, in contrast to Di Vaio et al. (2020) who looked at sustainable BM, we did not limit our research to a specific type of BM and, instead, adopted a broader approach to provide a comprehensive and multifaceted overview on the topic. Although the topics of sustainability and circular economy are also covered by this research, they constitute only a part of the research issues addressed in this field, thus providing a greater variety of other perspectives. Second, by undertaking a multi-method literature review, which combines a bibliometric analysis (e.g. Öztürk et al., 2024) with a systematic literature review (B-SLR), this study offers both a synthesis and a theoretical development in the field of AI and BM. Specifically, performing the co-citation analysis, we identified and clustered the literature streams underlying the AI and BM research field, shedding light on its intellectual structure. The co-occurrence analysis allowed the identification of the current and most prevalent topics within research at the intersection between AI and BM. Furthermore, given the rapid growth of the AI and BM literature in recent years, this study offers an up-to-date overview and localizes emerging topics and relevant references by performing the timeline view analysis. Third, by systematically and holistically analyzing the dataset and the bibliometric results, the study extends the research field by developing a framework for AI adoption in BM. Furthermore, by giving particular attention to the recent developments in the field, the study traces future roadmaps for advancing the scholarship.

To synthesize and advance the understanding of the intersection between AI and BM, we applied the B-SLR framework (Marzi et al., 2025). Whereas bibliometric methods provide quantitative indicators related to a research field and focus on the interconnections of the studies within a broader network, SLR involves a qualitative analysis of articles to develop new theoretical frameworks to understand phenomena. Marzi et al. (2025) suggest performing 10 methodological steps of the B-SLR to deliver meaningful and relevant contributions, which allow both a synthesis and a theoretical development of a research field. Table 1 summarizes the B-SLR process applied in this study.

As a first step, we scanned the literature on AI and BM to develop an overview of the research field and establish a set of representative keywords for database search. Following the B-SLR framework, we defined our inclusion/exclusion criteria before approaching the data collection process. In doing so, we relied on a diverse range of key definitions within the domain of inquiry (Clauss, 2017; Foss and Saebi, 2017; Teece, 2010; Wirtz et al., 2016), where BM is considered holistically as an aggregated representation of how a product or service is created, how it is offered to customers and how the value is captured, generating a competitive advantage for the company. As a result, we defined the search query as follows: (“artificial intelligence” OR AI) AND “business model,*” and targeted academic journal English-language articles in the business and management area (step 2).

For data collection, the Scopus database was selected, which is typically adopted within management studies (Balzano, 2022; Casprini et al., 2020). Scopus is globally recognized as a reliable scientific data source and covers academic literature in a broad array of management journals (Step 3). In January 2024, we ran the search query in the title, abstract and keywords, limiting the document type to article or review, the structure type to journal, the subject area to “Business, Management and Accounting”, and the documents to the ones published in English language (Kraus et al., 2020, 2022). As a result, we extracted a raw database of 193 documents (Step 4).

Subsequently, two researchers independently manually reviewed the titles and abstracts of the documents in the raw database to clean the data by removing papers not falling within the boundaries of the selected topic and to ensure that the articles are consistent with the inclusion/exclusion criteria. After the discussion and review of the full texts, when necessary, the number of retained documents totaled 112 (step 5).

Furthermore, to ensure rigor in the journal selection and identify the most impactful studies, we applied a journal ranking from the Academic Journal Guide (AJG) 2021 list developed by the Chartered Association of Business Schools’ (CABS). Specifically, we excluded journals that are not referenced in the list. This has resulted in a final dataset of 87 articles published in 58 journals. The metadata of these documents was retrieved from Scopus in February 2024 to be used in the software to perform bibliometric analyses (Step 5).

Figure 1 illustrates the growth in the number of publications analyzed in this study. The earliest article in our sample dates back to 2016. From that year on, our analysis includes papers published until February 2024, which were the most recent articles available at the time of writing this manuscript. There is a positive trend in the number of publications, which indicates a strong continuation of the development of the literature. This might be connected to the overall interest in AI application in business, which is observed in the scholarly environment. A significant increase in publications can be observed starting in 2019, with the number of publications growing significantly since then and reaching 31 articles in 2023. We recorded an annual growth rate of 18.92% throughout the whole period. Consequently, the average age of publications is relatively low at 2.29 years.

The analyzed articles were published in 58 journals by 277 authors. The top-five journals based on the total number of articles include Journal of Business Research (8 articles); Technological Forecasting and Social Change (8 articles); Business Strategy and the Environment (3 articles); Review of Managerial Science (3 articles) and Technovation (3 articles). Furthermore, we identified the authors who have published most frequently on the topic: V. Parida (8 articles), A. Brem (3 articles) and S. Sjödin (3 articles). V. Parida is the most influential author due to having published the highest number of articles. This is also supported by the fact that his five publications received over 50 citations each.

Table 2 presents the 10 most-cited contributions based on the number of total citations. The average number of total citations of articles in the sample is 42.51. The most cited article is Warner and Wäger (2019) which received almost three times more global citations (935 citations) than the others, both overall and annually. The second most cited article is Di Vaio et al. (2020), with 331 citations. About 56% of the publications in our sample were cited more than 10 times, 34% of publications were cited less than 10 times and 9 articles (10%) have not received citations yet, which could be due to their relative newness, being mainly from 2023.

In terms of country distribution, the analysis of global literature on AI and BM reveals that publications primarily originate from Germany, the USA, China, Sweden, the UK, Australia, Italy, Finland, South Korea, India and the Netherlands. The country distribution illustrates that the interest in AI and BM is global and spread across regions, with special interest from European countries. This aligns with the general forerunner position of European and American research in the management literature (Ferasso et al., 2020). Nevertheless, it is noteworthy that we also detected many publications from Asia and Oceania, which highlights the international attention to these topics. Figure 2 presents a country collaboration map. The international co-authorship is 45.98%. Joint research efforts are emphasized in Northern Europe between Finland and Norway, Finland and Sweden and the UK and the Netherlands, highlighting regional synergies. Additionally, the collaboration between the USA and the UK is highlighted, fostering international research.

Next, we moved to bibliometric analyses and performed a co-citation analysis, a keyword co-occurrence analysis and a timeline view analysis of a combination of keywords and references, providing a comprehensive understanding of the research field (Step 6). This approach is an appropriate way of uncovering patterns and trends that might be missed with manual methods when dealing with a broad scope of review and large datasets.

A co-citation analysis, with the aid of CiteSpace (version 6.3.R2), was conducted to elucidate the intellectual structure and identify seminal works within the research domain of AI and BM. To perform this analysis, metadata from 87 papers was downloaded in the *.CSV format, which was used in CiteSpace for network analysis. CiteSpace’s data conversion process generated 6,861 references from the documents, out of which 6,599 references (96%) were deemed valid for analysis, fitting the acceptable loss threshold of 5% (Chen, 2010, 2018). The software automatically set the analysis period from 2016 to 2024 based on the sampled documents. Time slicing was configured to one year per slice, with text processing applied to titles, abstracts and author keywords. The analysis was run multiple times to ensure accuracy and reliability, aligning with the guidelines provided by Chen (2018), Donthu et al. (2021) and Synnestvedt et al. (2005). The final parameters were set as follows: a Link Retaining Factor (LRF) of 3.0 (i.e. the total number of links is no more than three times the number of nodes), a maximum of 10 links per node (L/N), look-back years (LBY) of −1 (i.e. no citation age was restricted) and a parameter value of 2.0 (minimum citations) for the most frequently mentioned articles (TopN). The top 100% of the most cited or occurred items from each slice were selected, with a maximum of 100 selected items per slice. The network was constructed with “references” as node types, using “Cosine” for strength and “within slices” for scope. Finally, the co-citation network was displayed in an un-pruned and merged network view. This resulted in 80 nodes (references) and 246 links representing the frequency with which two references are cited together (Figure 3). Furthermore, the co-occurrence network was displayed in a cluster view, as shown in Figure 4.

A keyword co-occurrence analysis, performed using VOSviewer software (version 1.6.20), was conducted to uncover and map the conceptual structure of a research field of AI and BM by understanding the relationships among authors’ keywords. The metadata was imported from the *.CSV file, following the methodology outlined by Van Eck and Waltman (2023). The aggregation criteria adopted in VOSviewer was the co-occurrence of author keywords, employing a full counting approach. A VOSviewer thesaurus file was used to correct spelling differences (“artificial intelligence (AI)” and “AI” were merged with “artificial intelligence”, “digitalisation” – with “digitalization” and “business model” – with “business models”). A minimum of two occurrences per keyword was set as the threshold. Out of the 341 identified keywords, 40 keywords met the threshold. The resolution parameter was set to the default value of 1.00. The minimum cluster size was set as five items and small clusters were merged to achieve balanced sizes. As a result, the network consisted of 39 connected keywords and 168 links between them, forming five clusters (Figure 5) (Step 7). This network highlighted primary research themes and their interconnections within the field of AI and BM.

A network displayed in a timeline view, with “author keywords” and “references” as node types, was built to explore the current topics linked to the cited references. A link between a keyword and a reference means that a document with that keyword cited the reference. The analysis was performed using CiteSpace and focused on the recent developments in the field of AI and BM. For this, the following parameters were configured: CiteSpace was set up to show all the links in the network (LRF = −1), while it kept the most current cited references from the sampled documents (LBY = 3, i.e. the network displayed references published up to three years earlier than the citing article). The maximum number of links per node was not restricted (L/N = −1) and a parameter value of 2.0 was set, which is the minimum number of citations that an item in the TopN group must satisfy. As in the co-citation analysis, the selection criterion was set to top 100% of the most occurred items from each slice, with a maximum of 100 selected items per slice. As a result, 70 nodes (keywords and references) and 464 links between them were displayed in a timeline view, composed of five clusters and presented in layers, as shown in Figure 6.

After analyzing the bibliometric results and independently reading the extracted documents, we identified the key theme within each cluster in the field of AI and BM. We used the full sample of documents for SLR (Step 8). The thematic clusters have been analyzed by mapping the state of knowledge and grasping the content of the literature (Step 9). The theorizing perimeter has been established through a conceptual framework and a research agenda, offering deeper insights into the topic under investigation and contributing an original theoretical advancement to the field (Step 10).

The co-citation network, created by CiteSpace, incorporates 80 nodes and 246 links (Figure 3). The most influential references are identified by the number of citations as well as by the extent to which a particular reference is linked to the network. The size of the nodes is determined by the number of citations per reference. The betweenness centrality of a node measures the extent to which the node is a part of paths connecting other nodes in the network (Chen, 2010). Nodes with high betweenness centrality are indicated with the purple ring. They demonstrate a pivotal point of transformation (Beliaeva et al., 2022).

The most frequently cited reference is Teece (2010), which outlined the significance of BM and explored its connections with business strategy, innovation management and economic theory. The second most cited source is Barney (1991), who identified the sources of sustained competitive advantage and linked them to overall competitiveness. The third most cited work is Eisenhardt (1989). Although having the same number of citations as Zott et al. (2011), Eisenhardt (1989) is considered more pivotal for the research trajectory due to a greater betweenness centrality. This result is explained because it is a reference paper for the case study methodology and is indicative of the preferred methodology by citing documents in the sample. The reference of Zott et al. (2011) provided a broad and multifaceted overview of the development of the BM concept.

The color in the network shows when a connection was made for the first time, allowing the identification of the older and the newer parts of the network. As depicted in Figure 3, the earlier period is presented by the blue and green parts of the network, including the time before 2022, and the later period is presented by the orange and red parts of the network, covering the time after 2022. However, the color code is independent of the publication date. For instance, Schumpeter (1934) is shown in orange to indicate it was recently cited, even though the original work dates back to 1934. It is noteworthy that the blue/green network diffuses to a greater extent than the orange/red network.

The sub-network of earlier cited references (blue/green) includes the most cited references by Teece (2010), Barney (1991), Eisenhardt (1989) and Zott et al. (2011). The other encompassed topics are BMI (Foss and Saebi, 2017), qualitative rigor in inductive research (Gioia et al., 2013), BM research (Massa et al., 2017), digital servitization – the process of leveraging digital tools to transition from a product-centric to a service-centric BM (Sklyar et al., 2019) and BM design (Zott and Amit, 2007). In more detailed terms, Zott et al. (2011) noted that there is no consensus among scholars regarding the definition of BM. However, there are several common themes emerging from the literature. These include the view that BM represents a novel unit of analysis, an emphasis on how firms execute business, the influence of firm activities on BM and explanations of how value is created. In the systematic review, Foss and Saebi (2017) identified significant challenges in the existing literature on BMI, including a lack of clarity and the existence of significant research gaps. Gioia et al. (2013) introduced a novel concept development approach designed to enhance the quality and rigor of research. Massa et al. (2017) identified three different meanings of BM and the reasons for the disagreements in BM definitions and their relationship to strategy. Sklyar et al. (2019) argued that within-firm centralization and integration are crucial for the successful implementation of digital servitization. Zott and Amit (2007) demonstrated that a novelty-centered BM design is a significant factor in the performance of entrepreneurial firms.

The group of the later cited references (orange/red) includes research contributions on BM (Baden-Fuller and Morgan, 2010), a theory of economic development (Schumpeter, 1934), an evolutionary theory of economic change (Nelson and Winter, 1982), AI in marketing (Davenport et al., 2020), innovation through co-creation (Sjödin et al., 2020a, b), AI implementation (Reim et al., 2020), interpretations and implications of AI (Kaplan and Haenlein, 2019) and AI capabilities (Sjödin et al., 2021). To elaborate, Baden-Fuller and Morgan (2010) proposed that BM took various forms and could be utilized to classify businesses, serve as sites for scientific investigation or act as recipes for creative managers. Schumpeter (1934) viewed economics as a natural self-regulating mechanism, while Nelson and Winter (1982) discussed how firms and industries evolved over time. Davenport et al. (2020) developed a framework to understand the impact of AI by considering three dimensions: intelligence level, task types and the use of AI in robots. Sjödin et al. (2020a, b) explored how manufacturers and customers collaborated to create digital service innovations, addressing the challenge of balancing efficiency and flexibility in digital transformations. Reim et al. (2020) identified AI as a key driver for BMI and provided a roadmap for AI implementation. Kaplan and Haenlein (2019) clarified the distinctions between AI and related concepts such as Internet of Things (IoT) and big data, as well as the internal and external implications of AI for organizations. Sjödin et al. (2021) explored how manufacturing firms could develop AI capabilities for BMI to scale AI in digital servitization. Furthermore, a part of the later period is a small sub-network not connected to the main network. The ideas revolve around contributions to deep learning, particularly those by LeCun et al. (2015) who explained how deep learning methods and algorithms operated and how they led to significant breakthroughs in processing power.

The co-citation network was further divided into clusters to gain a deeper understanding of the intellectual structure of AI and the BM field. Five distinct clusters emerged from the references, as depicted in Figure 4.

The first cluster (red) is designated as “innovation” due to its predominant focus on innovation-related studies. The largest cluster comprises 19 references with an average publication year of 2009. Thematic key contributions include Chesbrough (2003), who introduced the open innovation paradigm, which integrated both internal and external ideas and technologies. Mariani et al. (2023) offered a systematic overview of innovation types in conjunction with AI. Ferreras-Méndez et al. (2021) demonstrated that BMI can effectively channel a firm’s entrepreneurial orientation into its innovation processes, thereby enhancing the success of new product development.

The second cluster (yellow) is titled “AI” and consists of 15 references with an average publication year of 2015. For example, Dwivedi et al. (2021) provided multidisciplinary perspectives on the emerging challenges and opportunities of AI, including generative AI (Dwivedi et al., 2023). Haenlein and Kaplan (2019) offered an exposition on the historical and future developments of AI. Burström et al. (2021) investigated AI for enabling BMI in industrial ecosystems.

The third cluster (green) is labeled “business models” and encompasses 14 references with an average publication year of 2008. Except the most cited works by Teece (2010) and Zott et al. (2011) belonging to this cluster, other references include Morris et al. (2005), who discussed the entrepreneurial BM, and Chesbrough (2010), who examined the opportunities and threats posed by BMI. Chesbrough and Rosenbloom (2002) further investigated how BM supported capturing value from early-stage technologies and innovations.

The fourth cluster (blue) is titled “case studies” due to its methodological focus. It consists of 13 references, with an average publication year of 2009. Eisenhardt (1989) is the most cited reference, contributing significantly to theory-building from case study research, further elaborated in Eisenhardt and Graebner (2007). Edmondson and McManus (2007) discussed methodological fit in management research, while Braun and Clarke (2006) outlined thematic analysis as a useful method for qualitative research. Parida et al. (2015) used an exploratory case study design to examine the capabilities and practices of multinational manufacturing companies and illustrate how they developed global service innovation capabilities.

The smallest cluster (violet), titled “digital servitization,” contains eight references with an average publication year of 2018. Ardolino et al. (2017) explored how digital technologies such as the IoT or predictive analytics transformed industrial companies. Vendrell-Herrero et al. (2017) empirically examined the impact of disruptive digitalization on B2B interdependencies, suggesting that digital servitization had a different impact on upstream and downstream companies within the supply chain. Kohtamäki et al. (2019) investigated the effects of digital servitization on BM and its application within various ecosystems.

Overall, the co-citation analysis revealed that the field of AI and BM is based on the seminal contributions in management and strategy (e.g. Barney, 1991), innovation (e.g. Chesbrough, 2003) and BM (e.g. Teece, 2010; Zott et al., 2011) literature. More recently, the literature on AI (e.g. Haenlein and Kaplan, 2019) and digitalization (e.g. Sklyar et al., 2019) fueled the research field with new research issues and approaches. Furthermore, key references on qualitative research and case studies (e.g. Eisenhardt, 1989; Gioia et al., 2013) constitute the methodological foundations of the research field.

The co-occurrence network of authors’ keywords highlighted the main research themes and their interconnections within the field of AI and BM, providing insights into topics of scholarly interest. The network, created by VOSviewer, is displayed in Figure 5. The analysis identified five key clusters formed by 39 nodes and 168 links.

Cluster 1 (red) focuses on customer relationships and comprises 11 items. Researchers explored how the adoption of AI technologies and service automation influence customer experience, engagement and business success. For example, AI, used on social media, can enhance customer experience, engagement, satisfaction and purchase intention (Bilal et al., 2024). AI can predict consumer preferences and allow achieving personalized service and precision marketing (Chiu and Chuang, 2021).

Cluster 2 (green) encompasses innovation and entrepreneurship with eight items. Prior studies have investigated BMI and AI applications in different industries. AI enables BMI and should be aligned with ecosystem innovation (Burström et al., 2021). Disruptive BMI such as AI-powered demand forecasting and product design provide effective operation models for handling demand uncertainty, inventory management and timely market responses (Jin and Shin, 2020). Additionally, generative AI has been viewed from a BMI perspective, driving the development of new and improvement of existing BM (Kanbach et al., 2023).

Cluster 3 (blue) centers on sustainability and business ethics, and also contains eight items. Within this research theme, circular economy was detected as the main topic. Chauhan et al. (2022) highlighted that interconnected technologies such as AI, IoT and big data were essential components that collectively enhanced the circular economy’s efficiency and effectiveness. Concurrently, the development of circular BM is supported by the strategic role of AI in data management, processing and mining (Rusch et al., 2023). From an ethical standpoint, it can be posited that the integration of AI into BM necessitates a focus on data privacy, bias, accountability, transparency and compliance (Breidbach and Maglio, 2020).

Cluster 4 (yellow) relates to Industry 4.0 and includes six items. It encompasses such topics as AI implementation in manufacturing and AI-driven operations and supply chain management. For example, Sjödin et al. (2021) identified AI capabilities that enable BMI in manufacturing. Helo et al. (2022) highlighted the value creation opportunities in supply chain management through automated infrastructures and optimized business procedures facilitated by AI.

Finally, Cluster 5 (violet) pertains to digitalization, with six items. Within this research, it should be highlighted that successful digital transformation hinges on overcoming cognitive barriers, reconfiguring digital routines and adopting new organizational forms (Volberda et al., 2021). Furthermore, digital transformation shapes the next generation of services via digital technology, digital strategy, customer experience and data-driven BM (Zaki, 2019).

Overall, AI enhances different BM components and functions, such as marketing and customer management, manufacturing and supply chain management and human resource management. Additionally, AI drives BMI and facilitates the transition towards circular economy BM. Viewed as a part of an overall digital transformation of an organization, the AI adoption impacts operational efficiency, performance, sustainability and has broader ethical and societal implications.

CiteSpace allows the presentation of a timeline view according to different topics and the influential references in each layer (cluster). In this functionality, it is possible to represent cited references and their relationships with authors’ keywords. This is particularly relevant to explore the current topics when setting up parameters to consider the references up to three years before the publication year of an article. Results are shown in Figure 6.

The cluster that grouped the most diversified cited references (Layer 1) evidenced that both AI and BM keywords are associated with them. While the most cited references in this layer are Davenport et al. (2020), Burström et al. (2021) and Kaplan and Haenlein (2019), this layer is cited by 18 documents referring to AI, digital platforms and digitalization (Agarwal et al., 2022; Battisti et al., 2022; Chauhan et al., 2022; Lv et al., 2023; Wang and Su, 2021) and AI and operations management (Helo and Hao, 2022; Mithas et al., 2022; Niu et al., 2023). This coverage is also in line with the reported keywords related to “electronic commerce,” “sales” and “digitalization,” emphasizing that AI is more related to these activities in narrow applications.

The second group of cited references (Layer 2) is represented by the “circular economy” keyword. In this layer, four references stand out, those of Iansiti and Lakhani (2020), Nishant et al. (2020), Awan et al. (2021) and Kolagar et al. (2022). This layer was formed based solely on two citing documents by Sjödin et al. (2023) and Fallahi et al. (2023). Both have studied AI for circular BM, relating BM to sustainability thanks to digital servitization and business ecosystems.

The third cluster (Layer 3) grouped the older references of the network, under the “digital transformation” and “digital servitization” keywords. The key references are those of Davenport and Ronanki (2018), Kohtamäki et al. (2019), Paiola and Gebauer (2020) and Sjödin et al. (2020a, b), among others. It was formed by five citing documents referring to AI capabilities (Sjödin et al., 2021), digital servitization (Chen et al., 2021; Sjödin et al., 2023) and AI-enabled BMI (Burström et al., 2021; Thomson et al., 2023).

The fourth cluster (Layer 4) was represented by the “decision-making” theme. This cluster grouped cited references including Nasiri et al. (2020), Sjödin et al. (2020a, b, 2021, 2023), Ferreras-Méndez et al. (2021), while it is formed by citing documents addressing resources re-orchestration (Attah-Boakye et al., 2023), AI and circular BMI (Sjödin et al., 2023) and Industry 4.0 (Ferreira et al., 2023). A cross-cluster analysis evidenced that the works of Sjödin and colleagues are the main references addressing AI in conjunction with different aspects of BM.

Lastly, Layer 5 is represented by the “generative AI” topic. Most representative references are from Reim et al. (2020), Stokel-Walker (2022) and Dwivedi et al. (2023). Formed by two citing documents, Budhwar et al. (2023) investigated HR management and generative AI and Kanbach et al. (2023) studied the generative AI and BMI.

Overall, the timeline view of the recent developments in the field of AI and BM highlights topics that represent current scholarly interest, such as narrow AI applications, the role of AI in circular economy BM and generative AI. Some of the identified topics, such as innovation, circular economy and digitalization, are also present in other networks signifying that they have achieved broad representation over time in the analyzed sample.

The previous results allowed the development of a theoretical framework and the identification of future research directions in the field of AI and BM. Figure 7 presents the theoretical framework of AI’s role in BM transformation comprising the antecedents of AI adoption, the AI’s impact on BM and the outcomes of such adoption; Table 3 synthesizes the main future research suggestions and provides further readings from selected literature.

This theme focuses on identifying individual, organizational and environmental drivers of AI adoption in organizations (Wang and Su, 2021), such as top management (Jorzik et al., 2024), dynamic capabilities (Warner and Wäger, 2019) or COVID-19 (Ivanov et al., 2022). Future investigations are needed to unveil the motivation of organizations and feasibility factors in adopting AI (Wang and Su, 2021). Another promising topic is to identify similarities and differences of the antecedents of AI adoption in different types of firms like state-owned and private, large companies and SMEs, early-stage and established firms and family businesses. Furthermore, the role of managerial support, AI education of employees, government policy (Wang and Su, 2021) and crises (Ivanov et al., 2022) in AI adoption will provide further insights in the field.

This theme focuses on how AI is integrated into BM, including various BM components, BMI, circular economy BM and digital transformation. The AI-enhanced BM components theme investigates the implementation of AI in different BM functions such as customer relationships (Bilal et al., 2024), operations (Mithas et al., 2022) or human resource management (Minbaeva, 2021). More investigations can explore the value co-creation enhanced by AI in business ecosystems and digital platforms (Battisti et al., 2022; Chiu and Chuang, 2021) and the AI-based value creation in different contexts and industries (Kulkov, 2021). Moreover, future studies may investigate how AI-personalized experiences influence consumers’ perception of value creation and brand loyalty. Other topics include AI applications on social media platforms (Bilal et al., 2024) and customers’ preferences of services automation versus human-based services. In the supply chain and operations, further studies can unfold the AI’s role in the identification and mitigation of risks in the supply chain, the deployment of configurations of AI technologies to improve operations (Mithas et al., 2022) and the integration of AI in operational workflows to increase scalability (Helo and Hao, 2022). Additionally, more knowledge is needed to understand how AI can be integrated into product design and impact the product lifecycle (Jin and Shin, 2020; Rusch et al., 2023). Furthermore, topics related to employees that deserve further investigation involve the AI use in employees’ routines and the AI’s impact on workforce skills acquisition (Minbaeva, 2021).

The AI-driving BMI theme encompasses studies exploring the role of AI technologies and capabilities in BM transformation and innovation (Burström et al., 2021; Sjödin et al., 2021). Promising future research directions include the role of stakeholders in BM transformation (Burström et al., 2021), social innovations fostered by AI (Battisti et al., 2022), and implications of generative AI for BMI in different industries and contexts (Kanbach et al., 2023).

The AI impacting circular economy BM theme focuses on whether and how AI influences the transition towards circular and sustainable BM (Chauhan et al., 2022; Di Vaio et al., 2020). Investigations are needed to understand how to choose appropriate AI technologies for circular BM (Chauhan et al., 2022), how resource efficiency, waste reduction and material recycling can be improved with AI, and how AI-based recovery practices can help solving issues with end-of-life products (Fallahi et al., 2023; Sjödin et al., 2023). Also, the challenges in implementing AI-based solutions in circular BM require further exploration.

AI as a part of the digital transformation theme is centered on the impact of digital technologies, including AI, on BM, and the implementation of digital transformation in a company, its opportunities and challenges (Chen et al., 2021; Zaki, 2019). More investigations are needed to understand how the convergence of different technologies (AI, IoT and big data analytics) help improve BM digitalization (Nguyen Dang Tuan et al., 2019). AI is also changing the organizational structure and relationships, and more studies are needed to understand the fluidity of firms’ and ecosystems’ boundaries and the new intra/inter-organizational forms in the digital era (Volberda et al., 2021).

Turning to implications of AI adoption in BM, scholars have investigated both organizational and societal consequences of AI, including operational efficiency, financial performance (Gebauer et al., 2020), environmental, social and ethical implications (Breidbach and Maglio, 2020; Ferreira et al., 2023). There is also a need to examine the first and second-order effects of AI implementation by companies. For example, the long-term impact of AI adoption on customers’ brand perception, loyalty and trust warrants deeper examination (Bilal et al., 2024; Morosan and Dursun-Cengizci, 2023). Additionally, an intriguing avenue to explore is whether and how it is possible to balance customization with efficiency using AI. It is also important to analyze AI implications for employees such as job security, compensation, satisfaction and well-being (Budhwar et al., 2023). Further investigations are also needed to deeper understand the AI’s role in firm’s sustainability, resilience and agility. For example, AI can facilitate social inclusiveness and foster corporate social responsibility (CSR). Moreover, the dark side of AI technologies and ethical concerns surrounding AI-driven BM compliance, data protection, privacy and transparency need to be explored in greater depth (Budhwar et al., 2023).

This research aimed to provide a thematic exploration and mapping of the developments at the intersection of AI and BM and to extend the existing knowledge base by providing a theoretical framework and a research agenda. This study conducted a B-SLR (Marzi et al., 2025) by analyzing the Scopus dataset with co-citation analysis, co-occurrence analysis, timeline view analysis and qualitative analysis of themes.

Findings revealed a growing interest from scholars since 2019 and the peak of publications in 2023, representing a need to develop this field further considering the impact of AI in many sectors. Analyzing the co-citation network in depth by forming five clusters, it was possible to identify that, despite the clusters of AI and BM, the field has its roots also in innovation and digitalization literature. One additional insight is the cluster related to the case study methodology, indicating that the field is at an exploratory stage. Furthermore, by exploring the keyword co-occurrence network, five thematic clusters were formed, in which it was possible to identify diversified arrays of AI’s impact on BM. Researchers are interested in understanding customer relationships, innovation and entrepreneurship, sustainability and business ethics, Industry 4.0 and digitalization topics. The timeline view provided a deeper exploration of recent references related to authors’ keywords and suggested future developments from these findings. Finally, after qualitatively analyzing the results, it was possible to propose a theoretical framework of the AI’s role in BM and future research agenda.

This research presented limitations. The first is related to the reliance on a single Scopus database instead of mixing the results from different databases. This is also a software limitation as different metadata files provided by databases could result in technical errors. Although Scopus covers a wide range of management journals, it does not include all possible journals in the field. Second, the research methodology could present different results because network analysis can produce multiple arrays of results and not represent, for example, the less cited references. The software parameters utilized in this research focused on readability for further exploration. Lastly, our interpretations and suggestions for future research directions can vary according to other researchers’ expertise (Kraus et al., 2024). The focus for future research directions was based on AI and BM perspectives, which could vary if taken from the Information and Technology viewpoint. Finally, we recommend researchers consider similar research to track further developments in the rapidly evolving research field of AI and BM in order to refresh our understanding of AI’s impacts and contributions to theory and practice.