Developing context-transcendent information systems for the future enterprise: Theory and insights from adaptive digital transformation

IS in the enterprise: The embedding premise and emerging challenges

The nature of the contemporary enterprise is undergoing a profound transformation as digital technologies reshape the sociotechnical environments in which organizations operate (Grover and Lyytinen, 2023; Ignatiadis & Nandhakumar, 2007). Rather than merely automating existing processes, these technologies engender new sociotechnical configurations that transform how value is created, which actors participate in that process, and the organizational arrangements through which it is delivered (Hinings et al., 2018; Packard and Clark, 2020). This reconfiguration renders the IS application context a consequential theoretical object: not a neutral backdrop to organizational action, but an open and evolving configuration of value imperatives, relational arrangements, and dynamic pressures that jointly condition what IS can meaningfully provide within a given business situation (Orlikowski and Iacono, 2001; Tan et al., 2020).

The relationship between IS and the contexts in which they operate has evolved significantly over the last two decades. Early IS research conceptualized IS in the enterprise as context-independent tools: rational instruments evaluated primarily on their technical functionality and information-processing efficiency, with context treated as an external condition to be controlled rather than a constitutive force to be theorized in IS use (Markus and Robey, 1988; Orlikowski and Iacono, 2001). IS value was accordingly understood as intrinsic to the technical features of the system itself, that is, a well-designed and satisfactorily implemented system would deliver predictable returns regardless of the specific users, organizational situations, or value imperatives it encountered (DeLone and McLean, 2003; Goodhue and Thompson, 1995).

A decisive theoretical shift occurred from the late 2000s onward as IS research moved away from treating systems as context-independent utilities toward recognizing context as the medium through which IS value is defined and realized. The foundational insight is that IS properties are not fixed in initial design but emerge through recurrent, situated interactions among human actors, technological artifacts, and institutional arrangements, making what IS does and what value it generates irreducibly context-dependent (Orlikowski, 2000; Volkoff and Strong, 2013). At the enterprise level, this embedding premise carries three implications for theorizing IS in the enterprise. First, the IS value is defined and redefined iteratively within business contexts rather than predefined outside them, as organizational sensemaking rooted in core business contexts determines which IS capabilities become meaningful and which remain latent (Mueller et al., 2025; Tan et al., 2020). Second, embedding is generative rather than neutral: once IS becomes entangled with local arrangements, it privileges certain value creation paths over others, generating path dependencies that make subsequent reconfiguration increasingly difficult (Rolland et al., 2018). Third, IS-context fit is inherently temporal: as business contexts evolve, IS deployments that cannot accommodate ongoing reconfiguration become progressively misaligned, which incurs additional coordination efforts for IS-business alignment (Aral et al., 2024).

Grounded in the embedding premise that IS creates value through a deep alignment with the business contexts it serves, enterprise IS research has moved beyond static system views toward a more dynamic understanding of IS. This premise has been highly productive because it explains why IS value is not inscribed in design alone, but constituted through ongoing alignment between IS and business contexts. Complex adaptive systems (CASs) research, exemplified by Nan’s model of bottom-up IT use (Nan, 2011), extends this view by showing that IS features are emergent rather than linearly determined, arising from recurrent interactions among agents within an environment. From this perspective, enterprise IS is dynamic, multilevel, and interaction-driven, while context primarily enters as a structure that shapes how actions and interactions unfold. Similarly, the digital infrastructure literature shows that these dynamics do not remain episodic or localized, but become sedimented in evolving sociotechnical arrangements of interdependent technologies, processes, and organizational controls (Hanseth and Lyytinen, 2010; Tilson et al., 2010). The enterprise problem is thus recast from designing fit-for-purpose systems to governing ongoing infrastructural evolution as interdependencies accumulate and become increasingly generative, interconnected, and path dependent.

This study builds on these traditions, but shifts the analytical focus to a more specific enterprise IS tension. Enterprise IS research has largely been grounded in the embedding premise that IS creates value through deep alignment with the business contexts it serves, and has accordingly moved beyond static system views toward a more dynamic understanding of the IS-context relationship. Yet in both CAS and digital infrastructure research, context is treated mainly as the environment within which enterprise IS evolves, rather than as the central problem to be theorized. Drawing on Chatterjee and Davison (2021)’s problematization approach, we revisit a more basic assumption embedded in the literature: that deeper IS embedding within a focal context is generally conducive to alignment and value creation. We argue that this premise becomes limited under conditions of contextual heterogeneity. When business contexts differ substantially in structures, demands, and operating logics, deeper embedding in one context may come at the expense of transferability, reconfigurability, and coherence across others. Contextual heterogeneity, therefore, offers a sharper point of departure because it directs attention to the conditions under which the IS-context relationship itself becomes a source of enterprise-level tension.

Our review of recent literature on IS deployment and value creation suggests that three forms of heterogeneity are especially consequential. First, structural heterogeneity arises as increasingly interconnected digital infrastructures make configurations optimized for one setting resistant to integration with others, thereby intensifying tensions between local responsiveness and enterprise-wide coherence (Hanseth and Modol, 2021). Second, institutional heterogeneity emerges when enterprises operate across distinct regulatory and normative environments, such that the same embedding logic that enables value creation in one setting may undermine legitimacy in another (Lok, 2010). Third, temporal heterogeneity becomes salient as AI-enabled IS continuously learns from real-world data and updates its parameters, progressively reshaping the very context it was introduced to serve (Benbya et al., 2024). Taken together, these conditions render embedding not merely insufficient, but increasingly self-undermining as a dominant logic of enterprise IS design and deployment.

In other words, enterprises no longer need IS that simply fits individual contexts; rather, they need IS that is capable of continuously transcending contextual boundaries to enable value creation and delivery across heterogeneous settings. We term this new paradigm context-transcendent IS and argue that realizing it requires a fundamentally different logic for coordinating IT artifacts, people, and organizational processes in the evolution of IS. DT, as an enterprise-level process through which organizations reconfigure and continuously reshape these elements in response to shifting contexts (Vial, 2019; Wessel et al., 2021), provides the most consequential setting in which this shift becomes visible. As DT makes observable the ways in which IS and context recurrently reshape one another through ongoing reconfiguration, it offers a natural empirical terrain for examining and theorizing context-transcendent IS. We turn next to the prevailing paradigms of DT.

How IS was deployed: Existing paradigms of digital transformation

DT represents a fundamental shift in how organizations engage with IS, which are strategic and involve redefining what an organization offers and how it operates (Vial, 2019). Enabled by the emerging technologies, DT unlocks new opportunities, fostering new models of stakeholder engagement and user involvement in value co-creation (Tilson et al., 2010; Yoo et al., 2010). The cumulative body of work on DT has identified two major paradigms: business-driven and IT-driven, which present distinct ways for enterprises to conceptualize and enact information systems in each context.

The business-driven perspective conceptualizes DT as a strategic and organizational transformation oriented toward redesigning business models, processes, and capabilities to enhance organizational competitiveness (Uhl and Gollenia, 2014). It follows an outside-in logic in which business needs and strategic priorities shape the development and deployment of information systems (Westerman et al., 2014), positioning DT as a vehicle for strategic renewal through managerial orchestration and deliberate planning. Within this view, IT is treated primarily as a strategic enabler that supports competitive positioning by delivering digital solutions aligned with business objectives (Morton et al., 2022). Accordingly, DT is expected to unfold through formal planning, structured change programs, and alignment mechanisms that connect digital initiatives to predefined business strategies (Hess et al., 2016). Hence, conventional wisdom positions business-driven DT as the dominant and most reliable pathway to organizational adaptiveness, grounded in the assumption that business units possess superior contextual knowledge because they are closest to market demands, customer expectations, and competitive pressures (Matt et al., 2015; Sambamurthy et al., 2003).

Business-driven DT becomes increasingly problematic under conditions of contextual heterogeneity. Its outside-in logic assumes that transformation objectives can be specified ex ante, IS can be designed around predefined plans, and outcomes can unfold in a predictable, sequential manner. This assumption breaks down where heterogeneity is most pronounced. Rigid planning cycles cannot accommodate the divergent value imperatives of different business units (Tallon and Pinsonneault, 2011), predefined alignment mechanisms struggle to keep pace with shifting institutional demands across markets (Boh et al., 2023; Liang et al., 2017), and business leaders often lack the expertise to recognize and evaluate the technological affordances that heterogeneous contexts render salient (Hsu et al., 2018). As AI systems become increasingly opaque across such conditions, this capability gap widens further, making formal alignment between business ambitions and IS capabilities progressively more fragile (Berente et al., 2021; Lebovitz et al., 2022).

IT-driven DT offers a contrasting logic in which digital capabilities, rather than predefined business objectives, become the primary force reshaping value propositions, business models, and organizational arrangements (Besson and Rowe, 2012; El Sawy and Pavlou, 2008). This inside-out orientation foregrounds the generativity of digital infrastructure and the ongoing recombination of capabilities, enabling IS to be reconfigured rapidly across contexts and opening innovation trajectories that business-driven planning could neither foresee nor direct (Bharadwaj et al., 2013a; Mikalef et al., 2020). By loosening the grip of locally dominant business preferences, IT-driven DT enables organizations to identify overlooked opportunities, experiment at scale, and pursue more radical forms of innovation under dynamic conditions (Ismail et al., 2017; Tan et al., 2009). Its distinctive strength amid contextual heterogeneity lies precisely in this cross-context generativity: rather than optimizing IS for a single setting, it leverages generative digital infrastructure to create value across heterogeneous contexts.

Yet IT-driven DT has its own structural limit under contextual heterogeneity. It’s inside-out logic, though generative across contexts, shifts the burden of interpretation and absorption onto business units, making outcomes difficult to predict ex ante (Bharadwaj and Dong, 2013; Kaganer et al., 2023). This challenge becomes even more pronounced when reusable digital capabilities extend to ecosystem partners, where differences in priorities, interpretive frames, and complementary resources make cross-context translation even less certain (Benbya et al., 2020). The result is a central paradox: the same generativity that enables rapid recombination across contexts can, without sufficient coordination and architectural ordering, produce fragmented architectures and isolated solutions that fail to consolidate into business value (Hanelt et al., 2021).

Our review of existing DT paradigms reveals a shared theoretical limitation: both reproduce the embedding premise in different forms. Business-driven DT subordinates IS to a strategically defined context, whereas IT-driven DT subordinates context to technologically defined possibilities. In both, the relationship among people, IT artifacts, and context remains one of one-sided conformity rather than mutual constitution, rendering each paradigm structurally inadequate in the face of contextual heterogeneity. Business-driven DT fragments the shared digital capabilities required for cross-context coordination; IT-driven DT generates capabilities that business units often cannot interpret, govern, or translate into locally meaningful value. Neither resolves these tensions, and their coexistence without a coordinating logic often reproduces them. What is required instead is a paradigm in which people, IT artifacts, and organizational processes are continuously reconfigured across heterogeneous contexts through the dynamic orchestration of business-driven responsiveness and IT-driven generativity. We term this adaptive digital transformation (adaptive DT). This construct differs from adaptive structuration theory (DeSanctis and Poole, 1994), which explains technology-practice integration within a single organizational setting. Adaptive DT, by contrast, concerns enterprise-level IS governance across heterogeneous organizational and ecosystem boundaries.

To explore adaptive DT as a new paradigm, a theoretical perspective is needed to illuminate how it unfolds in practice. Because our interest in DT lies in the organizing structures and implementation processes surrounding IT-based activities, an IT governance perspective provides a particularly valuable lens as it focuses on how organizations design, coordinate, and oversee IT-related decisions and actions to ensure that smooth value creation from IT activities (Tiwana et al., 2014; Wu et al., 2015). Given that the central challenge of adaptive DT concerns the governance of direction, integration, and alignment, we adopt this lens as a “sensitizing device” (Weick, 2007: p. 16) to guide an inquiry into how adaptive DT processes unfold under contextual heterogeneity.

Theoretical foundation: IT governance theory

IT governance (ITG) refers to the organizational arrangements through which enterprises direct IS deployment toward strategic objectives, assign accountability for technology-related decisions, and realize business value through cost efficiency, asset utilization, and organizational flexibility (Tiwana et al., 2014; Weill and Ross, 2004). ITG is therefore not a fixed administrative structure, but a contingent organizational logic: effective governance depends on aligning decision rights and accountability structures with strategic priorities while continuously adapting them to technological complexity and shifting organizational conditions (Weill and Ross, 2004; Xue et al., 2008, 2021). To map this broad lens, this study draws on (Gregory et al., 2018) three-dimensional framework, which specifies: (1) the focus of governance: the governance objects, or the IT artifacts and activities that require oversight; (2) the scope of governance: the actors and organizational units involved in exercising decision rights and responsibilities; and (3) the approaches to governance: the mechanisms through which IT-related decisions are coordinated and enacted.

Over the past three decades, the focal concerns of IT governance have undergone a substantive evolution. In the 1990s and early 2000s, research and practice centered primarily on IT investment decisions, emphasizing how organizations could justify, allocate, and measure the returns of technology expenditures (Sambamurthy and Zmud, 1999). During the 2000s through the mid-2010s, attention shifted toward IT infrastructure and IT-business relationships, reflecting the growing importance of enterprise-wide architectures, integration capabilities, and alignment processes needed to support increasingly complex business operations. Since roughly 2015, the rise of pervasive digital technologies, cloud services, analytics, and platforms has redirected focus toward decentralized IT use and platform governance, highlighting distributed innovation, local experimentation, and the orchestration of digital resources in multi-actor environments (Tiwana and Kim, 2015). This trajectory illustrates how IT evolved from a budgeted organizational resource to a strategic asset that shapes how enterprises operate and compete.

IT governance archetypes vary along a centralization–decentralization spectrum, each allocating decision rights in distinct ways. At the centralized end, a business monarchy vests authority in senior business executives, often through executive committees that include, but are not dominated by, the CIO, thereby supporting strong strategic alignment and value realization in transformation contexts (Gerth and Peppard, 2016; Lowry et al., 2025). An IT monarchy, by contrast, concentrates decision rights within IT leadership, leveraging technical expertise to coordinate architecture and integration, though it risks becoming detached from enterprise strategy when technology trajectories diverge (Lorenz and Buchwald, 2023). Between these poles, federal configurations and IT duopolies distribute decision authority across corporate executives, business units, and IT leaders, balancing enterprise-wide standardization with local responsiveness; however, their collaborative structures are difficult to operationalize due to conflicting incentives and the need for robust coordination mechanisms (Brown and Grant, 2005). Toward the decentralized end, feudal governance grants local units’ autonomy over IT decisions, supporting responsiveness but often undermining enterprise coherence and cross-unit synergies. At the extreme, anarchy disperses decision rights to individual actors, enabling rapid action but sacrificing coordination, and is generally tolerated only in exceptional circumstances requiring ultra-fast responsiveness (Salameh and Bass, 2022).

The approaches to IT governance, in turn, are underpinned by a coordinated system of structural, processual, and relational mechanisms (Wu et al., 2015). These mechanisms collectively shape how IT decisions are made, enacted, and enforced across the organization, facilitating the deployment of decision-making structure, alignment processes, and communication approaches (Gregory et al., 2018; Weill and Ross, 2004). Structural mechanisms establish the formal allocation of responsibilities, roles, and decision rights, and are typically implemented through governance bodies such as IT steering committees, architecture councils, and digital investment boards. It therefore defines who has the authority to make decisions (Gregory et al., 2018) and provide institutional support for accountability and oversight (Weill and Ross, 2004). Based on the chosen governance structure, processual mechanisms codify the routines and procedures that underpin IT governance activities, including planning cycles, portfolio evaluation, and project tracking frameworks, to align the effective management of IT with enterprise-wide objectives (Weill and Ross, 2004). These mechanisms operationalize governance by translating strategic intent into daily actions (Weill and Ross, 2004), ensuring that decisions are executed consistently and monitored for performance (Huang et al., 2017). They also establish feedback loops that assess alignment and channel insights back into the strategic process, supporting ongoing adjustments and ensuring that governance remains responsive (Weill and Ross, 2004). Relational mechanisms foster effective two-way communication and active participation, and collaboration between business and IT stakeholders (Wu et al., 2015). These mechanisms enhance IT governance effectiveness by enabling management to proactively and formally communicate the purpose, functioning, and expected outcomes of governance arrangements (Weill and Ross, 2004). In doing so, relational mechanisms cultivate trust, shared understanding, and commitment between business and IT leaders, thereby ensuring social alignment and strengthening business-IT relationships (Schlosser et al., 2015).

While each mechanism provides distinct contributions, their effectiveness ultimately depends on how well they operate in an integrated and complementary manner (Buchwald et al., 2014). Such coordinated integration within a coherent governance configuration (Tiwana and Konsynski, 2010; Turel and Bart, 2014), forms a critical antecedent shaping the success of digital transformation initiatives. Therefore, adopting an IT governance lens enables an examination of the strategic direction, control, and flexibility required for enacting DT (Mulyana et al., 2021), and thus guiding this study’s investigation into how existing DT approaches can be reconfigured innovatively to achieve adaptive DT and enable the development of the context-transcendent IS for future enterprises to grow and thrive under contextual heterogeneity.

Data collection

The data collection process unfolded in two phases. The initial preparation phase, from July to August 2024, involved gathering secondary data to prepare the research team for fieldwork. This phase included collecting organizational documents from Geely’s official website, industry reports, and news articles (Gioia et al., 2013). The secondary data provided essential background information and informed the development of our interview protocol and questionnaire (Eisenhardt, 1989; Pan and Tan, 2011).

The primary data collection took place in September 2024 at Geely’s headquarters in China, involving a team of five researchers. The involvement of multiple researchers enabled the triangulation of observations and interpretations (Klein and Myers, 1999). This phase employed semi-structured interviews with informants from business units of the Digitalization Center and Digital Service Sharing Center, who have led and experienced Geely’s DT. Informants were identified through chain referral sampling based on a preliminary set of interview questions sent to a designated gatekeeper (Pan and Tan, 2011). This sampling strategy was appropriate given that the research team had no prior access to internal information to identify suitable informants directly.

Following the interview protocol developed in the preparatory phase (see Appendix A for the sample protocol), we asked standardized core questions about Geely’s digital transformation strategies, evaluations of these strategies, and the challenges encountered in the adaptive DT process. Additional questions were tailored to each informant’s role. For instance, senior management informants were asked about the formation of adaptive DT strategies, while first-line managers were interviewed about the execution and implementation of the ITG mechanisms. On average, interviews took around 60 minutes and were audio-recorded and transcribed for analysis and concept framing (Pan and Tan, 2011).

After data collection, each research team member independently analyzed the data and exchanged notes to ensure consistency in interpretation (Klein and Myers, 1999). In total, 21 informants were interviewed. Among them, 10 were senior management executives responsible for formulating and governing DT strategies, while 11 were first-line managers responsible for strategy execution and implementing governance mechanisms within their respective business units. Appendix B provides a detailed list of the informants.

Data analysis

Since our research aims to understand how adaptive DT is achieved, we employed narrative and visual mapping strategies as an initial step in analyzing the raw data (Langley, 1999). The narrative strategy involved constructing a story that provides contextual details of the adaptive DT phenomenon using “thick description” (Langley, 1999: p. 695). Meanwhile, the visual mapping strategy involved creating chronological event timelines and diagrammatic sketches to capture and document our emergent theoretical ideas. These initial sketches facilitated the organization of rich interview data by illustrating relationships between concepts at a higher level of abstraction, thereby enhancing the validity of our data interpretation and laying the foundation for theory development (Klein and Myers, 1999). After constructing the narrative and visual maps, we verified them with relevant informants and revised them as needed based on additional data. This process ensured the validity of both our interpretations of the informants’ accounts and the emerging theory (Pan and Tan, 2011).

Based on our initial process mapping, we adopted grounded theory techniques to code our data using open, axial, and selective coding (Strauss and Corbin, 1998). In the open coding, we focused on identifying and assigning descriptive labels derived directly from the data to form first-order concepts (Gioia et al., 2013). We delineated these labels from our data excerpts and conducted comparisons at the conceptual level (Corbin and Strauss, 2014). We sought to preserve the informants’ terms to ensure a truthful interpretation of the labels. Therefore, some labels were directly taken from the informant’s words, including “massive IT investments of 1000 projects per year across the whole corporation,” “anchor IT projects on driving business value creation,” and “business value network.” When a new data segment was conceptually similar to a previously coded label, we retained the same concept label to maintain consistency (Corbin and Strauss, 2014). As a result, our opening coding identified a total of 40 first-order concepts.

Axial coding was then used to relate the first-order concepts to a number of second-order themes (Gioia et al., 2013), we first applied axial coding to establish relationships between concepts (Strauss and Corbin, 1998), allowing us to describe and explain conditions, actions, interactions, and consequences of our investigated phenomenon (Strauss and Corbin, 1998). For example, concepts such as “massive IT investments of 1000 projects per year across the entire corporation” and “disconnected resource allocation between business and IT” were coded as “contextual pressures” because they capture the scale, fragmentation, and coordination demands that pressure IT governance and alignment within DT. In this process, we placed particular emphasis on concepts emerging from our data that were underdeveloped in existing literature or existing concepts that stood out beyond their original studied context (Gioia et al., 2013). For example, the second-order theme of “value anchoring” was derived from concepts such as “align business opportunities with IT capabilities for potential value realization” and “anchor IT projects on driving business value creation.” This theme reflects Geely’s approach to managing digital transformation through an IT-polycentric governance logic, requiring a joint force between IT and business to define how value is conceptualized and to determine the potential value contribution of IT initiatives and projects, a gap that remains underexplored in the existing DT literature. As a result of axial coding, we identified a total of 10 second-order themes.

Finally, through selective coding, we further examined the data to integrate and refine our first-order concepts and second-order themes. When data challenged previously identified concepts or themes, we revised the coding schema by adding, removing, or modifying codes as needed. In cases of disagreement within the research team, we revisited the data and resolved inconsistencies by refining theme names, discarding low-agreement themes, or breaking down complex themes until consensus was reached (Gioia et al., 2013). This stage distilled the findings into four aggregate dimensions (Gioia et al., 2013), extending our findings beyond descriptive accounts of the phenomenon and developing the theoretical conceptualization of adaptive DT.

As part of this process, we also iteratively compared our emerging concepts and themes with relevant literature to distinguish existing knowledge from novel insights (Corbin and Strauss, 2014). This iterative refinement shaped our coding structure until the theoretical saturation state was reached, where our inductively derived theory provided a comprehensive explanation of the investigated phenomenon, and further incremental learning became minimal due to established precedents in the existing literature (Eisenhardt, 1989; Gioia et al., 2013). These inductive iterations between theory and data gradually developed our theoretical framework, which is ultimately structured in our coding schema, as presented in Appendix C.

Triggers: Contextual pressures and opportunities from growing heterogeneity

Geely’s adaptive DT was precipitated by an escalating structural tension between its functionality-oriented enterprise architecture and growing contextual heterogeneity. From 2002 to 2015, rapid multi-brand expansion led to growing divergence among business units in operating logic, data standards, and value-creation processes. In response, Geely deployed standardization-oriented enterprise systems (e.g., mySAP ERP and SCADA) to enforce common processes and unified data structures. Although these systems stabilized early-stage operations, they institutionalized a process-centric architecture that became progressively misaligned with differentiated and evolving business requirements.

As product portfolio complexity deepened after 2015, this misalignment became acute. Despite sustained IT investment and an internal IT workforce exceeding 1400 employees, Geely remained unable to accommodate the scale and heterogeneity of business demands. The underlying constraint was structural: early governance arrangements premised on a vendor–client logic treated business needs as discrete, stable requests to be fulfilled by IT. As demands became increasingly context-specific, this logic fragmented coordination and reinforced an interpretive gap between IT and business departments, producing a vicious cycle of fragmented requests, an overloaded project pipeline, and progressively diminishing effectiveness in value creation. As the Chief Digitalization Director observed:

“We were handling nearly 1000 projects per year in the automotive group alone, and despite having a 1000-person IT team, the business pressure was immense… Business doesn’t understand what IT can offer; the IT group doesn’t even understand business value…They have their own understanding, and they can always propose a functional requirement, but is it the most suitable one? Not necessarily, because he does not have a deep understanding of IT.”

These conditions constituted what we term contextual pressures, that is, the structural demands generated by heterogeneous and proliferating contexts that progressively erode the effectiveness of context-bound IT arrangements. As business units proliferated and operating contexts diverged, process variants multiplied, such that the marginal contribution of process-specific IT solutions to enterprise-level value creation declined.

Yet the same heterogeneity revealed a countervailing dynamic. Operating across an extensive multi-brand portfolio, even marginal improvements in shared enterprise-level capabilities could generate disproportionately large gains when diffused across the entire organization for contextual opportunities: the generative potential latent in heterogeneous contexts, whereby scale and diversity become the basis for enterprise-level value creation. As the Chief Digital Officer illustrated in the context of talent recruitment:

“We recruit nearly 30,000 people each year across highly diverse business units. The real opportunity is not in monitoring individual cases, but in improving the recruitment process itself. If we benchmark best practices across brands and units, identify where hiring cycles are shortest, candidate quality is highest, and onboarding is most effective, and then systematically transfer those capabilities across the portfolio, a single process improvement scaled across 30,000 hires can generate hundreds of millions in value.”

Recognizing these intertwined pressures and opportunities, Geely’s leaders concluded that incremental IT fixes were fundamentally inadequate. What was required was an enterprise-wide transformation capable of reconstituting the organization’s IT foundation and coevolving with shifting business contexts rather than merely responding to them. This recognition initiated the first cycle of adaptive DT, beginning with value anchoring: a deliberate process of reading the contextual pressures and opportunities generated by heterogeneous business demands and, on that basis, anchoring the direction of value creation that would orient subsequent IS-business interaction.

DT outcomes: Context-transcending IS

Taken together, the three stages constitute adaptive DT: an enterprise-wide transformation logic oriented not toward a fixed endpoint, but toward the ongoing reconfiguration of the conditions of value creation. The outcome is what we term context-transcending IS: an IS configuration that is not confined to isolated contexts, but continually carries capabilities across heterogeneous settings, where they are recombined and renewed to generate value beyond the limits of any single context. Two interdependent dimensions define this configuration: a Polymorphic Capability Base, which enables capabilities to be abstracted, recombined, and redeployed across shifting contexts; and a Reconfigurable Digital Core, which provides the architectural coherence that keeps such recombination governable and scalable.

The first is a polymorphic capability base: a structured enterprise-level asset pool from which capabilities can be abstracted, recombined, and deployed across heterogeneous business contexts without being re-engineered from scratch. This dimension emerged through the cumulative logic of the three stages. Value anchoring replaced fragmented, function-specific demands with a unified evaluative frame for identifying what capabilities the enterprise actually required. Capability orchestration operationalized this frame by abstracting capabilities across organizational boundaries and rendering them reusable beyond their original deployment contexts. Asset platformization then consolidated these capabilities into standardized, API-accessible components embedded within a governed digital infrastructure, forming a coherent enterprise capability map that persists independently of any individual project or business unit. This consolidation represents a structural transformation in how capabilities are held: no longer residing within units or initiatives, they constitute a generative enterprise asset base that underpins value creation across all contexts. As the Director of the Digitalization Center described:

“Through this planning, we created an enterprise capability map that now drives our entire application architecture. It forced us to ask a different question - not ‘what system do we need to build?’ but ‘what capabilities does the enterprise actually possess, and how can they be made available across contexts?’ It integrates previously scattered capabilities into a coherent whole. And the starting point of this entire structure is not technology, it is our shared value logic.”

The second dimension is a reconfigurable digital core: a digital infrastructure that evolves through use, continuously reshaped by the demands and feedback generated across heterogeneous deployment contexts. As capabilities were modularized and embedded within the platform’s boundary resources, the architecture acquired properties beyond static modularity, recombining components in novel configurations and refining its functioning through accumulated use. When capabilities circulated across Geely’s broader ecosystem, they elicited performance expectations and workflow logics that exposed latent mismatches with existing architectural assumptions. These mismatches functioned as generative feedback signals, compelling architectural recalibration and expanding the infrastructure’s capacity to capture emergent value opportunities. The result was a transition from static modularity to reflexive design: an architecture whose evolution is driven by the heterogeneous environments in which it operates, rather than by internally prescribed development roadmaps. As a platform architect described:

“Every time our capabilities enter a new scenario, the platform is forced to update itself. New requirements reveal what our current models cannot handle, gaps we would never have discovered by looking inward. We then evolve the architecture accordingly, not as a planned upgrade cycle, but as a direct response to what the environment is telling us. In that sense, renewal isn’t an add-on—it is built into how the system works.”

Through this recursive process, Geely’s IS evolved into a configuration that does not embed into a pre-given context but continuously transcends successive contextual boundaries through iterative cycles of value anchoring, capability orchestration, and asset platformization. Context-transcending IS, as we theorize it, is an information systems configuration whose value creation logic is not anchored in any single context but is constituted through the recurrent renegotiation of IT artefacts, people, and organizational processes across heterogeneous business situations, such that IS does not merely serve existing value creation logics but actively generates new ones that no single context could produce in isolation. This concept departs analytically from the dominant IS literature, which theorizes IS as context-dependent artifacts designed to fit or align with relatively stable organizational environments (Markus and Robey, 1988; Orlikowski, 1992). Under contextual heterogeneity, this embedding premise is structurally violated: no single context can serve as a stable anchor for IS design, and value creation requires IS to operate meaningfully across structurally, institutionally, and temporally divergent settings. Context-transcending IS addresses this condition precisely: its polymorphic capability base and reconfigurable digital core convert contextual pressures into architectural renewal imperatives and contextual possibilities into generative resources for value creation, transforming heterogeneity from a governance liability into the primary engine of enterprise-level generativity.

Reflecting on adaptive DT: The nature and insights

Our case study of Geely reveals adaptive DT as a theoretically distinct paradigm whose adaptiveness resides not in any single stage or single entity but in the self-reinforcing logic that connects IT, business, and the heterogeneous contexts in which they operate into a continuously evolving whole. Unlike business-driven DT, which embeds IS into contexts defined by strategic intent (Bharadwaj and Dong, 2013; Uhl and Gollenia, 2014), or IT-driven DT, which embeds contexts into the possibilities defined by IS (Besson and Rowe, 2012; Tan et al., 2020), adaptive DT reconstitutes the relationship between these two logics entirely: rather than privileging one over the other, it institutionalizes their productive tension as the organizing principle through which IS capabilities are continuously reconfigured across heterogeneous contexts. Under adaptive DT, this tension is not a problem to be resolved but a structural condition to be governed, precisely because neither business-driven responsiveness nor IT-driven generativity alone can remain coherent across contexts that differ fundamentally in their value imperatives and institutional pressures.

Before elaborating on the theoretical logic of the three stages, two features of the model merit clarification. First, the contextual conditions that initiate each cycle, contextual pressures and contextual possibilities, are not separate antecedent variables, but dialectically related manifestations of the same structural condition: contextual heterogeneity. They are two sides of the same phenomenon because the challenges arising from divergent contexts are inseparable from the potential that their diversity also creates (See Nambisan and George (2024)). Second, although the model presents three stages in sequential order within each cycle, this linearity is theoretically necessary rather than empirically incidental. Each stage creates the substantive preconditions for the next. Value anchoring establishes a shared evaluative frame and governance legitimacy, without which capabilities cannot be recognized, prioritized, or coordinated across heterogeneous contexts. Capability orchestration then transforms these aligned value commitments into organized, reusable capability configurations, which in turn provide the material basis for asset platformization. Platformization is therefore not possible ex ante, but depends on capabilities first being made visible, governable, and recombinable. The sequence is therefore cumulative rather than merely chronological: each stage resolves the central governance problem that would otherwise block the next.

The intra-cycle linearity also reflects a logic of governance sequencing rather than an artificial simplification of organizational life. Across cycles, by contrast, the model is non-linear because each cycle does more than resolve an existing governance problem; it also generates the conditions of the next. As capabilities are accumulated, platformized, and deployed across new contexts, the enterprise confronts broader value demands, more varied usage conditions, and more complex coordination requirements than before. Each subsequent cycle, therefore, reactivates the same governance logic at a qualitatively higher order of complexity. What emerges is not recurrence in the sense of repetition, but a generative spiral in which contextual heterogeneity intensifies as transformation extends the enterprise’s reach across increasingly diverse contexts (Pavlou and El Sawy, 2011; Zollo and Winter, 2002).

The IT governance practices of synchronizing, recombining, and diversifying operationalize this logic, but their relationship to the three stages requires careful theoretical specification. Each practice is most prominently activated at its corresponding stage: synchronizing at value anchoring, recombining at capability orchestration, and diversifying at asset platformization. However, this correspondence reflects governance primacy rather than exclusivity. Synchronizing, which aligns actors’ interpretive frames around enterprise-level value, does not terminate at value anchoring; its logic persists as a background condition throughout capability orchestration and asset platformization, ensuring that architectural decisions and ecosystem relationships remain anchored in shared evaluative commitments. Similarly, recombining’s architectural logic extends into platformization, where the modular reconfiguration of capabilities provides the structural substrate upon which diversifying operates. The governance practices thus function not as episodic compliance mechanisms but as continuous background enablers that sustain productive tension between business-driven responsiveness and IT-driven generativity at every stage of the cycle, preventing the structural myopia of context-specific embedding while ensuring that cross-context generativity remains strategically directed rather than architecturally fragmented.

The two dimensions of context-transcending IS that emerge from this process, a polymorphic capability base and a reconfigurable digital core, are not separate outcomes but mutually constitutive dimensions of the same configuration. The former provides the capabilities that the latter organizes and renders deployable; the latter provides the architectural infrastructure through which those capabilities are recombined and renewed across heterogeneous contexts. This mutual constitution is generative: deployment in novel contexts produces feedback that refines the capability base and expands the reconfiguration possibilities of the core (Rolland et al., 2018; Tiwana et al., 2010). Importantly, the effects of one cycle do not simply flow outward to alter external contextual conditions. They transform the enterprise’s internal capability conditions, thereby reshaping how contextual heterogeneity is perceived and what counts as contextual pressure or possibility in the next cycle. Context-transcending IS is thus not merely the outcome of adaptive DT, but its generative infrastructure, enabling each successive cycle to operate at a higher level of contextual complexity and value-creation potential (Bygstad, 2017; Henfridsson and Bygstad, 2013).

This second-order capability unfolds through three theoretically necessary stages that together point to a counterintuitive insight: IS generates value not by transcending contexts from above, but by engaging them deeply enough to extract what can travel across them. Value anchoring provides this interpretative foundation by establishing that generative transcendence depends on deep contextual immersion (Similar to IT enactment proposed by Orlikowski (2000)). Capability orchestration then converts such contextual immersion into organizational value by identifying and recombining transferable capability patterns across heterogeneous contexts, thereby shifting the locus of IS value from system functionality to cross-context capability transferability. The generativity, in the form of a platform architecture, thus becomes an organizational accomplishment rather than a purely technical feature. Asset platformization completes the logic by turning these accumulated capabilities into a reconfigurable digital core. Instead of imposing a standardized platform on heterogeneous contexts, adaptive DT allows the core to emerge through repeated cycles of immersion, extraction, and abstraction, making its generativity a product of the context-transcending practices from which it was built.

Together, what emerges from adaptive DT is not merely a refined account of digital transformation practice but a fundamental challenge to the embedding premise that has governed IS theorization. As contextual heterogeneity deepens into the structural condition of the contemporary enterprise, the question is no longer how IS can be better embedded, but what becomes theoretically possible and necessary once the embedding premise is released. It is to this question that we now turn.

The future of IS in the enterprise: From context-embeddedness to context-transcendence

The preceding analysis has established that enterprise IS scholarship has long been organized around a foundational premise: IS value is constituted through embedding within specific business contexts, and the task of IS design, governance, and innovation is to achieve and sustain this embedding as effectively as possible (e.g., IT-business alignment, see Luftman et al. (2017)). The insights from Geely’s adaptive DT reveal the structural limits of this premise under conditions of contextual heterogeneity and point toward a fundamentally different trajectory for IS theorization in the future enterprise. Rooted in the context-transcending IS that emerged through Geely’s adaptive DT, we develop context-transcendent IS as a theoretically distinct and increasingly necessary form of enterprise IS for the future. This distinction is important. Context-transcending IS emphasizes movement, that is, the process through which IS stretches beyond any single business context. Context-transcendent, by contrast, captures the more stable conceptual outcome of that process: an enterprise IS form deliberately organized to maintain coherence, transferability, and reconfigurability across heterogeneous contexts. In the following paragraphs, we elaborate on this concept and explain why it matters.

We define context-transcendent IS as a paradigm of information systems whose value is generated through the recurrent traversal of heterogeneous contexts rather than through embeddedness in any single one. Where embedding-based IS treats context as a relatively stable condition to which IS must conform (Benaroch et al., 2006; Tarafdar and Tanriverdi, 2018), context-transcendent IS treats heterogeneous contexts as generative resources: extracting transferable capability patterns from each, abstracting them into reconfigurable assets, and redeploying them as cumulative generative infrastructure across the enterprise. This reconceptualization shifts the locus of IS value from the functional fit between a system and its host context to the cross-context transferability of the capability configurations that IS generates and accumulates (Henfridsson and Bygstad, 2013). The unit of value creation is accordingly reconstituted: no longer confined to discrete value points within a single context, but constituted across an expanding plane of interrelated contexts whose generative potential compounds as IS traverses settings of increasing heterogeneity.

With these disruptive characteristics, context-transcendent IS carries two important implications for existing intellectual conversations on enterprise IS design and development. Existing enterprise IS has often been organized around an integrated operational backbone, through which firms standardize core processes, consolidate data, and use modular interfaces and architectural controls to maintain enterprise-wide coherence while accommodating limited local variation (Ng and Gable, 2010; Wagner et al., 2012). Despite the value-creation performance achieved through local context embedding in existing enterprise IS implementation practices, it is important to reconsider the IS-context relationship within the enterprise for sustained value creation amid contextual heterogeneity in two ways. First, as business increasingly needs to handle the pressures and opportunities of new contexts, it is of strategic value to shift from context-specific fit alone to assessing whether such embeddings also produce enterprise capabilities that remain transferable across heterogeneous contexts, without reverting to repeated one-off customization (Rai et al., 2010). Second, along with this shift, it suggests that IS value should be theorized less as a static achievement within a focal context and more as a broader value surface unfolding across contexts over time. As enterprise IS increasingly takes the form of a nexus of interrelated systems rather than a single bounded system, value may reside not only in local performance improvements, but in the continued articulation, redeployment, and recombination of capabilities across multiple operating environments (Baskerville et al., 2022).

In the development of context-transcendent IS, although the Geely case reveals only partial manifestations of context-transcendent IS, our analysis identifies three constitutive dimensions that define its cross-context generative logic. The first is context permeability, which refers to the structural openness of IS to heterogeneous contexts. It captures the capacity of IS to treat differences in value imperatives, relational arrangements, and dynamic pressures not as frictions to be suppressed, but as generative inputs from which new capability patterns can emerge. This logic distinguishes context permeability from two adjacent concepts: IS flexibility, as conventionally theorized, refers to a system’s capacity to adapt to changing requirements within a given context (Byrd and Turner, 2000; Tallon and Pinsonneault, 2011). It assumes a relatively stable contextual reference point against which adaptation is evaluated, and its generative scope remains bounded by that reference point; Comparatively, IS agility emphasizes the speed and responsiveness with which IS reacts to environmental change (Overby et al., 2006; Sambamurthy et al., 2003), but it does not capture the structural capacity of IS to remain open to contexts that differ fundamentally in kind rather than merely in degree. Comparatively, context permeability rests on a different logic: It treats irreducible contextual difference not as a constraint on adaptation, but as a source of generativity. While flexibility and agility could improve IS performance within an existing contextual range, wide touch and permeation expand the range of contexts from which IS can learn, extract, and generate value. The theoretical basis of context permeability lies in the recognition that IS capabilities are not pre-given system properties, but achievements of situated enactment (Gruys et al., 2008). Specifically, capabilities become visible, and thus available for abstraction, only through deep and sustained engagement with specific contexts. Context permeability is therefore not passive receptiveness, but an active governance achievement that requires IS to engage deeply with contextual differences rather than smoothing them away in advance. We therefore propose:

Proposition 1

The generative potential of context-transcendent IS is conditioned by its structural openness to heterogeneous contexts, that is, the broader the permeability of IS in contextual variation, the richer and more transferable the capability patterns available for abstraction and redeployment across the enterprise.

Second, context immersion refers to the depth of enacted engagement through which IS is shaped by, and responsive to, the specific logic of each context it enters. The theoretical foundation of this dimension lies in sociomateriality: IS does not possess fixed architectural properties that are deployed into contexts, but properties that emerge from and are continuously reconfigured through entanglement with specific organizational arrangements, human practices, and institutional logics (Orlikowski and Scott, 2008). This distinguishes immersion from IS implementation, which treats context as a target environment to be managed around a relatively stable IS configuration (Berente et al., 2019; Markus et al., 2000): where implementation presupposes IS as the active subject and context as the passive object, immersion inverts this directionality, treating each context as the generative source from which IS is continuously reconstructed. The deeper the immersion in a context, the richer the capability patterns that become visible, extractable, and available for cross-context abstraction. We therefore propose:

Proposition 2

The abstraction potential of context-transcendent IS depends on the depth of its constitutive engagement with each context it enters. The deeper the engagement, the richer the capability patterns available for abstraction and redeployment across the enterprise.

Where permeability and immersion establish the conditions under which IS can be shaped by heterogeneous contexts, context transferability completes the generative logic by determining what IS can do with what it has learned. It refers to the capacity of IS to extract, abstract, and redeploy capability patterns across heterogeneous contexts, such that the generative potential developed through immersion in one context becomes productive infrastructure in others. This dimension is analytically distinct from IS scalability, which extends an established logic across a larger number of similar contexts (Bharadwaj et al., 2013b). Scalability presupposes contextual similarity because its value derives from replication: the more stable and comparable the contexts, the more efficiently an existing capability can be reproduced at scale. Transferability rests on a different logic. It treats contextual heterogeneity as generative because differences across contexts reveal which elements of a capability are locally bounded and which can be abstracted, recombined, and redeployed elsewhere. Its value, therefore, compounds not through repetition, but through reconfiguration. Each new context does not simply receive an existing capability, but helps refine and expand the portable capability patterns available for subsequent use. The theoretical basis of this distinction lies in the dynamic capabilities view (Teece et al., 1997): the ability to reconfigure capabilities across fundamentally different environments constitutes a second-order organizational competence whose value increases with heterogeneity rather than being constrained by it. In context-transcendent IS, this second-order competence is embedded in the system’s architecture rather than left to episodic managerial intervention. We therefore propose:

Proposition 3

The generative potential of context-transcendent IS compounds with the degree of heterogeneity traversed, that is, the greater the structural, institutional, and temporal differences among the contexts through which IS moves, the richer the accumulated capability patterns and the more expansive the generative infrastructure available for future contexts.

Having established the three constitutive dimensions of context-transcending IS, we now turn to the structural conditions that make its emergence organizationally possible. From the case study of Geely, we identified key conditions that differ from what is widely accepted in enterprises but are important for the enactment of these practices, enabling IS to navigate diverse contexts for value creation. The first condition is the decomposability of business processes. Conventional enterprise IS was built on the assumption that business logic should be encoded in tightly integrated, process-centric architectures, that is, a design principle that maximized operational efficiency within stable contexts but rendered reconfiguration across heterogeneous ones structurally prohibitive (Davenport, 1998). What the Geely case reveals is that context-transcendent IS becomes organizationally achievable only when business processes can be disaggregated into modular, recombinable capability units that retain coherence independently of the specific contextual arrangements that produced them. As such, decomposability is the organizational precondition for context transferability: capabilities locked within monolithic process architectures cannot be extracted, abstracted, or redeployed across contexts, regardless of how immersive or permeable the IS configuration may otherwise be (Sunyaev et al., 2023). This is not merely a technical design choice but a governance commitment: decomposability requires that the enterprise actively resist the organizational tendency toward process entrenchment and capability boundedness that conventional IS governance reinforces (Rinta-Kahila et al., 2023; Rolland et al., 2018). We therefore propose:

Proposition 4a

Decomposability of business processes is a necessary condition for the emergence of context-transcendent IS: the greater the modularity of business capabilities, the greater IS’s capacity to extract, abstract, and redeploy transferable patterns across heterogeneous contexts.

The second condition is the institutionalization of heterogeneity as an operant resource. Prevailing IS governance has treated contextual heterogeneity as a structural liability to be minimized through standardization, harmonization, and the enforcement of uniform operating logics across business units (Kettinger et al., 2010). This governance orientation reflects that contextual difference is a coordination cost to be reduced rather than a generative condition to be cultivated. What Geely’s adaptive DT demonstrates is that context-transcendent IS requires a fundamental reorientation of this logic: heterogeneity must be actively institutionalized as an operant resource: recognized in governance arrangements and embedded in the organizational routines through which IS capabilities are developed and deployed. This reorientation matters at two levels. At the governance level, enterprises must resist premature standardization, namely the impulse to suppress contextual tension through uniform IS logic before its generative value has been extracted (Hanseth and Modol, 2021). At the architectural level, IS must be designed to accommodate contextual variation rather than convergence. Otherwise, heterogeneous contexts will continue to be treated as barriers rather than generative resources, and the conditions for context immersion and permeability will remain systematically underdeveloped. We therefore propose:

Proposition 4b

Context-transcendent IS is conditioned by the institutionalization of heterogeneity as an operant resource: enterprises that govern contextual heterogeneity as a strategic asset rather than a governance burden are better positioned to cultivate the permeability, immersion, and transferability through which context-transcendent IS generates and accumulates value.

The two organizational conditions identified above, the decomposability of business processes and the institutionalization of heterogeneity as an operant resource, have not yet been widely met across enterprises of varying size and digital maturity. Their co-presence in the Geely case reflects the organizational consequences of operating at the intersection of extreme business complexity and accelerated digital maturity: Geely’s multi-brand, multi-market architecture rendered the tension between IS and heterogeneous contexts structurally unavoidable, while its digital maturity enabled the rapid reconfiguration of business processes in response to shifting value priorities. Yet as more enterprises grow through diversification, multi-market expansion, and the multiplication of business models, contextual heterogeneity is becoming a structural condition that no enterprise can indefinitely defer. Thus, Geely is not an anomaly but an early and unusually visible instance of a trajectory that is becoming generalized: one whose organizational complexity has rendered the future of IS legible ahead of its time, and whose adaptive practices provide the empirical foundation for our theorization of context-transcending IS. What this theorization ultimately traces is an evolutionary arc in the relationship between IS and value creation: from IS as the automation of discrete processes, to IS as the enabler of value within a given context, to IS as the generator of value across heterogeneous contexts.

Theoretical implications

Practical implications

Beyond its theoretical implications, our work offers actionable guidance for organizations across different stages of maturity. For large incumbents operating at scale, our findings illustrate how IT can be repositioned from a reactive service provider to a proactive orchestrator of transformation by leveraging its generative potential across heterogeneous business contexts. In such environments, value cannot be predetermined through fixed process designs for predefined business contexts; instead, identifying and cultivating core digital and business capabilities as strategic assets becomes essential. This shift requires rethinking the division of labor and accountability between IT and business units. Rather than isolating IT at the periphery of business decision-making, incumbents should establish a more agile governance structures that enable continuous joint sensemaking and co-ownership of value creation, ensuring that digital transformation remains both technically feasible and strategically coherent at scale.

For start-ups and digital-native firms, the model highlights the importance of building modular, recombinable capability cores early, even before rapid growth and expansion generate a heterogeneity of contexts that incur architectural debt and coordination tensions. While start-ups often prioritize speed and experimentation, our findings suggest that they benefit from investing in flexible capability architectures that can accommodate future heterogeneity without costly redesign. By adopting governance practices that support shared interpretation of value (synchronizing), structured capability reuse (recombining), and broad participation in innovation (diversifying), start-ups can scale more smoothly and avoid the architecture lock-in that often hinders later-stage digital transformation.

For policy makers and industry regulators, the model underscores the need to view digital transformation not merely as technology deployment but as the cultivation of adaptive digital capability ecosystems. Policy interventions should therefore facilitate the development of shared digital infrastructures, interoperable standards, and cross-organizational coordination mechanisms that support distributed innovation. Encouraging modularity, open interfaces, and the recombination of inter-firm capabilities would lower barriers to collaboration and enable both large enterprises and smaller firms to participate in innovation-driven ecosystems. As digital infrastructures become more complex and interdependent, policies that promote transparent governance, cross-sector learning, and capability sharing are essential to sustain competitiveness and ensure that digital transformation yields broad societal benefits.

Limitations

Our study is not without limitations. First, the findings are derived from a single case within the automotive industry, which raises the familiar challenge of external validity in interpretive case research (Walsham, 2006). Although our model provides a theoretically informed account of adaptive DT, its robustness and boundary conditions must be further examined through comparative or multi-case designs across industries with different levels of complexity, regulation, and technological maturity (Eisenhardt and Graebner, 2007). Future research could apply or extend our framework to sectors such as finance, healthcare, logistics, or public services to assess its scope of applicability and refine its generalizability.

Second, our theorization is inevitably shaped by the empirical materials available to us. While we observed a rich, longitudinal process of adaptive DT in Geely’s practice, other configurations, that is, whether IT-driven or business-driven, may unfold differently and also support the development of IS in the future enterprise. Variations in sociotechnical elements (Sarker et al., 2019), such as organizational culture, governance structures, legacy architectures, and market conditions, could produce alternative pathways that were not captured in our data. Subsequent studies should therefore explore different transformation trajectories, with additional forms of IS development, to challenge and extend the model proposed here.

Together, these limitations underscore the need for future work to deepen, refine, and broaden the empirical base for understanding how enterprises build adaptive information systems under increasingly dynamic and heterogeneous conditions. Despite these limitations, we believe our findings will contribute to understanding the future of IS in the enterprise and how it can be developed through a new pathway of DT for competency building.