Human and technological actors shaping digital transformation capability: an integrative review

The digital economy, accelerated by the global adoption of artificial intelligence (AI), poses unprecedented challenges to the competitive capabilities of incumbents (Caputo et al., 2021; Holmström, 2022; Krakowski et al., 2023). To adapt to rapid technological advancements like the rise of cognitive technologies, firms must embrace DT (Biloslavo et al., 2024). Cognitive technologies are highly adaptable and autonomous technologies that emulate and augment human cognitive processes like information perception, memory, and representation (Brea and Ford, 2023) and include technologies like AI, machine learning (ML), Internet-of-Things (IoT) and related interfaces. DT is a priority for incumbents, requiring them to fundamentally renew their capabilities, using cognitive technologies, to radically improve their innovation potential and reconfigure value creation and delivery (Gong and Ribiere, 2021; Verhoef et al., 2021; Volberda et al., 2021). Industry expenditure on DT is projected to more than double by 2027, reaching 3.9 trillion US dollars by 2027 (Statista, 2024). However, many firms struggle to achieve their DT ambitions (Block, 2022; Gillani et al., 2024; Neiroukh et al., 2024), compelling managers to reassess how well they can leverage human–machine collaboration (Agostini et al., 2023; Li et al., 2023; Montreuil, 2023).

Incumbents’ challenge is further intensified by competition from born-digital startups who are harnessing cognitive technologies. Startups like Schein, and NotCo demonstrate how novel human–machine collaborations can reshape value creation and delivery. Schein, a Chinese e-commerce leader, uses cognitive technologies to identify social media “micro-trends” and rapid testing-and-repeating cycles, introducing thousands of new fashion products daily, overtaking competitors like H&M and Zara (Uchańska-Bieniusiewicz, 2024). NotCo, a Chilean food tech unicorn, leverages machine learning to develop plant-based food alternatives, generating 18.5 billion US dollars in revenue in 2019 (Bedoya et al., 2022; Mansilla-Obando et al., 2023). In contrast, incumbent firms often struggle to integrate human and technological actors in their DT efforts as the roles, mechanisms, and mindsets of human and technological actors are unclear, underscoring the need for the current study.

DT benefits from a corporate entrepreneurship perspective, integrating human and technological actor dynamics such as management and employee roles, learning and knowledge relevant for the transformation process (Nadkarni and Prügl, 2020), enabling incumbents’ sensitivity and responsiveness to new value creation opportunities and avoid becoming obsolete (Gillani et al., 2024; Kraus et al., 2022). McKinsey & Company, for example, developed Lilli an AI team-mate, to support their global network of consultants by leveraging the firm’s intellectual property to creatively prepare for partner meetings, and “sense check” proposals, thereby serving as a catalyst to reshape work practices. Unlike digitalization studies, which explain how firms enhance existing processes with digital technologies (Annarelli et al., 2021; Caputo et al., 2021), DT, as used in this study, demands overcoming entrenched mindsets, beliefs and behaviours to embrace cognitive technologies (Gong and Ribiere, 2021; Verhoef et al., 2021). As DT research has surged (Vial, 2019; Volberda et al., 2021), studies are increasingly fragmented across multiple disciplines (Troise et al., 2022). Therefore, this study conducts a systematic review (Kraus et al., 2020; Tranfield et al., 2003) and integrates findings from the disparate DT literature related to cognitive technologies, focusing on incumbents. We take a socio-cognitive, rather than a socio-technical, perspective and address the following research questions:

The socio-cognitive perspective, with foundations in psychology and social cognition research (Bandura, 1989), offers a framework for examining micro-level individual and collective group cognitive perceptions and behaviours relevant to organization-wide DT processes and outcomes (Hadjielias et al., 2021). By recognizing cognitive technologies as performing high-level autonomous tasks, it emphasizes how employees’ beliefs, attitudes, and mental models influence their interactions with these technologies (de Paula et al., 2022). This perspective allows scholars to explore how ingrained attitudes and values, anchored in established norms, can inhibit firm renewal (Smith and Lewis, 2011; Wimelius et al., 2021). In contrast, the socio-technical perspective has been critiqued for its dominant focus on technical systems and performance-driven improvements, often overlooking human learning due to inherent biases (Turnheim and Sovacool, 2020). Thus, the socio-cognitive perspective is more appropriate for this study and has been applied to explain several phenomena in organisational management (Wood and Bandura, 1989), from entrepreneurial intentions (Boyd and Vozikis, 1994; Krueger, 2000; Shepherd and Krueger, 2002), adaptive cognition (Haynie and Shepherd, 2009), to human–machine convergence in next generation computing (Jha and Singh, 2022).

We review 83 articles using content analysis and thematic coding (Gioia et al., 2012; Hsieh and Shannon, 2005) to contribute to the DT literature in the three ways. First, we address the fragmentation in the current DT literature regarding cognitive technologies. By synthesizing the DT literature at a micro-foundational level, we aim to offer a more nuanced understanding of how human–machine collaboration relates to strategic flexibility and firm renewal (Chen et al., 2023; Li et al., 2023; Rialti et al., 2020). Prior studies on DT capabilities, such as knowledge search (Gavetti, 2005), knowledge management (Cepeda and Vera, 2007; Nonaka et al., 2016), and transactive memory systems (Argote and Ren, 2012), have focused on human actors without considering the role of cognitive technologies that until now have been the purview of human actors. Our review seeks to bridge this gap, offering a synthesis of the multidisciplinary literature and closing the DT literature lag in addressing this emergent phenomenon.

Second, we clarify the ambiguity surrounding employees’ mindsets in enabling or impeding firms’ DT capability (Hildebrandt and Beimborn, 2022; Laamanen and Wallin, 2009). While previous research indicates that an appropriate mindset is crucial for firms to adapt routines and commit to structural changes (Mugge et al., 2020; Volberda et al., 2021), it is unclear what constitutes such an appropriate mindset. DT relies upon integrative knowledge flow, both enhancing and benefitting from corporate entrepreneurship (Chen et al., 2024; Cheng et al., 2024), yet conflicting beliefs about technology (Solberg et al., 2020) and the challenge of balancing the contradictory demands of exploration and exploitation activities create obstacles to DT (Smith and Beretta, 2020). This review clarifies how employees’ mindsets affect the integration of cognitive technologies into DT capabilities, as reported in the current literature.

Third, we integrate organizational and individual perspectives using a socio-cognitive perspective to analyse the tensions that emerge as firms adapt their capabilities to leverage cognitive technologies, extending the practical implications of DT paradoxes (Qin, 2023). Socio-cognitive paradoxes encapsulate underlying tensions that stem from differing employee attitudes, values and beliefs across organizational levels, and are often resistant to change (Ertl et al., 2020; Vial, 2019). Socio-cognitive paradoxes encapsulate underlying tensions that stem from differing employee attitudes, values and beliefs across organizational levels, and are often resistant to change (Ertl et al., 2020; Vial, 2019). By focusing on the human element of DT, this review offers a novel approach to understanding how socio-cognitive paradoxes impact firms’ renewal efforts (Smith and Beretta, 2020; Smith and Lewis, 2011).

Digital transformation (DT) is defined as a process of strategic renewal on the basis of cognitive technologies’ ability to radically improve the firm’s innovation capability and redefine its value creating potential (Gong and Ribiere, 2021; Verhoef et al., 2021; Volberda et al., 2021). Strategic renewal refers to entrepreneurial behaviour within established firms and has been denoted by several different terms such as corporate entrepreneurship, intrapreneurship, corporate venturing, and strategic entrepreneurship (Glinyanova et al., 2021). Nadkarni and Prügl (2020) recommend examining DT from a corporate entrepreneurship perspective, as it is beneficial to address the three DT aspects of interest to this study– micro-foundations, mindsets, and paradoxes.

Micro-foundations of DT. This review examines the micro-foundations of DT as a dynamic capability, revealing how firms structure their capabilities, processes, and routines to adapt to technological change (Eisenhardt and Martin, 2000; Teece et al., 1997). Dynamic capabilities theory explains how firms sense, seize, and reconfigure capabilities in response to changing environments (Schilke et al., 2018; Teece et al., 1997), particularly pertinent for firms operating in technology contexts where persistence in the same routines is hazardous (Zollo and Winter, 2002). Scholars have called for a focus on micro-foundations as mechanisms that ground dynamic capabilities’ higher order concepts empirically (Annarelli et al., 2021; Bojesson and Fundin, 2020; Eisenhardt and Martin, 2000).

We extend the micro-foundations of dynamic capabilities to DT as a capability by three dimensions: knowledge search, knowledge management, and transactive memory. Traditionally orchestrated by human actors, these dimensions also relate to cognitive technologies as key technological actors. First, knowledge search involves seeking external knowledge to drive innovation (Gavetti, 2005; Gavetti et al., 2012; Wang et al., 2024; Zollo and Winter, 2002). For example, Netflix employs a sophisticated AI-driven recommendation system that continuously searches for patterns in user behaviour and content characteristics to personalize customer offerings (Neiroukh et al., 2024). Second, knowledge management coordinates people, technology, and processes for the acquisition, transfer, and storage of knowledge (Cepeda and Vera, 2007). For example, Rialti et al. (2020) found big data analytics and knowledge management is related to innovation capability and strategic flexibility in firms. An example of this is US copper mining firm, Freeport, who deploys machine-learning algorithms to leverage knowledge from the firm’s metallurgists and operators, radically enhancing the agility of their milling processes (Buckley et al., 2023). Transactive memory systems refer to collective systems for storing and sharing knowledge (Argote and Ren, 2012; Bryant, 2014; Lewis and Herndon, 2011; Lewis et al., 2005). McKinsey’s AI assistant Lilli, described as a “thought-sparring partner” (McKinsey & Company, 2023), exemplifies this. As illustrated through the three dimensions of knowledge search, knowledge management, and transactive memory systems, cognitive technologies as actors can function as tools, assistants or collaborators with the organisations and employees which use them (Rouse and Spohrer, 2018).

Mindsets in DT. Literature shows that mindsets, people’s beliefs about their own resources and the resources in their environment, are crucial in shaping their understanding of DT initiatives (Cetindamar Kozanoglu and Abedin, 2021; Solberg et al., 2020). These beliefs can significantly influence whether individuals support or resist firm DT efforts (Solberg et al., 2020). Indeed, literature suggests that understanding how “digital mindsets” can be cultivated to explain the benefits of DT to key stakeholders is a crucial step (Kontić and Vidicki, 2018). For example, employees within an organisation with a mindset of being open or growing new knowledge or being adaptive may be more likely to embrace new cognitive technologies and support corporate entrepreneurship initiatives aimed at DT (Nadkarni and Prügl, 2020). Others equipped with higher levels of proactiveness or risk-taking may even initiate new ventures and renewal activities, reflecting an intrapreneurial mindset (Cheng et al., 2024; Glinyanova et al., 2021). In contrast, employees with a fixed mindset, may withdraw or be less inclined to engage with cognitive technologies, hampering DT efforts for an organisation. Ritala et al. (2021) examined how 166 individuals’ entrepreneurial orientation, as mindset influenced their performance in relation to a manufacturing firm’s digital strategy goals and found that proactiveness and risk-taking positively affected performance, while innovativeness did not, indicating the need to understand employee mindsets as influencing firms’ DT efforts (Solberg et al., 2020; Kontić and Vidicki, 2018). This recognition motivates the current review to explore the presence of these mindsets in the existing body of literature.

Paradoxes in DT. Paradoxes, defined as contradictory yet interrelated socio-cognitive tensions that exist and emerge as firms change, renew, and innovate (Smith and Lewis, 2011). It is vital in understanding why firms face inertia despite devoting considerable resources to leverage their dynamic capabilities (Qin, 2023; Teece et al., 2016; Volpentesta et al., 2023). For instance, firms may navigate the tension of wanting stability, yet needing to innovate with cognitive technologies (Wimelius et al., 2021). While organisations aim to embrace these technologies and adapt to changing market conditions by driving innovation through DT, doing so often requires disrupting established practices, processes, and systems. This disruption may lead to resistance among some employees (Klein et al., 2024; Smith and Beretta, 2020), highlighting the important role managerial and employee cognition and behaviour play in DT (Teece et al., 2016). Literature linking DT and dynamic capabilities to explain the foundations of socio-cognitive paradoxes assumes employees, individually or in groups, act based on their existing attitudes, values, and beliefs (Ertl et al., 2020). It is due to instances like this that the literature suggests considering paradoxes is critical for understanding DT (Danneels and Viaene, 2022; Klein et al., 2024; Soh et al., 2023; Volpentesta et al., 2023; Wimelius et al., 2021). This rationale also justifies the current review’s examination of paradoxes in the existing DT literature (Qin, 2023; Wimelius et al., 2021).

Prior DT reviews, as shown in Table 1, have emphasised organisational renewal (Hanelt et al., 2020; Verhoef et al., 2021; Vial, 2019; Volberda et al., 2021), innovation (Appio et al., 2021; Lanzolla et al., 2020), and the actors involved (Larson and DeChurch, 2020; Nadkarni and Prügl, 2020) as dominant perspectives. Such studies have aided the accumulation of knowledge in keeping with a key purpose of literature reviews focused on DT that is disparate and crosses disciplinary boundaries (Snyder, 2019).

As shown in Table 1, existing reviews have only examined the elements of micro-foundations in isolation, primarily at the firm level (Vial, 2019). This overlooks how these mechanisms, related to knowledge and learning, interact, and contribute to the practical applicability of dynamic capabilities (Felin and Foss, 2009; Vial, 2019) for leveraging DT effectively (Canhoto et al., 2021; Teece et al., 2016; Wielgos et al., 2021). Understanding how these routine-centred mechanisms induce cognitive challenges for employees is crucial for advancing DT research (Vial, 2019; Volberda et al., 2021).

When considering the first dimension of DT micro-foundations, prior studies show how knowledge search occurs when firms are motivated to look beyond their internal boundaries to acquire new knowledge for innovation (Lanzolla et al., 2020). Knowledge management, encompassing people, technology, and processes to support the acquisition, transfer, and storage of tacit and explicit sources of knowledge are crucial mechanisms underpinning innovation capability (Annarelli et al., 2021; Appio et al., 2021). Whereas Nadkarni and Prügl (2020) distinguish between technology and actors as separate entities in DT, Larson and DeChurch (2020) conceptualise cognitive technologies as both, whereby technological actors possess the potential to serve, work alongside, and even replace, human actors as tools, assistants, and collaborators (Rouse and Spohrer, 2018). These authors show how transactive memory systems as the third dimension of DT micro-foundations considered in this review play an important role in facilitating the cognitive emergent state required by employees to embrace such novel modes of interaction and identification with technological actors. While literature shows knowledge-based micro-foundations can be shared, augmented, or replaced by cognitive technologies (Faraj et al., 2018), this remains unaddressed by extant reviews as shown in Table 1. A gap therefore remains in the current state of the field of DT and associate reviews on this topic shedding critical insight into the role of knowledge search, knowledge management, and transactive memory systems individually, and together as mutually overlapping knowledge-based micro-foundations, in renewing the firm’s innovation and value-creating capabilities.

The reviews in Table 1 acknowledge the importance of employees’ mindsets to support the firm’s transformation (Hanelt et al., 2020; Nadkarni and Prügl, 2020; Vial, 2019), yet remain unclear on what it constitutes (e.g. how it is measured) or how it can be achieved. Firms are known to struggle with overcoming path dependencies, cognitive limitations of their members, and established attitudes toward technology (Ertl et al., 2020; Helfat and Peteraf, 2015; Vial, 2019). Yet, a question remains as to what can be gleaned from the DT literature as to the appropriate mindset best suited to resist inertial forces and enable DT to succeed (Mugge et al., 2020; Volberda et al., 2021). Our review aims to go beyond the current state of the field in DT and previous reviews on employees’ mindsets, while also encouraging future research to address remaining gaps.

Prior reviews reveal how DT has the potential to impact innovation processes in ways that are complex and causally ambiguous (Appio et al., 2021; Lanzolla et al., 2020), yet these reviews have not synthesised how socio-cognitive paradoxes manifest within the DT process. For example, in their systematic review of 171 articles Lanzolla, Pesce and Tucci (2020) focused on the mechanisms underlying DT that enable different types of knowledge search and recombination and the innovation outcomes they produce. They propose a taxonomy of innovation models that relate to knowledge-based tensions that emerge at the intersection of digitalization and innovation activities such as reinforcing versus overturning existing knowledge structures, substituting versus complementing existing competences, and increasing versus decreasing cognitive and emotional costs. These reviews identify a research gap in explaining how emerging technologies can sustain and change the foundations of organisational learning and the capabilities required to adapt in innovative ways (Appio et al., 2021; Lanzolla et al., 2020). At this intersection, we identify additional insight beyond those offered by current DT reviews, drawn from existing studies on the socio-cognitive paradoxes firms face at multiple levels as they sense and seize future opportunities, and ultimately, transform.

In sum, Table 1 demonstrates that a further review of the fragmented, multi-disciplinary DT literature is warranted, thus motivating the current review, in line with the recommendations of Paul et al. (2021). This conceptual review consolidates and synthesises existing knowledge to provide state-of-the-art understanding of the DT phenomenon, identify extant knowledge gaps, and highlight avenues for future research, aligning with the aims of the systematic review method (Hulland and Houston, 2020; Paul and Criado, 2020). Notably, unlike previous DT reviews, this review examines the extent to which cognitive technologies, mindsets, and socio-cognitive paradoxes have been explored in the literature. By addressing all these aspects this review offers a more holistic understanding of DT and its complexities.

This review presents the findings of our systematic review by integrating prior DT research, following the guidance of seminal authors (Kraus et al., 2020; Paul et al., 2021; Tranfield et al., 2003). While bibliometric-systematic reviews (Marzi et al., 2024) have been used to map the historic evaluation of fields and provide insight into publications trends, this paper follows an integrative approach (Cronin and George, 2023), given the fragmentation in the multidisciplinary DT literature. Our methodological approach closely aligns with related studies (Acciarini et al., 2021; Hanelt et al., 2020; Jeon and Maula, 2022). As such we follow the recommended steps of 1) identifying the scope of the review using research questions (outlined in the introduction), 2) defining the search strategy using systematic criteria for the inclusion of studies (i.e. keywords and search criteria for exclusion in Section 3.1), and 3) retrieval of studies and data cleaning, shown in Figure 1 following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) framework, 4) content analysis and thematic coding of papers in the dataset, and 5) synthesising the findings to propose future research directions. Following these steps ensures our review is guided by a transparent and replicable process (Page et al., 2021; Tranfield et al., 2003).

The search string deployed to index Scopus, EbscoHost, and Proquest databases was digit* OR digit*ation AND transformation AND capabilities AND technolog*. These databases were suitable as they have been utilised by prior systematic reviews (Annarelli et al., 2021; Hanelt et al., 2020; Nadkarni and Prügl, 2020). Multiple variations and combinations of search terms were deployed to evaluate their suitability based on the research aims, resulting in the final choice of search string. Due to the inconsistency of related terms used for DT (Verhoef et al., 2021) the truncated form of digital was used to capture related terms such as “digitisation” and “digitalisation” to scan the breadth and potential relevance of available literature that could be then filtered by the researchers’ use of a consistent definition to clearly distinguish DT from these different, yet related concepts. “Capabilities” was also a key word that focused the literature stream based on the scope of this investigation related to how firms innovate and transform (Helfat and Peteraf, 2015; Teece et al., 1997). Furthermore, the truncated form of “technology” was deployed to capture variations such as “technologies” and “technological” based on the context in which DT is triggered by emerging technologies. The search was filtered in each database for peer-reviewed articles from scholarly journals in English only. No date constraints were applied to capture the full breadth of literature available and show the scholarly evolution when filtered by our selection criteria.

Selection criteria (inclusion and exclusion) were applied to limit the articles of interest based on the theoretical framing and objectives of this review. Inclusion criteria related to articles addressing DT (as defined in this review) and articles at the firm level of analysis, excluding those that focus on platforms and marketplaces, supply chains, and broader political/economic eco-systems, or government/financial systems that operate at the macro level. Exclusion criteria were articles that 1) focus on social media or devices in the workplace for activities such as social media marketing, ergonomics, or e-learning, as these are more marketing and training related; 2) articles that narrowly focused on a specific industry with bounded implications that are non-generalisable or relate with public sector, government organizations; 3) articles focused on start-ups or “born digital” firms producing digital products (e.g. software), or operating in a primarily digital sector as these firms are unlikely to pursue DT, as they were established as digital since their inception, and 4) articles with Scimago Journal Rankings (SJR) lower than the second quartile as an indicator of quality, to ensure quality and rigour in the sample (Xiao and Watson, 2019), consistent with prior reviews in this field (Good et al., 2019).

The dataset includes articles published from 2017 to 2022. The initial search was conducted in August 2021. Prior to the implementation of inclusion and exclusion criteria, the search yielded 719 search results. A database of 444 unique articles was attained after 275 duplicate references were removed. Applying the selection criteria to the titles and abstracts subsequently resulted in removal of 299 articles. Full text articles were sought for the remaining 145 records, resulting in 112 articles removed based on criteria. Only one full text paper was unable to be retrieved. 8 articles were added in April 2022 to complete the year of 2021 and 43 articles published in 2022, were added in July 2023. The decision to include articles until the end of 2022 is appropriate as the proliferation of research related to large language models (generative AI) since the launch of ChatGPT in November 2022 is likely to produce a new stream of articles to be captured in future studies. Following the PRISMA steps prescribed by Page et al. (2021), a final sample of 83 articles were included for analysis as shown in Figure 1.

A descriptive analysis of the corpus of papers included in this review is presented in Section 4, while Section 5 outlines the content and thematic analysis to address the three research questions. Consistent with other systematic literature reviews (Riedel et al., 2022; Whittaker et al., 2023), we employed various coding methods, including directed content analysis and summative content analysis (Hsieh and Shannon, 2005) and three-order thematic analysis, as recommended by Magnani and Gioia (2023) to address the research questions, given the type of data and insights that could be drawn from the reviewed papers, as shown in Table 2.

To clarify the extent to which knowledge search, knowledge management, and transactive memory systems were addressed as DT micro-foundations in the literature, (research question 1) directed content analysis was used based on prior research (Hsieh and Shannon, 2005). Directed content analysis was deemed appropriate because it combines qualitative and quantitative approaches and enables coding to stay true to the literature in the process of relating articles to conceptually congruent categories ( Appendix 1, Table A1). Next, to quantify and summarise the extent to which the reviewed literature explores the employee mindset required for DT (research question 2), summative content analysis was used ( Appendix 2, Table A2). This analytical method was deemed appropriate, as it enabled the concrete categorization of articles in our dataset to concrete definitions and dimensions, describing certain attributes, rather than merely identifying the importance of these concepts (Finfgeld-Connett, 2014). Finally, to address the third research question, the literature was synthesised by identifying the role of socio-cognitive paradoxes that manifest during DT. Since this is a novel theoretical approach to the fragmented DT literature, a rigorous three-order coding process (Magnani and Gioia, 2023) was followed to achieve theoretical parsimony and address the fragmentation. First-order concepts were coded from across the data corpus and combined to create second-order themes and aggregated to reveal paradoxes that simultaneously persist during DT as recommended by Gioia et al. (2012), by drawing upon dynamic capabilities’ higher-order dimensions of sensing, seizing, and reconfiguring (Teece et al., 1997), hence an abductive process was followed (Rinehart, 2021).

The analysis reveals an increase in the number of articles published since 2019 as shown in Figure 2. These articles are distributed among 55 academic journals from primarily strategic management, information systems, and entrepreneurship perspectives with 12 journals publishing more than one article (Figure 3). This reflects the multidisciplinary interest in DT research and underscores the need to consolidate a growing body of literature prone to fragmentation (Snyder, 2019; Tranfield et al., 2003).

The findings pertaining to the three research aims: knowledge-based DT micro-foundations, employee mindset, socio-cognitive tensions are subsequently presented.

Addressing the first research question of determining the extent to cognitive technologies relate to the knowledge-based micro-foundations of DT capabilities, 31 of the reviewed articles show the extent to which knowledge search, knowledge management, and transactive memory systems are reflected in the literature corpus as DT micro-foundations, including how cognitive technologies are related to these themes. Only two conceptual articles relate these DT micro-foundations to cognitive technologies demonstrating the current DT research has not kept pace with emergent perspectives to show how human and technological actors are related to these DT processes. These two articles highlight the use of sophisticated algorithms and AI capabilities that leverage and learn from data, and the features of individual and social workplace technologies that also anticipate employee and organisational needs, connect people with knowledge, and in some cases perform management functions (Baptista et al., 2020; Elia and Margherita, 2021). Despite the dearth of articles examining cognitive technologies, Figure 4 demonstrates the inter-related nature of knowledge search, knowledge management, and transactive management systems related to DT.

The knowledge search theme was evident in 14 articles, reflecting that search practices occur when firms are motivated to look beyond their internal boundaries to acquire and generate new knowledge (Ghosh et al., 2022). Firms can leverage emerging technologies to act proactively and prospectively toward new opportunities (Arcidiacono et al., 2022). For example, a quantitative study of 210 managers of Italian firms by Tortora et al. (2021) found acquiring and generating knowledge combined with market sensing capabilities enables digital innovation, which Matarazzo et al. (2021) found is enhanced by using digital technologies. The articles included in this review highlight how firms differ in searching for new knowledge along near and distant horizons. For example, big data analytics capabilities enable and enhance sensing and opportunity awareness close to organisational boundaries based on the existing business (Arias-Pérez et al., 2021a; Arias-Pérez et al., 2021b; Chirumalla, 2021; Troilo et al., 2017). In contrast, open innovation and external networks are identified as an important means of sensing new opportunities and generating new knowledge beyond the boundaries of the existing firm (Arias-Pérez et al., 2021a; Arias-Pérez et al., 2021b; Steiber and Alänge, 2021).

Knowledge management is addressed in 25 articles. Underlying this theme is the conceptualisation of the firm as a cognitive unit which involves new knowledge creation and absorption processes from different sources (Ambos and Tatarinov 2022; de Paiva Britto et al., 2019). However, human cognitive limitations constrain activities that involve knowledge, which Cetindamar and Abedin (2020) demonstrated how employees’ perceptions and values determine their behaviour related to technology. Li et al. (2021) found digital technologies determine the way firms collect, store, analyse, and disseminate information, shaping the way communication and collaboration occurs among individuals and groups. Cloud-based technologies (Cetindamar and Abedin, 2020) and big data analytics are indicated as key enablers of knowledge management processes (Arias-Pérez et al., 2021a; Troilo et al., 2017), however, leveraging the transformative value of these technologies is also moderated by individual and group-level mindset factors (Tortorella et al., 2020; Troilo et al., 2017). More recently, Chatterjee et al. (2022) demonstrated how developing the competencies of employees to leverage cognitive technologies is enhanced by management support to achieve competitive advantages through technology adoption. In these articles, employees are the agents of DT and therefore the absorptive capacity of the firm relies primarily on human actors for knowledge management (de Paiva Britto et al., 2019; Demeter et al., 2021; Rocha et al., 2021; Tortora et al., 2021).

Transactive memory systems were identified in 15 articles. As this type of knowledge is typically tacit, it is more likely to result in unique, non-replicable, rare knowledge, provided the firm has high knowledge-sharing capabilities (Cetindamar and Abedin, 2020). In their quantitative study, Li et al. (2021) found DT processes suffer when relying on codified forms of knowledge. In contrast, socialisation knowledge, processed through conversation and interaction, directly influences the firms’ proactiveness, and ultimately, DT. Interestingly, Nasution et al. (2020) found a stark knowledge gap between top-level managers and employees at lower levels of firm digital maturity suggesting these firms should take advantage of existing employees’ knowledge to motivate other employees in transforming their firm. In their investigation of Medtech incumbents, Pundziene et al. (2022) found technology platforms facilitate dynamic capabilities through fostering connectedness, breaking down silos, and transforming organisational boundaries.

Employee mindsets play a crucial role in enabling successful DT, with more than half of the reviewed articles (n = 45) emphasizing this factor (see  Appendix 2, Table A2). Table 3 indicates the literature conceptualizes mindsets in three distinct ways: digital, entrepreneurial, and risk-taking. A digital mindset enables value creation through technology adaptation and business model reconfiguration, with employees acting as workplace technology consumers who leverage their technological knowledge for DT (Ahmed et al., 2022). This mindset, reflecting self-efficacy and personal innovativeness, helps breakdown knowledge silos and enhance collaboration across organizational boundaries (Ghosh et al., 2022; Liu et al., 2021; Rocha et al., 2021). The entrepreneurial mindset, focused internally on the firm’s strategic objectives, emphasizes embracing uncertainty and fostering innovativeness and adaptability (Baptista et al., 2020; Colli et al., 2021; Thayla Tavares et al., 2020; Warner and Wäger, 2019), while the risk-taking mindset balances innovation within organizational constraints to promote organizational agility (Ciampi et al., 2022). These cognitive characteristics collectively enable DT through employees’ technology directed entrepreneurial behaviour within firms.

Socio-cognitive paradoxes arise from tensions in the DT processes and address the third research question shedding light on the conflicting, yet simultaneous, interrelated elements that focus the challenges firms face when striving to resolve differences between the demands of DT and the status quo (Eisenhardt, 2000; Smith and Lewis, 2011; Teece et al., 2016; Volberda et al., 2021). These paradoxes are enabled or impeded by employees’ mindsets, with all articles reviewed (n = 83), revealing three aggregate dimensions. These dimensions are a sensing paradox of renewal drivers, a seizing paradox of innovation posture, and a reconfiguring paradox of structural malleability reflecting these polarities as shown in Figure 5. The inertial forces that reside within these tensions contribute to our understanding of the complexity firms engaged in DT face and provide the basis for future research to advance knowledge.

Our analysis reveals a sensing paradox of renewal drivers, which highlights the simultaneous combination of motivations and inhibitions firms face as they look beyond their internal boundaries for new sources of knowledge (Ellström et al., 2022). DT is an existential problem for firms triggered by environmental changes that represents game-changing opportunities on one hand and existential threats on the other (Cui et al., 2021; Fachrunnisa et al., 2020; Matarazzo et al., 2021; Naimi-Sadigh et al., 2021). Despite the importance of scanning the environment for new opportunities (Liu et al., 2021), the literature also shows firms need to overcome inhibitions related to rapid changes in emerging technologies (Saarikko et al., 2020; Tavoletti et al., 2021), perceived costs and high failure rate (Thayla Tavares et al., 2020), and confusion about what is required to digitally transform (Ghosh et al., 2022). Customers’ behaviour related to technological adoption and changing expectations also motivates firms to change (Garbellano and Da Veiga, 2019; North et al., 2020; Warner and Wäger, 2019). However, managers’ long-term and short-term perceptions shapes the extent to which firms are able to sense and capitalise on new opportunities (Colli et al., 2021).

The seizing paradox of innovation posture comprises tensions related to firms’ leaning toward incremental and radical knowledge to explore uncertainty and experiment contrasted to exploiting existing domains of mastery. While radical innovation is shown to be more challenging (Sund et al., 2021), incremental knowledge accumulates in value and strengthens firms’ absorptive capacity over time (Depaoli et al., 2020). Firms can take advantage of disruptions in the environment to embrace more radical innovations that would otherwise be more difficult (Al-Edenat, 2021; Gholampour Rad and Nisar, 2017). This can improve firm performance and enhance R&D capabilities (Konlechner et al., 2018), enabling firms to develop new products and services by drawing upon knowledge combinations from different technology fields (Demeter et al., 2021). Managers are key to integrating and coordinating innovation (Matarazzo et al., 2021; Naimi-Sadigh et al., 2021; Saarikko et al., 2020), but conversely also risk limiting the firm’s innovative potential by their own limited knowledge of technology and cognitive biases (Li et al., 2018). Knowledge exchange and transfer in diverse domains enhances pursuing new opportunities (Cetindamar and Abedin, 2020; Konlechner et al., 2018; Nasution et al., 2020), altering routines, and creating a sense of urgency among managers for DT (Ellström et al., 2021). Arcidiacono et al. (2022) propose managers can foster these attributes by placing their firms in a learning mode to embrace mystery (uncertainty) rather than rely on their mastery derived from prior experience.

The reconfiguring paradox of structural malleability describes the tensions firms face when the need for continuous adaptability conflicts with proven ways of working that may reinforce the status quo. The literature reveals that DT places demands on firms in terms of continuous adaptability (Ellström et al., 2022) in ways that can simultaneously challenge managers’ past experiences as an impediment (Chirumalla, 2021; Colli et al., 2021; Depaoli et al., 2020; Dwipayana et al., 2021; Ellström et al., 2022; Ghosh et al., 2022; Konlechner et al., 2018; Li et al., 2018; Oliver, 2018; Rocha et al., 2021). Past performance reinforces existing path dependencies that may not be conducive to the emergence of new opportunities (Demeter et al., 2021; Dwipayana et al., 2021; Gholampour Rad and Nisar, 2017; Saarikko et al., 2020; Warner and Wäger, 2019). This is further exacerbated by DT complexities (Naimi-Sadigh et al., 2021; Saarikko et al., 2020). Troilo et al. (2017) found data-driven cultures offer employees a new kind of power to challenge embedded mental and behavioural routines by shifting the emphasis from intuition and personal feelings to solid data analysis and modelling. It also fosters learning-oriented routines that enable ongoing innovation (Colli et al., 2021), and brings managers closer to new technologies, customers, and markets (Thayla Tavares et al., 2020). However, these changes require the removal of functional silos, openness to change, supportive culture, collaborative knowledge management, and data transparency to be effective (Ghobakhloo and Fathi, 2020). Warner and Wäger (2019) found top-level managers struggle to make radical reconfigurations when the firm’s capabilities are too tightly tied to its values, history, collective memory, routines, and emotions. Incumbents must balance their own path dependent tendencies with an adaptive mindset as organization structures become inert and weaken over time (Liu et al., 2021; Warner and Wäger, 2019). An adaptive mindset is reflected in improvements to management cognitions that lead to capability reconfigurations (Chirumalla, 2021; Gholampour Rad and Nisar, 2017; Ghosh et al., 2022; Steiber and Alänge, 2021).

This review synthesized the fragmented DT literature to address three key objectives: the role of cognitive technologies in DT, the employee mindsets required for DT, and the socio-cognitive paradoxes firms need to manage as they transform. By integrating the findings across these objectives, our review contributes to the growing scholarly interest in DT and proposes a future research agenda.

Our review clarifies how extant studies lag the emergent phenomenon of cognitive technologies, whose unique properties (e.g. autonomous learning and responsiveness) strengthen the link with traditionally human-only orchestration of knowledge search, knowledge management, and transactive memory systems as crucial mechanisms underlying DT capability. Cognitive technologies can automate and augment opportunity sensing capabilities by widening external knowledge search or deepening internal knowledge discovery and aggregation, especially in larger and more complex firms with siloed or archived knowledge. Our findings support related studies that have found knowledge management also plays a pivotal role in enabling firms’ strategic flexibility and capability to deal with the competing demands of innovation and seizing new opportunities (Chen et al., 2023; Neiroukh et al., 2024; Rialti et al., 2020). Notably, cognitive technologies’ ability to access and divulge knowledge encoded in natural language challenges the notion of transactive memory systems as an intrinsically human-only domain, especially pertinent given radical innovation capability is typically linked with tacit rather than codified knowledge (Lopes-Bento and Simeth, 2024; Zahra et al., 1999).

Based on our review, we propose a DT mindset can be defined as an integrated cognitive schema combining technological adaptability, entrepreneurial agility, and risk intelligence to drive organizational innovation and value-creation. Technological adaptability involves a deep understanding of how cognitive technologies and platforms can be leveraged to create value, coupled with the ability to diffuse technological knowledge across organizational boundaries. Entrepreneurial agility refers to the capacity to continuously evolve business practices and models while maintaining a flexible posture toward market uncertainties and technological changes. Risk intelligence means calculated risk-taking that encourages employee initiative and experimentation (including learning from temporary setbacks) in response to changing conditions in relation to firms’ strategic objectives. Subsequently, our review responds to calls for a more multi-dimensional perspective to understand how employees’ mindsets toward cognitive technologies are shaped by both their entrepreneurial behaviours and cognitive expressions of DT in the workplace (Ahmed et al., 2022). Our review also quantifies a substantial gap between the emphasis on fostering a DT mindset and the emergent realization that the locus of entrepreneurship in firms is fundamentally shifting due to cognitive technologies (Holmström, 2022; Recker et al., 2023). Reconciling this divide requires new conceptualizations to extend literature on digital corporate entrepreneurship (Nadkarni and Prügl, 2020).

Finally, our review addresses the socio-cognitive paradoxes that arise when knowledge-based mechanisms are enacted in DT. Responding to calls for research that conceptualizes firms as cognitive enterprises grounded on a system of processes defined by human–machine interactions, our findings highlight how employee mindsets comprised of behaviours, attitudes, values, and beliefs toward cognitive technologies can enable and impede DT (Elia and Margherita, 2021). By deploying a socio-cognitive perspective, our analysis draws together organizational and individual perspectives to elucidate the complex, social context of DT, incorporating aspects of the firm environment and individual behaviours and cognitions, highlighting a sensing paradox of renewal drivers, a seizing paradox of innovation posture and reconfiguring paradox of structural malleability. Our findings reflect a socially-situated context (Mitchell et al., 2011; Orlikowski and Scott, 2008), wherein entrepreneurial agency is considered among both human and technological actors (Nambisan, 2017), with far-reaching ramifications for the future of DT. Our review extends prior research in three ways related to dynamic knowledge-based micro-foundations (Chen et al., 2023), the role of human–machine collaboration (Li et al., 2023; Neiroukh et al., 2024), and firms’ strategic flexibility (Agostini et al., 2023; Montreuil, 2023; Rialti et al., 2020).

Our findings offer three managerial implications. First, as micro-foundations are useful for grounding higher-order concepts like dynamic capabilities, managers can move beyond abstract conceptualizations of cognitive technologies and AI by focusing on how specific mechanisms can address task and domain specific problems. For example, cognitive technologies not only improve existing processes, but also expand what is possible beyond traditional human constraints. Second, by automating and augmenting the ability to sense and seize opportunities, these technologies can enhance firms’ innovation potential and value creation and delivery opportunities. Managers should foster an entrepreneurial mindset across all levels of the firm and remain mindful of their own blind spots, as employees with a greater readiness to adopt cognitive technologies may perceive opportunities they do not (Chen et al., 2023). Finally, managers should sharpen their attention on the social aspects of DT, including the entrepreneurial opportunities that arise from human–machine collaboration, and configure firm environments with cognitive technologies to promote shared learning, and the attitudes employees hold of these technologies.

As the findings and synthesis of extant DT research reveal critical research gaps, we propose a three-pronged research agenda in Table 4 with related research questions to stimulate future research. This review showed DT research related to cognitive technologies is in a nascent stage, therefore proposing future research avenues is useful to steer research that is theoretically significant and practically relevant for firms.

First, DT studies can advance research by addressing the sensing paradox of renewal drivers. Our synthesis of existing DT literature demonstrates empirical studies are lacking to show how cognitive technologies can augment and automate the way firms look beyond their boundaries for new sources of knowledge and opportunities. Firms might be motivated to embrace cognitive technologies to leverage their potential for renewal through enhanced knowledge search and knowledge management, and yet fail to realize their ambitions for transformation due to cognitive limitations of employees at different firm levels. The ability for cognitive technologies to also augment tacit knowledge as a conduit among employees holds novel contributions for existing firms. Socio-cognitive theory (Bandura, 1989) provides a suitable basis for empirical studies to probe the emergent phenomenon of human and technological actors in DT.

Second, limited empirical DT studies examine the seizing paradox of innovation posture. Considering innovation posture tensions, we propose research questions that examine the implications of cognitive technologies. Cognitive fit (Vessey and Galletta, 1991) provides a promising theory to investigate the inertial properties of employees’ motivating and inhibiting mindsets that enable or impede their firm’s DT processes, especially in the context of radical and incremental approaches to innovation. For example, person-job fit, person-organization fit, and person-supervisor fit have been studied as important interactions in management studies to explore the consequences of individuals’ fit at work (Kristof-Brown et al., 2005). Karran et al. (2022) also expounded on the fit between AI systems and the way their agency (i.e. decision-making) is represented to human actors, thereby demonstrating how trust mediates how much employees are willing to rely on cognitive technologies. However, a potential blind spot to resolve discrepancies in fit that our research agenda addresses relate to a socio-cognitive perspective, which has been overlooked (Li et al., 2023; Montreuil, 2023).

Finally, this review identifies the need for future research that empirically investigates the reconfiguring paradox of structural malleability. The DT literature is yet to investigate how structural relations within existing firms are impacted by cognitive technologies that emulate and augment human cognition. Our proposed research agenda suggests how power is distributed among employees and the extent to which existing structures are flexible to adapt to the emergent role of cognitive technologies in the workplace is important. By deploying a paradox lens as recommended in the case of complex phenomenon like DT (Smith and Lewis, 2011), future research can shed light on the cognitive aspects of human and technological actors in DT. We thus urge scholars to move beyond conceptual or exploratory studies that have only begun to consider the implications of cognitive technologies on the ability of firms to overcome path dependencies and continuously evolve.

All research has limitations and so too this integrative review. First, it is important to acknowledge the DT landscape is ever changing and thus the need for consolidation and accumulation of knowledge will continue. Our review is based on a systematically conducted search and analysis guided by criteria that can be replicated. It is therefore possible that other studies involving different literature may glean additional or varying insights. Given our goal is to stimulate further research based on evidence from synthesizing the DT literature, we encourage other researchers to extend and refine our findings by drawing on different sources or analytical approaches. Second, the use of qualitative methods is subject to subjectivity in the way articles have been integrated to derive the themes used in this study. As a research team we met regularly to discuss the findings presented and refine theoretical interpretations to mitigate these concerns. Third, this review acknowledges the abstract nature of dynamic capabilities theory that has historically generated critique among scholars (Barreto, 2009; Helfat and Peteraf, 2009; Kuuluvainen and Hussler, 2012). We addressed these concerns by emphasising the importance of DT micro-foundations to ground the theoretical dimensions of dynamic capabilities in mechanisms and routines that are observable (Eisenhardt and Martin, 2000). As cognitive technologies are rooted in knowledge, researchers can further probe how cognitive technologies relate to knowledge search, knowledge management, and transactive memory systems as DT micro-foundations. Fourth, we also acknowledge that studies examining generative AI, will have surged since 2023 given the dramatic rise in interest fuelled by technological breakthroughs of large language models (LLMs). We believe this strengthens the choice to include literature up until the end of 2022 to accumulate and synthesise the knowledge to this point and draw from it important inferences to guide further research that might otherwise be overlooked.

Despite these limitations, the current review aims to advance the understanding of the fragmented multidisciplinary DT literature. While existing literature offers valuable insights, there is a need for clarification to fully leverage its contribution to emerging research issues. For scholars exploring the ramifications of rapidly evolving technological landscapes, particularly spurred by cognitive technologies, our study offers a holistic and synthesized overview of the current knowledge base. Furthermore, it identifies opportunities that warrant future research avenues, especially in areas where empirical studies are still in their nascent stages.