The Role of Artificial Intelligence in Business Model Innovation of Digital Platform Enterprises

Abstract

The advancement of artificial intelligence (AI) offers new opportunities for business model innovation in digital platform enterprises. Despite growing interest in AI applications, the specific mechanisms through which digital platform firms leverage AI to drive business model innovation remain insufficiently explored, particularly from the integrated perspective of resource mobilization and organizational capability reconfiguration. To address this gap, this study conducts a single-case analysis of a highly successful digital platform enterprise in China. This study explores how digital platform enterprises can effectively utilize AI technologies to support business model innovation. The findings reveal that AI technologies enable digital platform enterprises to develop organizational capabilities in intelligent connectivity, intelligent development, and intelligent governance. AI-enabled organizational capabilities in digital platform enterprises evolve through three progressive stages: AI-assisted, AI-augmented, and AI-integrated. At each stage, these capabilities are shaped through different types of resource actions—namely, entry-oriented resource patchwork, depth-oriented resource arrangements, and coordination-oriented resource orchestration. This study offers practical insights for digital platform enterprises seeking to leverage AI technologies for business model innovation. By integrating the concepts of resource actions and organizational capabilities, it provides a dynamic explanation of how AI drives innovation in digital platform business models. The research contributes to the theoretical advancement of human-AI integration and resource action frameworks, offering actionable intelligence for the broader industry.

1. Introduction

The deep integration of artificial intelligence (AI) across various industries continues to reshape organizational behavior and drive socio-economic transformation, constituting a critical component of the ongoing global technological revolution and industrial transformation. AI’s unprecedented impact on society fuels new dynamics that transcend traditional boundaries [1]. As a novel form of “new infrastructure” facilitating the convergence of the digital and physical economies, digital platforms have emerged as a key organizational form in industrial transformation [2]. A growing number of digital platform enterprises adopt AI technologies to develop proprietary AI models or vertical AI applications, thereby enhancing their competitive capabilities in cloud computing, networks, devices, chips, and blockchain. This synergy fosters complementary capabilities across platforms, promotes coordination, and drives the transformation of operational and business models. Within this context, understanding the processes and mechanisms by which AI enables business model transformation has become a central concern for both academia and industry.

Business model innovation refers to the process through which enterprises redesign or optimize their products, services, value chains, and revenue models to adapt to changing market demands and competitive environments [3]. Digital platforms, as a novel form of business model, leverage privately controlled digital infrastructures such as websites or applications to transform enterprise operations. Through business model innovation, digital platform enterprises can improve operational efficiency and create new avenues for value generation [4], foster value co-creation, and gain a competitive advantage, ultimately maximizing overall value [5]. The rapid development of AI technology provides new pathways for business model innovation in digital platform enterprises. Existing research highlights AI’s role in enhancing organizational capabilities, enabling enterprises to accurately identify market opportunities, optimize resource allocation, and improve decision-making efficiency, thus promoting business model innovation by enhancing AI capabilities [6,7]. However, most studies treat AI as an external technological context, focusing on its facilitation of process optimization and efficiency gain [8,9,10], leaving the mechanisms through which enterprises leverage AI to drive resource actions, capability enhancement, organizational coordination, and overcome competitive challenges largely unexplored.

Whether AI can play a pivotal role in driving business model innovation in digital platform enterprises depends not only on the AI technologies themselves [11] but also, and more critically, on how these enterprises mobilize AI resources and develop organizational capabilities to enable value co-creation between humans and AI [12]. AI-driven business model innovation necessitates a stronger focus on the interactions between humans, AI technologies, and tasks [13]. These interactions are inherently complex and dynamic, requiring intentional organizational design and continuous adaptation [14]. To effectively drive such innovation, digital platform enterprises must actively orchestrate AI-related resource actions and dynamically reconfigure organizational capabilities to accommodate the evolving demands of AI-driven business models [15]. Understanding the nature and typology of AI capabilities across different stages of development is, therefore, critical for analyzing how AI supports innovation.

Prior research has examined how digital platform enterprises leverage AI and data analytics to enhance network effects and generate user value in the context of business model innovation [16,17,18,19], with a predominant focus on user-centric and market-oriented perspectives [20]. However, much of this literature treats AI as an external technological enabler, overlooking its interactive and co-evolutionary relationship with internal organizational systems. Moreover, limited attention has been given to the internal mechanisms underlying the evolution of AI capabilities within enterprises [21,22]. While existing studies have explored AI’s role through lenses such as institutional logic, digital network effects, and socio-technical systems [23,24,25], they fall short of elucidating the specific ways in which AI contributes to business model innovation and commercialization pathways. Resource mobilization theory’s link to organizational capability development in platform contexts remains under-theorized, and the underuse of the architecture view limits a holistic understanding of AI’s role in digital platform business model innovation. In response to the identified theoretical gaps and practical needs, this study investigates Quwan Technology Group, a Guangzhou-based company in China, using a case study approach to trace its evolution from TT Voice to the development of a digital human resources platform. From the perspective of integrating AI resource actions and organizational capabilities, this research explores two critical questions: How do digital platform enterprises, through AI-enabled resource actions and capability reconfiguration, drive the formation and dynamic evolution of business model innovation? What role does AI play in this innovation process?

By answering these questions, this study contributes to the literature in three key ways. First, it proposes an integrated conceptual framework that systematically links AI resource mobilization with the dynamic reconfiguration of organizational capabilities, thereby deepening our understanding of the longitudinal processes through which AI drives business model innovation. Second, it empirically identifies a typology of AI-enabled organizational capabilities and explicates their distinct roles and progressive evolution throughout the innovation lifecycle. Third, through an in-depth case study of a digital platform enterprise, the research uncovers micro-level mechanisms illustrating how AI resource actions and capability reconfiguration integration overcome competitive challenges and enable effective value co-creation between the organization and AI. By offering this nuanced, process-oriented perspective on AI-driven business model innovation, the study not only extends existing theoretical models but also provides actionable insights for digital platform firms seeking to leverage AI technologies to sustain innovation and competitive advantage in rapidly evolving markets.

2. Theoretical Background

2.1. AI Enablement in Digital Platform Enterprises

AI and Digital Platform Enterprises represent two fundamental forces reshaping the economic and organizational landscape [25,26,27]. AI, broadly defined as a system capable of interpreting data, learning from it, and autonomously completing specific tasks, serves as an umbrella term encompassing a range of advanced technologies, including machine learning, deep learning, and generative models [10,28]. Digital platform enterprises, which serve as ecosystem orchestrators by enabling interactions, knowledge sharing, and collaboration, increasingly integrate AI to enhance operational efficiency and personalize services [29,30]. The convergence of AI and digital platform enterprises represents not merely a technological integration but a dynamic interplay that drives the reconfiguration of organizational resources and the transformation of organizational capabilities [29,31,32,33].

Many studies have highlighted that the convergence of AI and digital platform ventures has significant implications for innovation [20,34,35], particularly as evolving interactions between human and machine intelligence reorganize the mechanisms of value creation and capture in digital platform business models [36]. Malgonde et al. (2020) conceptualize digital platforms as Complex Adaptive Business Systems, highlighting the dynamic evolution of the goals, preferences, and constraints of multilateral participants [37]. They also emphasize the presence of irreducible uncertainty within these systems—uncertainty that cannot be resolved through traditional data collection and processing methods. In this context, AI-powered recommender systems enable platforms to adapt dynamically by prompting agents within the system to co-evolve and learn collaboratively [37]. However, the continuous advancement of AI technologies introduces new layers of complexity to platform integration—beyond technical challenges, this also involves evolving organizational interactions with AI systems [38,39,40]. To better understand this transformation, an expanding body of research now investigates how AI can empower digital platform enterprises [16,41,42,43].

Existing studies have systematically examined how digital platform firms facilitate organizational transformation through the empowerment of AI [10], particularly from the perspectives of institutional theory, digital network effects, and socio-technical systems [23,24,25]. First, some studies have explored the process of embedding and institutionalizing AI technologies in digital platform firms. For instance, Sandvik et al., within the framework of institutional theory, argue that platform companies overcome commercialization barriers through four key steps: shaping value logics, cultivating demand, forming ecosystem alliances, and catalyzing institutional change [23]. Gregori et al. highlight that sustainable business models for digital platforms integrate multiple institutional logics, including environmental, social, and commercial, while the emergence of digital technologies such as AI has introduced new digital logics [44]. These digital logics are reflected in hybrid value propositions, integrative value creation, and multidimensional value capture; yet, they also present both complementarities and potential tensions with environmental and social logics [44]. While these studies offer valuable insights into the role of AI in platform organizations, they primarily focus on how AI technologies are legitimized within institutional contexts. However, the existing literature has not sufficiently addressed how platform firms can gain and sustain competitive advantage through AI—particularly regarding the specific mechanisms by which AI drives business model innovation and supports commercialization. In particular, the core process through which platform firms cultivate and integrate strategic resource actions and organizational capabilities to develop distinctive AI-enabled organizational capabilities remains underexplored.

Second, a number of studies have emphasized the central role of AI in catalyzing distinct forms of data-driven network effects [45]. While traditional network effects emphasize value creation through increased user numbers, AI-driven data network effects focus on the cyclical value generated from data accumulation and intelligent analytics [24]. Gregory et al. (2021) propose that platform AI capabilities—defined as a platform’s ability to learn from data in order to continuously improve its products or services—give rise to new forms of platform externalities [16]. These externalities manifest when the utility a user derives from the platform becomes a function of the scale and effectiveness of data-driven learning and AI-enabled improvements [46]. Such improvements are reflected in enhanced product functionality, platform quality, and user experience [24,47]. Although these studies have advanced the understanding of AI-driven data network effects, limited attention has been paid to how platform firms can strategically cultivate, manage, and leverage these effects to establish and sustain competitive advantage. This is particularly pertinent given the increasing diversification of data sources and the tightening of data governance regulations. Further research is needed to investigate the strategic decisions and organizational practices required to navigate and capitalize on AI-driven data network effects under these evolving conditions.

Third, the increasing autonomy of AI systems necessitates a reexamination of existing knowledge regarding how organizations interact with AI [25]. As AI systems gain greater autonomy, novel patterns of interaction are emerging between AI agents and human actors [48]. These developments have drawn significant attention from researchers and practitioners seeking to understand the evolving modes of human–algorithm collaboration in contexts marked by algorithmic autonomy and shifting levels of control [49,50]. Recent research has explored how humans and AI can operate in hybrid configurations, including through mutual augmentation or joint decision-making arrangements between human actors and information systems [51,52,53,54]. This literature has primarily focused on human-computer interaction (HCI) models involving either individual or team-based interactions with algorithmic agents [54,55,56]. However, AI systems are increasingly deployed as multi-agent systems (MAS), wherein the digital platform environment comprises not only human users and platform operators, but also multiple AI algorithms with varying levels of autonomy [57,58]. These configurations give rise to complex, many-to-many interaction patterns that extend far beyond traditional one-to-one or many-to-one HCI frameworks. How such distributed, multi-actor interactions lead to emergent behaviors—and how these behaviors impact platform efficiency, innovation capacity, and resilience—remains an open and critical area for future research.

To address this research gap, this study adopts a technology empowerment perspective by categorizing the impact of AI on digital platform enterprises into four types of organizational capability reconfigurations: AI-assisted, AI-augmented, AI-integrated, and AI-substitutive, as illustrated in Figure 1. The objective is to examine how digital platform enterprises can embed AI technologies into their products, services, processes, or activity systems to develop AI interaction capabilities that support business model innovation, thereby contributing to the theoretical discourse on AI in the context of digital platforms. This study focuses specifically on the AI-assisted, AI-augmented, and AI-integrated capabilities, aiming to explore the dynamic co-creation relationship between AI and humans. The AI-substitutive capability is excluded, as its emphasis does not align with the central theme of value co-creation between digital platform enterprises and AI.

2.2. Business Model Innovation from the Perspective of Resource Mobilization

Traditional resource-based theory (RBT) views the endowment and possession of organizational resources as key drivers of innovation and competitive advantage [59]. It posits that a firm’s competitive advantage arises from resources that are valuable, rare, inimitable, and non-substitutable, emphasizing the role of static resource heterogeneity in developing core competencies—an essential theoretical foundation for business models [60]. However, RBT has been critiqued for its limitations in explaining how firms dynamically align internal capabilities with external conditions to achieve sustained competitive advantage. Specifically, it tends to overlook the dynamic and evolutionary nature of resources [61]. In response, scholars have increasingly turned their attention to the processes by which organizations acquire, integrate, and reconfigure resources over time [62].

The dynamic capabilities perspective, an extension of the resource-based theory (RBT), has enriched our understanding of the dynamic aspects of resource allocation. However, it has not fully clarified the specific mechanisms by which resources are deployed [63]. To address this gap, the concept of resource mobilization has been introduced as a refinement of traditional RBT and dynamic capabilities theory [62,64,65]. Resource mobilization refers to the organizational activities aimed at leveraging both internal and external resources to enable enterprises to align effectively with evolving resource contexts [66]. By following the logical chain of “resource context–resource mobilization–action outcomes,” this perspective offers insights into how firms orchestrate resources to drive business model innovation and enhance their competitive advantage [67].

The evolution of resource mobilization in response to shifting resource characteristics across different stages of enterprise development can be divided into three processes: resource bricolage, resource orchestration, and resource synchronization. These processes correspond to different stages in the organizational lifecycle, from startup to maturity [62]. While all three involve resource-related actions, they vary in application depending on the enterprise’s stage of development [68]. Resource bricolage is especially relevant during the startup phase, where new ventures mobilize resources in environments of scarcity [69]. Resource orchestration, rooted in resource management and asset orchestration models, involves the strategic recombination of resources to ensure efficient utilization, particularly when an enterprise is still in the process of developing or refining its core capabilities [70]. This process includes orchestration routines, coordination among those routines, and their dynamic alignment with changing organizational contexts, and is most prominent during the growth phase [71]. Resource synchronization, formally introduced by Sirmon et al. (2011), emphasizes how managers actively align and synchronize resources to establish competitive advantage [62]. This process includes mechanisms such as resource utilization and applies across various organizational contexts where proactive resource management is essential. The structured processes of resource bundling and orchestration have become central themes in research on business model innovation within the resource-based view [72].

Recent literature highlights the importance of resource bricolage, orchestration, and synchronization in overcoming growth barriers and facilitating business model innovation [67,69,71]. These processes have become integral to understanding how organizations optimize resource allocation to drive innovation and promote sustainable development [73,74]. However, while existing research has examined the relationship between resource orchestration and capability development—such as how enterprises navigate transformation and growth—there is still limited discussion on how resource mobilization can drive business model innovation, particularly from the perspective of building organizational capabilities. Therefore, it is essential to further investigate the mechanisms of AI-driven resource mobilization in the context of business model innovation, in order to deepen our understanding of how digital platform enterprises can strategically allocate AI-related resources and cultivate organizational capabilities that foster sustainable competitive advantage and contribute to the achievement of sustainable development goals [74].

2.3. The Impact of AI Empowerment on Business Model Innovation in Digital Platform Enterprises

Digital platform enterprises represent a novel organizational form arising from digital technological transformation, distinguished by decentralized governance and multi-sided architecture structures [75]. Existing research on business model innovation within these enterprises can be broadly categorized into two analytical perspectives: the component view and the architecture view [76]. The component view emphasizes the individual elements of business models, focusing on how specific components evolve over time [77,78,79]. Much of the current literature on AI empowerment aligns with this perspective, analyzing the influence of AI on particular facets of business model innovation.

In the realm of value proposition innovation, AI empowers digital enterprises to transition from the traditional product-centric logic to a service-oriented and ecosystem-driven logic. [76,80,81]. Regarding product and service innovation, AI serves two key functions: first, AI-empowered platforms enhance the depth and quality of interactions between enterprises and users [82]. Second, AI technologies allow firms to extract deeper insights from user data, analyze behavioral patterns, and mitigate information asymmetries, which collectively foster continuous product innovation [18,83]. In the context of boundary shaping, AI-driven empowerment facilitates the redesign of traditional business processes through intelligent data acquisition and automated workflows, significantly enhancing operational efficiency [84].

While the component view has significantly advanced our understanding of how AI influences individual elements of business model innovation, it often overlooks the interdependencies and complementarities among these elements, limiting the ability to fully grasp the complexity inherent in business models [85]. In response, business model innovation research has increasingly shifted toward an architectural view [86]. Rooted in a value system perspective, the architectural view examines the holistic configuration of interdependent and complementary components within the business model, emphasizing how their integration contributes to innovation outcomes [87]. For instance, Madsen et al. (2020) investigated how firms introduce new core components to align internal structures with evolving external environments [88], while Peprah et al. (2022) demonstrated how business model evolution in startups operating in weak institutional environments can trigger cascading changes across value proposition, value creation and delivery, and value capture [89].

Despite these advances, limited research has explored the influence of AI empowerment on business model innovation in digital platform enterprises from the architectural perspective. This gap constrains a systematic understanding of how AI affects the overall configuration and transformation of business models. Therefore, it is crucial to examine AI empowerment through the lens of business model architecture, as this approach can reveal how AI fosters the coordination and integration of business model components, ultimately driving comprehensive transformation and strategic upgrading in digital platform enterprises.

2.4. Literature Review and Research Framework

A review of relevant theoretical perspectives suggests that AI can support digital platform firms in innovating their business models by enhancing operational efficiency and enabling the delivery of novel services [90]. First, prior studies have explored how platform firms adopt and utilize AI through various lenses, including institutional logic, digital network effects, and socio-technical systems theory [23,44]. However, institutional theory overlooked the endogenous evolution of AI applications within platform enterprises and fails to account for the disconnects between strategy and execution, cognition and outcomes, and technology and commercialization. The data network effects perspective tends to oversimplify the relationship between data and competitive advantage, neglecting the contextual adaptability and integrative role of AI-related technical resources and organizational capabilities [24,45,47]. Similarly, socio-technical systems theory predominantly focuses on user-side dynamics, giving insufficient attention to the staged nature of AI deployment and its complex interplay with the architecture of organizational capabilities [91]. As a result, the underlying mechanisms by which digital platform companies leverage AI to build and sustain competitive advantage—particularly the dynamic, iterative processes through which AI enables business model innovation and commercialization—remain insufficiently explored.

Second, in the domain of business model innovation, the concept of resource mobilization has emerged as a significant extension of traditional resource-based theory (RBT) and dynamic capabilities theory. It offers a novel lens through which to understand how enterprises drive business model innovation via dynamic and context-sensitive resource reconfiguration. However, the mechanisms through which resource mobilization fosters business model innovation—particularly from the perspective of organizational capability development—remain in an early stage of exploration.

Finally, regarding digital platform enterprises, existing research on business model innovation can be broadly categorized into two perspectives: the component view and the architecture view. The component view emphasizes discrete elements such as value proposition innovation, product and service innovation, and boundary redefinition, yet it often overlooks the interdependencies and complementarities among these components. In contrast, the architecture view highlights the holistic configuration and integration of interrelated components, emphasizing how their synergistic interactions contribute to innovation outcomes. Despite its conceptual value, this architectural perspective has been underutilized in studies examining the influence of AI on business model innovation in digital platform enterprises, thereby constraining a more comprehensive understanding of AI’s integrative and transformative potential.

This study seeks to fill these theoretical gaps by exploring how AI enablement drives business model innovation in digital platform enterprises via the interplay between resource actions and the integration of organizational capabilities. Drawing on the resource-action evolutionary process model proposed by Jiang, Y. et al. (2023)—which follows the logical sequence of “resource situation perception–resource action implementation–action outcome presentation”—this study focuses on digital platform enterprises as the primary context for investigating the process of AI-enabled business model innovation [92]. Such innovation is shaped by dynamic internal and external contextual factors that influence how organizations perceive opportunities and respond to challenges. Within this context, resource strategy serves as a critical input. Firms must adopt dynamic resource action strategies to integrate external AI technologies with internal systems, thereby generating new knowledge and enhancing organizational capabilities. To capture this complex process, the study introduces a theoretical lens centered on the integration of resource action and organizational capability. Organizational capabilities support further resource actions, thereby achieving innovative outcomes. As contextual conditions evolve, a firm’s ability to adapt its resource action strategy becomes essential for developing AI-related capabilities and achieving effective business model innovation. The direct outcome of this process is the continuous evolution of platform-based business models. Building upon the classic “situation–strategy–outcome” logic, we propose an extended analytical framework: “resource situation–resource action–organizational capability–innovation outcome.” The corresponding research framework is presented in Figure 2.

Specifically, we first examine how dynamic internal and external contextual factors influence business model innovation. Second, we analyze the resource action strategies adopted by digital platform firms at different stages of development and their impact on innovation trajectories. Third, we investigate the coupling of resource actions with organizational capabilities to cultivate AI-enabled competencies. Finally, we explore the evolutionary patterns of business model innovation outcomes within digital platform enterprises.

3. Materials and Methods

This study seeks to answer the questions of how digital content platforms achieve scenario-based business model innovation, which are typical “how” and “why” inquiries. We employ a grounded theory (GT) approach [93], widely used in the field of information and communication technologies, as it allows for “the co-creation of theory with data and participants” [94]. The choice of this method stems from its openness to multiple perspectives, avoiding preconceived notions, and letting the data “speak for itself,” while integrating literature, data, and experiential insights [93]. Additionally, this study involves an exploratory analysis of the detailed processes of the case enterprise to investigate the mechanisms through which digital platform enterprises realize business model innovation empowered by AI. The single-case study method is particularly suited for this research as it allows for in-depth process mechanism analysis, leading to valuable insights. Consequently, this approach is well-aligned with the objectives of the study.

3.1. Case Selection

First, this study identifies and illustrates the distinct bottlenecks encountered by digital platform enterprises at various developmental stages in order to map the evolving internal and external contextual environments faced by the focal firms. This contextual mapping allows for an in-depth analysis of how these environmental dynamics influence AI-driven business model innovation and the specific roles played by different contextual factors. Second, we examine the resource action strategies and/or organizational capability reconfigurations undertaken by the case enterprise during its AI-enabled business model innovation processes. By identifying the resource actions and capability transformations across different stages, we assess their impact on business model innovation. Finally, based on the above analyses, we characterize the features of each stage of AI-driven business model innovation and elucidate the dynamic evolutionary patterns that underlie this process.

Case selection adheres to the following criteria:

Industry Representation: In recent years, China’s Internet social platform industry has experienced rapid development, accompanied by significant regulatory, market, and technological transformations. As an emerging industry, it offers a valuable lens for examining how shifting internal and external environments influence business model dynamics. Moreover, the sector has been at the forefront of adopting cutting-edge technologies such as big data and AI, fostering diverse AI-enabled business model innovations that align well with the thematic focus of this study.

Enterprise Representativeness: Founded in 2014, Quwan Technologies has undergone three distinct phases of business model evolution, each marked by significant transformation. The company was recognized among the “Top 10 AI Technology Companies in Guangdong Province” in 2023 and has been listed in China’s Top 100 Internet Companies for six consecutive years. It also holds titles such as “National High-Tech Enterprise” and “Professional, Specialized, and Innovative SME.” Quwan’s sustained exploration of AI technologies and continuous innovation in both technology and business models render it a representative case of AI-driven digital platform enterprises, offering valuable insights for constructing new theoretical frameworks rooted in real-world best practices.

Data Accessibility: Our research team has established a collaborative relationship with Quwan Technologies, granting access to rich first-hand data through in-depth field investigations. The company’s growing prominence in the AI sector also ensures the availability of extensive public data, allowing for robust triangulation between primary and secondary sources.

3.2. Case Overview

Drawing on key enterprise events and field observations, Quwan Technology’s process of leveraging AI for business model innovation can be delineated into three distinct stages, each aligned with a transformation in the company’s value proposition. These phases—Startup, Growth, and Transformation—correspond to successive business model configurations: the Tool Product Phase, the Social Platform Phase, and the Ecosystem Empowerment Phase.

• First Stage: Startup Phase (2013–2019)

Quwan Technology commenced its operations in 2013 with the launch of TT Voice, a real-time voice communication tool specifically tailored for multiplayer gamers. The product addressed a critical pain point in online gaming: seamless and high-quality in-game communication. By integrating speech recognition technology to enhance voice clarity and by partnering with popular online games, Quwan rapidly attracted a user base. During this initial phase, the firm adopted a vertical market penetration strategy, focusing on niche user demands and cultivating growth through community-driven word-of-mouth marketing. The business model during this stage was product-centric, with a clear emphasis on functional utility and user acquisition.

• Second Stage: Growth Phase (2019–2022)

In 2019, Quwan evolved from a communication tool provider into a comprehensive social platform, enriching TT Voice with features such as voice chat rooms, community forums, and content-sharing functionalities. The incorporation of AI-driven recommendation systems significantly enhanced personalization, deepened user engagement, and improved retention rates. During this stage, Quwan leveraged social network effects and product differentiation to expand its user base and strengthen its market position. The shift in business model reflected a transition from single-purpose utility to multi-sided platform dynamics, emphasizing interaction, community building, and increased user stickiness.

• Third Stage: Transformation Phase (2022–Present)

From 2022 onwards, Quwan embarked on a strategic transformation towards ecosystem empowerment, integrating digital human platforms and expanding its capabilities in virtual human technologies across diverse application domains. AI became the central driver in enabling platform openness and modularity, fostering co-creation with third-party developers and strategic partners. This phase marked a pivot from user-centric services to platform-centric ecosystem orchestration, as Quwan pursued diversified monetization models, collaborative innovation, and long-term sustainability. The enterprise’s focus on building a robust product matrix and attracting ecosystem participants significantly enhanced its platform competitiveness and adaptability in a fast-evolving digital landscape.

3.3. Data Collection

This study employs a qualitative case study approach, with Quwan Technology as the focal enterprise. The case data were primarily derived from semi-structured interviews with mid- and senior-level managers, complemented by secondary sources and field observations, thereby ensuring triangulation of data and enhancing the validity of the findings. To construct a comprehensive understanding of Quwan Technology and its AI-driven business model innovation, data collection proceeded in two stages.

Prior to fieldwork, the researchers conducted a systematic review of publicly available information regarding Quwan Technology. This included industry reports, news releases from the company’s official website, and other reputable online sources. These materials provided foundational insights into the firm’s operating context, development trajectory, and technological deployments, and informed the formulation of the interview guide. The research team conducted two rounds of in-depth, face-to-face semi-structured interviews over a two-day period with senior managers and project leaders at Quwan Technology. These sessions captured rich, decision-level insights into the company’s strategic initiatives, resource configurations, and business model evolution processes. All interviews were audio-recorded, subsequently transcribed, and compiled into a structured documentary database.

During the data analysis phase, the primary interview transcripts formed the core material for coding and theoretical abstraction. Meanwhile, secondary data were cross-referenced to reduce recall bias, corroborate key events and timelines, and verify financial or performance-related information. In instances where information gaps or ambiguities emerged during the analysis, the researchers conducted follow-up communications via WeChat and email with relevant company personnel to obtain clarifications and confirm key details. This iterative approach ensured the completeness, accuracy, and credibility of the case data. The data collection is shown in

Table 1. Data Sources.

Data Source	Interviewees/Source	Descriptive Statistics	Interview Content/Data Function
Primary Data	Founder and CEO	~150 min, ~21,000 words transcribed	Overview of enterprise development, business model evolution, key milestones, and strategic events
	Head of the AI Research Center	~250 min, ~32,000 words transcribed	Internal and external environmental changes; technological progress during business model innovation
	Head of Strategic Sector	~270 min, ~35,000 words transcribed	Organizational adjustments and external context during business model innovation
	Project Leaders (6 individuals)	~200 min total, ~29,000 words transcribed	Specific implementation strategies and operational measures for business model innovation
Secondary Data	Internal Materials (executive speeches, internal documents, corporate website, etc.)	~15 h of video footage, 24 documents (~190,000 words total)	Triangulation, supplementation, and validation of primary data
	External Materials (news articles, industry reports, etc.)	2 industry reports and 73 web-based sources (~250,000 words total)	Triangulation, supplementation, and validation of primary data

.

3.4. Data Analysis

The data analysis in this study adopts the procedural coding method of grounded theory as proposed by Strauss and Corbin (1990) [95]. The analytical results are presented through structured graphical representations, following the approach outlined by Langley (1999) [96], as illustrated in Figure 3. The coding process consists of three sequential steps. First-order coding seeks to preserve the original language and expressions of the interview respondents as much as possible, thereby capturing the authentic behaviors and experiences of the case enterprises. Second-order coding involves elevating these descriptive codes to a conceptual level, thereby uncovering the underlying theoretical logic and meaning behind the observed behaviors. Finally, these second-order concepts are further distilled into aggregate dimensions, which reveal the broader patterns and relationships among the conceptual categories.

In the first phase of the analysis, we chronologically traced the timeline of business model innovation at Quwan Technology, identifying key milestones and progressively constructing an analytical framework through iterative data coding. This process led to the classification of the firm’s development into three distinct stages. The first stage, the tool product phase, was characterized by the development of applications such as TT Voice, designed to serve as social tools attracting users via diversified social interaction scenarios, thereby achieving initial scalability. The second stage, the social platform phase, saw the platform evolve beyond functional products, integrating social activities such as cultural tourism and esports to cultivate new revenue streams. In the third stage, the ecosystem empowerment phase, the firm expanded its digital human services and collaborated with ecosystem partners to develop an AI-driven platform, ultimately forming an independent technological empowerment system.

In the second phase, we conducted a detailed analysis and abstraction of the data, constructing a data structure diagram to semi-structurally code the contextual conditions, resource actions, AI capabilities, and business model innovations corresponding to each developmental phase. Initially, we applied first-order coding to extract raw constructs from the empirical data, drawing upon established theoretical frameworks. We then identified distinct characteristics of these constructs across the different phases through keyword and pattern analysis, which informed the second-order coding. Based on these coded elements, we employed a “context–action–outcome” analytical logic to explore: (1) how internal and external contextual variables influenced the dynamism of business model evolution; (2) how different AI technology embedding strategies shaped innovation pathways; and (3) how the nature and level of intelligent capabilities changed across phases.

In the third phase, we synthesized the emergent theoretical dimensions to construct a process model of AI-enabled business model innovation. We aggregated relevant second-order codes into three overarching theoretical dimensions, which serve as the foundation of the proposed model. To enhance the robustness and credibility of the findings, we adopted several validation measures. First, we invited external scholars uninvolved in the original data collection to review and discuss the emergent constructs and models, refining any ambiguous components. Second, we maintained ongoing communication with senior executives at Quwan Technology, sharing our preliminary constructs and process model to solicit feedback. This feedback informed the iterative refinement of both the data structure and the overall model.

4. Case Findings

Digital platform enterprises continue to encounter both internal and external challenges in the course of business model innovation driven by AI. These challenges necessitate distinct resource actions to develop AI-related organizational capabilities and realize business model innovation.

In the first stage, platform construction faces a startup dilemma characterized by abundant market opportunities but limited internal resources. To survive the entrepreneurial phase, platform enterprises must adopt a market-oriented approach by developing user-centric tools that cater to personalized needs. AI serves an assistive role within the organization, and freemium strategies for tools and products encourage user participation, enabling initial platform growth.

In the second stage, platform development confronts an expansion dilemma stemming from intense industry competition and increasingly complex user demands. Sustained growth requires a resource-deepening strategy, enabling horizontal expansion across the industry. Simultaneously, the platform must address the intricacies of user needs within the broader social context by leveraging AI-enhanced organizational capabilities, thus facilitating innovation toward a social-platform business model.

In the third stage, platform enterprises face a transformation dilemma, driven by rapid technological advancements and evolving structural demand. As user bases grow and digital reconstruction deepens, personalized demands increase exponentially due to heightened user awareness. Given the limited resources of the organization, platform enterprises must orchestrate coordination-oriented resource orchestration based on demand trajectories. AI-convergent organizational capabilities become essential to enabling ecosystem-wide, synergistic business model innovation.

The following section will examine Quwan Technology’s behavioral processes in employing AI to achieve business model innovation.

4.1. Coupled Requirements Development Tool Platform

During the start-up phase, although the digital platform enterprise experienced growth, the social needs of its user community remained largely unmet. Constrained by limited internal resources, Quwan Technology actively engaged in entry-oriented resource patchwork operations to achieve its market entry objectives. By leveraging AI-assisted organizational capabilities, the company positioned the provision of differentiated functionalities as its core value proposition. This approach enabled value creation through a freemium model based on tool-oriented offerings, significantly amplifying network effects and thereby driving business model innovation. The relevant evidence cited at this stage is shown in

Table 2. Typical examples of a coupled requirements development tool platform.

Aggregated Dimension	Second-Level Theme	First-Level Coding	Typical Evidence Citation
Function Bottlenecks	Rich Market Opportunities	Unmet Demand	“In the early days of the mobile internet, numerous mobile game products emerged. Although these companies provided voice interaction services, their marginal functionality teams failed to generate revenue and incurred costs, resulting in poor voice quality and limited features that did not meet player expectations.”
		Low Technical Requirements	“At that time, intelligent voice technology was regarded as a generic solution. Many cloud service providers offered such services, and by integrating cloud platforms, carriers, devices, and AI-based voice technologies, a baseline level of functionality could be achieved.”
	Lack of Enterprise Resources	Insufficient Digital Resources	“The entire industry remained in an exploratory stage. Data accumulation and technology integration had not yet scaled, and we lacked both sufficient user data and mature AI technologies to support precise optimization.”
		Inexperience in the Market	“We initially lacked experience in AI commercialization. As a result, we prioritized implementing basic intelligent voice functions to build our reputation before expanding to interactive and entertaining features.”
Entry-Oriented Resource Patchwork	Improvised Development	Active Trial and Error	“We adopted a ‘learn by doing’ approach, rapidly releasing features for testing. This iterative process, driven by trial and error, started with a basic real-time voice feature to collect authentic user feedback and identify initial issues.”
		Rapid Iteration	“Rather than seeking a perfect product from the outset, we pursued rapid iteration. With each new piece of feedback, we quickly assessed and translated it into targeted improvements.”
	Patching Together Internal and External Resources	Making Do with Internal Resources	“Initially, our team was small—just six members who multitasked across roles such as coding, product design, and marketing. To avoid ineffective large-scale investment, developers also contributed to algorithm design.”
		Borrowing External Resources	“In-house development alone was insufficient. Therefore, we extensively explored open-source algorithms to test with our data sets.”
AI-Assisted Capabilities	Facilitating Real-Time Interaction	Enhancing Fault Diagnosis Efficiency	“When users reported severe lag at midnight, we implemented a root-cause localization algorithm to accurately diagnose anomalies and timings, thereby improving operational efficiency.”
		Dynamic Load Resource Allocation	“To reduce resource wastage, we introduced a resource recommendation system that used multidimensional metrics and historical data to suggest optimal configurations.”
	Assisting in Technical Visualization	Mass Multi-Modal Data Integration	“We gathered data from user and service endpoints, middleware, and infrastructure. This included links, logs, metrics, and event-based information for effective change tracking.”
		Machine Learning Correlation Analysis	“We used a sequential machine learning algorithm to detect anomalies in system metrics. By integrating our observability framework with historical fault data, we avoided misjudgments caused by periodic or trend-based fluctuations.”
	Supporting Relationship Management	User Relationship Mapping	“We analyzed user social behavior using algorithms and sentiment analysis to identify high-value and at-risk users, strengthening relationships and reducing churn.”
		Fault Prediction Analysis	“We developed a public opinion analysis model to detect various issues—such as recharging failures, voice issues, or access problems. Historically, this model effectively captured most user feedback-related faults.”
Tool Product Business Model Innovation	Differentiated Function Proposition	High-Quality Team Voice Function	“Our initial focus was on social scenarios in team-based competitive games, emphasizing real-time user interaction.”
	Function Value Creation	Voice Communication	“The primary requirement was simple: to enable voice chat and casual conversations.”
		Immediate Interaction	“In designing the product, we prioritized scenarios that allowed instant interaction, whether in games or light engagements, to foster immediate user connection.”
	Freemium Value Capture	Value-Added Services	“When a feature provides sufficient value, it should generate revenue directly. We designed our innovations as independent modules to clearly demonstrate value to users.”

.

At this stage, Quwan Technology adopted a strategy described as “Coupled Requirements Development Tool Platform.” This strategy involved close alignment with user needs and improvisational development as the primary means of market entry. Through the integration of internal and external resources, the company gradually developed AI-assisted organizational capabilities, which in turn supported the iterative development and commercialization of its tool-based products (see Figure 4).

4.1.1. Functional Bottlenecks

In its early development, Quwan Technology operated within a complex market environment. On one hand, there existed a substantial number of unmet user needs, low technological entry barriers, and vast market opportunities. On the other hand, the company faced significant constraints in accumulating digital resources and developing market experience. These contradictory conditions created a functional bottleneck, compelling the company to explore innovative breakthrough paths.

Although several companies had already launched products with similar functionalities, the specific needs of certain user groups remained underrecognized and insufficiently addressed. Particularly in the early stages of the mobile internet era, the market was saturated with mobile game products, and many gaming companies had integrated voice interaction services. However, these services failed to achieve effective commercial operation. As a result, the voice functionalities were often limited in quality and scope, falling short of meeting the increasingly diverse and complex needs of users.

Quwan Technology also faced internal constraints during its initial development phase. In particular, the company had notable limitations in two key areas: digital resource reserves and market operation experience. At the time, intelligent voice technology was already widely regarded as a mainstream tool within the industry, and numerous cloud service providers were capable of delivering related services. From a technical perspective, the integration of cloud services, carrier resources, device performance, and AI voice technology could achieve a basic level of functional performance. Nevertheless, Quwan Technology’s limited quantity and quality of digital resources, along with its relatively underdeveloped market experience, hindered its ability to scale and accelerate growth. As such, the company was required to exert additional effort to overcome these challenges on its path to market entry.

4.1.2. Entry-Oriented Resource Patchwork Strategy

To overcome developmental bottlenecks, Quwan Technology adopted a resource patchwork strategy for market entry. This approach emphasizes improvisational development by integrating internal idle resources with externally accessible ones, enabling the company to respond rapidly to market demands. Through this dynamic resource aggregation, Quwan Technology flexibly allocated personnel, technological assets, and data to establish a product development system aligned with market requirements, thereby laying the groundwork for future business model innovation. The relevant evidence cited at this stage is shown in

Table 3. Typical examples of decoupling precipitate specialized competencies.

Aggregated Dimension	Second-Order Theme	First-Order Coding	Typical Evidence Citation
Professional Bottleneck	Intense Industry Competition	Severe Product Homogeneity	“At the same time, there were approximately 20 similar products in the market. The gaming peripherals ecosystem was highly competitive, with at least five segmented subfields—each comprising around five companies of similar scale.”
		Diverse Competitive Entities	“Various voice-based social platforms evolved into numerous apps and user scenarios, though their core strategies remained largely similar.”
	Complex User Needs	Changes in User Demand	“Initially, user needs were basic—simple chat or casual conversation sufficed. However, with increased internet exposure over the years, users have become more discerning, knowing precisely what they want.”
		Frequent Crisis Events	“TT Voice was mandated to rectify issues related to non-compliant content and the unlawful collection of personal information by the Cyberspace Administration of China. Our application was also ordered to correct content compliance issues by relevant authorities.”
Depth-Oriented Resource Arrangements	Internal Resource Exploration	Focus on Core Resources	“We once explored live streaming, believing it held significant potential. However, despite its novelty, live streaming relies heavily on traffic and is inherently a consumable product. To sustain revenue, a large user base is needed. “
		Extensive Field Trials	“We launched a variety of interest-based scenarios—such as dubbing, murder mystery games, and karaoke—to help users build social networks. This diversity of offerings was part of our efforts to explore rich and varied user engagement formats.”
	External Resource Acquisition	Investment in Technology	“As the company scaled, we faced challenges in management and organization. While we had previously relied on strong operational capabilities, greater emphasis on technology and content development became necessary.”
		Participation in Policy Projects	“Guided by the Guangzhou government’s goal to establish a ‘National E-sports Industry Center,’ Quwan Technology launched strategic initiatives in the e-sports ecosystem and created the TT E-sports brand in 2019.”
AI-Augmented Capabilities	Interaction Experience	Understanding User Intent	“Users often do not explicitly articulate their needs in feedback, preferring to send screenshots instead. Therefore, AI systems must have multimodal recognition capabilities to accurately interpret user intent.”
		User Intelligent Matching	“We employ recommendation systems integrated with natural language processing and sentiment analysis to analyze users’ historical behaviors and real-time interactions. “
	Accelerated Iteration	Efficient Trend Grasping	“Through user behavior and social media sentiment analysis, our development team can more efficiently identify emerging user needs and market trends, allowing for timely feature optimization.”
		Improving Development Efficiency	“AI technologies have significantly accelerated our development cycles. Generative AI models, in particular, have enhanced the efficiency of coding and debugging foundational modules.”
	Contract Governance	Intelligent Content Review	“As business scenarios become more diverse, content governance becomes more complex. To manage content ecosystems and promote positive values,”
		Risk Warning Analysis	“Our security review system, ‘T-Net,’ integrates intent recognition, risk image detection, and audio event analysis.”
Social Platform-Based Business Model Innovation	Differentiated Experience Proposition	Meeting Diverse Interests	“We hope user interactions go beyond fleeting entertainment to form lasting social capital. Therefore, we continuously experiment with new game formats and explore integrating offline activities online to appeal to broader interests.”
	Social Value Creation	Strengthening Emotional Connections	“The core value of social platforms lies in fulfilling emotional needs. Our goal is to build emotionally rich experiences that foster authentic user connections.”
		Establishing Deep Relationships	“When users establish at least three reciprocal friendships and start interacting on their first day, retention rates significantly increase.”
	Community Value-Added Model	Virtual Gifts	“Purchasing and consumption scenarios are seamlessly embedded into our platform’s social and entertainment ecosystems.”

.

As part of this resource integration process, Quwan Technology began by conducting a comprehensive audit of its internal assets, identifying underutilized resources such as existing technical infrastructure, available human capital, and preliminary user data. Concurrently, the company actively pursued external resource supplementation by forming partnerships with entities such as cloud service providers and AI technology firms to access advanced technological capabilities and enriched datasets. By successfully integrating both internal and external resources, Quwan Technology developed foundational intelligent voice functionality that initially met market expectations for voice interaction services.

During this early phase, Quwan Technology also encountered a lack of experience in AI commercialization. To address this challenge, the company prioritized the deployment of core intelligent voice features to enhance its market credibility and user adoption. This strategic focus not only enabled Quwan Technology to establish a foothold in a competitive market environment but also facilitated the accumulation of valuable operational experience and user feedback. These insights became essential inputs for subsequent product iterations and the expansion of feature sets.

4.1.3. Building AI-Assisted Organizational Capabilities

To systematically develop AI-assisted organizational capabilities, Quwan Technology focuses on three core dimensions: enhancing real-time performance, enabling technical visualization, and supporting relational governance.

• Enhancing Real-Time Performance

In this phase, AI-enhanced intelligent interaction capabilities focus on optimizing the responsiveness of platform tools. Real-time interaction is pivotal for improving user experience and increasing platform engagement. Quwan Technology applies intelligent algorithms and big data analytics to streamline message transmission pathways and minimize system response times. Notably, the implementation of AI-powered root cause analysis algorithms enables precise fault detection and rapid problem localization. Furthermore, AI is employed to continuously monitor system load conditions, forecast peak user traffic, and facilitate dynamic, automatic resource allocation. These mechanisms ensure platform stability during high-traffic periods. As a result, TT Voice has achieved marked improvements in message latency and user engagement, thereby strengthening its real-time interactivity and enhancing its competitive positioning in the market.

• Facilitating Technical Visualization

Quwan Technology’s AI-enabled intelligent development capabilities aim to bolster real-time monitoring and system responsiveness through enhanced technical observability. The TT Voice platform integrates large-scale, multimodal datasets to conduct associative analytics using machine learning algorithms. Through the unification of diverse data types—such as logs and metrics—the enterprise constructs a comprehensive data view that enables deeper insight into system performance and user behavior. Machine learning models uncover latent patterns and identify causal relationships between operational factors. According to Quwan’s Head of Technical Operations, the deployment of AI-powered data visualization and sentiment analysis tools has significantly reduced the product failure rate from 38% to 3.8% and shortened average system response time from five minutes to one minute. These advances have considerably improved platform resilience and responsiveness.

• Supporting Relational Governance

AI-enabled intelligent governance capabilities support the effective management of user relationships in digital social environments. Quwan Technology utilizes AI to perform multidimensional analyses of user behavior and social interaction patterns. By constructing user relationship graphs and applying AI algorithms in conjunction with sentiment analysis and whitelist mechanisms, the enterprise can identify both high-value and high-risk users. This facilitates the formulation of personalized feature strategies aimed at enhancing user satisfaction, loyalty, and retention. Additionally, real-time sentiment monitoring on external social media platforms enables the timely identification of emerging topics and user concerns. Trend forecasting and sentiment modeling inform strategic responses, allowing the enterprise to optimize community governance. According to executive leadership, AI-driven relational governance mechanisms have been instrumental in reinforcing social bonds among users, thereby reducing churn and strengthening user engagement.

4.1.4. Business Model Innovation Based on Tool Products

Quwan Technology has developed a business model centered on the core elements of “differentiated functional experience–functional value creation–free value acquisition” during the tool product stage. Functional differentiation is achieved through AI empowerment to address users’ personalized needs. By leveraging a freemium model, the enterprise attracts users with basic functions offered free of charge. Subsequently, the network effects inherent in digital platforms enable user base expansion and enhancement of business value, thereby facilitating the evolution from tool product development to comprehensive business model innovation.

4.2. Decoupling Precipitates Specialized Competencies

During its growth stage, Quwan Technology faced a rapidly evolving market landscape characterized by increasingly complex user needs, expanding AI application scenarios, and intensifying industry competition. In response, the enterprise adopted a deep cultivation orientation, actively engaging in resource integration efforts to enhance its AI-augmented capabilities. Quwan Technology built a social platform-oriented business model innovation system, generating social value through differentiated experiential value propositions and achieving profitability via a community-based value-added model. The relevant evidence cited at this stage is shown in Table 3.

At this stage, Quwan Technology implemented the strategy of “Decoupling to Precipitate Specialized Competencies.” This involved identifying and decoupling specific user groups, abstracting common features from generalized user demands, and executing a comprehensive resource allocation strategy that combined internal resource exploration with external resource acquisition. The enterprise leveraged internal assets—such as user behavior data mining and technology patent reserves—while incorporating external AI technologies to develop modular AI augmentation capabilities. This enabled the creation of differentiated user experiences focused on intelligent interaction and precision services, thereby fulfilling the demand for deep social engagement and supporting innovation in the social platform business model (see Figure 5).

4.2.1. Specialization Bottleneck

The company’s AI-enabled functionalities, initially developed through ad-hoc resource combinations, began to reveal limitations in integration and coherence. As user expectations shifted toward seamless, personalized, and scenario-driven experiences, fragmented capabilities hindered the platform’s ability to deliver consistent value at scale.

The industry landscape during this period was marked by high homogeneity and intense competition. In particular, the gaming ecosystem exhibited significant saturation, with numerous players competing across niche sub-sectors. In the voice-based social platform space, a multitude of applications emerged based on chat and interaction models. Despite diversity in form, these offerings began to converge on similar strategic approaches, leading to severe product homogeneity and escalating market competition.

Evolving user demands further reflected a clear transition from superficial to more sophisticated social needs. In the early phase of internet development, users were content with basic chat functionalities or casual conversations. However, as digital literacy improved, users became more mature and rational, exhibiting significantly higher expectations for precision and personalization in social interactions. Today’s users possess a clearer understanding of their own needs, thereby placing increased demands on platforms’ service capabilities and their ability to deliver tailored experiences. Meanwhile, Quwan Technology also faced mounting regulatory pressures, which introduced a new layer of operational risk. In August 2019, TT Voice was ordered by the Cyberspace Administration of China to rectify compliance issues due to non-compliant content and the illegal collection of personal information. In September 2021, the broader game companion sector underwent a regulatory crackdown, with some products being indefinitely removed from app marketplaces. TT Voice was again required to address content compliance deficiencies. These regulatory developments underscore the growing importance and urgency for platform enterprises to operate in full alignment with legal and ethical standards.

4.2.2. Depth-Oriented Resource Arrangements Strategy

To overcome the specialization bottleneck and support its transition into a scalable and differentiated platform, Quwan Technology adopted an In-depth Resource Arrangement strategy. This strategic shift entailed a dual approach: internal resource exploration and external resource acquisition, enabling a comprehensive restructuring of capabilities and laying the foundation for sustained business model innovation.

Internally, Quwan Technology established a cross-departmental collaboration mechanism integrating its data center, algorithm R&D team, and user operations unit. This system aimed to build an agile, user-centered response mechanism, enabling the company to better sense user demands and quickly iterate solutions. A critical turning point came from an internal review of live broadcasting. While initially effective in attracting users, it proved heavily reliant on traffic and limited in fostering lasting social bonds. Recognizing this, Quwan strategically shifted from large-scale broadcasts to more intimate, small-group or one-on-one interactions, which better supported meaningful social connections. This pivot led to the development of scenario-based features, such as dubbing sessions, murder mystery games, and singing competitions, encouraging user co-creation and role-playing, thereby enriching emotional engagement and enhancing the platform’s social value.

Externally, Quwan expanded its innovation capacity through strategic investment, technological collaboration, and multi-stakeholder engagement. The company invested in content-oriented enterprises to co-develop AI-driven interactive content, gaining access to advanced technologies and industry data. It also aligned with the Guangzhou Municipal Government’s initiative to build a National E-Sports Industry Center, launching the TT E-Sports brand in 2019 and playing a key role in the regional e-sports ecosystem. To integrate academic and policy resources, Quwan partnered with the School of Tourism Management at Sun Yat-sen University, jointly establishing a College Student Internship Base and conducting research on the intersection of e-sports, local culture, and tourism. These collaborations broadened Quwan’s industry reach, enhanced its public image, and strengthened its innovation ecosystem.

4.2.3. Building AI-Augmented Capabilities

At this stage, Quwan Technology has established a competitively advanced AI-enhanced capability system focused on three core dimensions: interaction experience, functional iteration, and contract governance.

• Enhancing Interaction Experience

At this stage, the focus of intelligent interaction capabilities is on enhancing the platform’s social interaction experience through AI technologies [97]. Firstly, understanding user intent [98]. By deeply analyzing user behavior data, preference settings, and interaction history on the platform, AI systems can gain insights into users’ real needs and interests, thereby enabling the accurate recommendation of personalized services. Secondly, intelligent user matching. Utilizing collaborative filtering algorithms and content recommendation algorithms, the platform can efficiently match users within a large user base, creating more targeted social networks. Finally, by enhancing the platform’s interaction experience through AI, users can enjoy more engaging interactive content and establish richer and more meaningful social relationships on digital platforms.

• Accelerating Functional Iteration

At this stage, intelligent development capabilities primarily focus on enhancing information acquisition and increasing technical development efficiency through the use of AI technologies. Enterprises utilize AI technologies to analyze massive datasets, allowing them to quickly identify shifts and dynamics in the social attitudes and preferences of various groups. Through the automation provided by AI algorithms, technical teams can intelligently handle repetitive tasks such as code generation and testing processes, reducing the need for manual intervention. In this way, enterprises can accelerate the delivery of new features and modules, enhancing their responsiveness to market changes.

• Strengthening Contract Governance

At this stage, intelligent governance primarily aims to ensure platform compliance and enhance user trust through the application of AI and blockchain technologies [99]. Enhancing contract governance not only improves the platform’s operational efficiency but also strengthens users’ sense of security and satisfaction. First, intelligent content review utilizes AI-driven systems to monitor and analyze user-generated content in real time, automatically detecting violations. The implementation of such intelligent review mechanisms creates a safer communication environment and boosts users’ confidence in the platform. Quwan Technology has developed systems, including the “T-Net” intelligent content safety review system and the “T-Shield” one-stop content review management platform. Second, risk warning analysis leverages big data and machine learning technologies to identify and assess potential risks in real time, providing timely warnings. By comprehensively analyzing user behavior, transaction data, and external factors, the platform can accurately evaluate risk factors and implement preventive measures before issues arise.

4.2.4. Social Platform-Based Business Model Innovation

Through scenario-based applications of AI capabilities, Quwan Technology has developed a differentiated experience system driven by the integration of “technology and emotion.” By leveraging AI-enabled content ecosystem construction, Quwan Technology has established a value network that synergizes the development of User Generated Content (UGC) and Professionally Generated Content (PGC), thereby promoting knowledge sharing and cultural dissemination. Simultaneously, based on a community value-added model, the enterprise achieves a virtuous cycle of social and commercial value through diversified revenue streams, including virtual goods trading, membership services, and advertising and marketing.

4.3. Building a Loosely Coupled Ecosystem

In the ecosystem empowerment stage, rapid technological advancements have led to increasingly diversified application scenarios for AI. Digital platform enterprises are adopting coordination-oriented resource orchestration strategies, aiming to enhance capabilities in human-computer collaboration, hyper-modular development, and ecosystem governance. AI serves as a critical integrative force in constructing differentiated value propositions and enabling scenario-based, collaborative ecosystem value creation. This, in turn, supports platform enterprises in generating added value through network-based value creation models and driving innovation in business models. The relevant evidence cited at this stage is shown in

Table 4. Typical examples of building a loosely coupled ecosystem.

Aggregate Dimensions	Second-Order Themes	First-Level Coding	Typical Evidence Citation
Model Bottlenecks	Rapid Shifts in Technological Competition	Technological Innovation	“The new wave of AI-driven technological revolution is accelerating, with continuous emergence of technologies and applications such as digital humans.”
		Intense Competition	“A fierce competition—dubbed the ‘thousand-model war’—is unfolding, led by enterprises such as Tencent, Huawei, Alibaba, and JD.”
	Changes in Demand Structure	Shift in Demand Preferences	“Users now expect virtual environments with comprehensive social systems, tailored to their highly individualized needs (‘a thousand faces’), and featuring complex interaction mechanisms.”
		Mismatch Between Demand and Capability	“We believe digital humans will become foundational infrastructure for the digital world.”
Coordination-Oriented Resource Orchestration	Strategic Alliances	Cross-Industry Aggregation	“Leveraging its strong foundation in AI, audio technology, and big data, Quwan Technology actively pursues strategic collaborations with leading industry clients to expand application scenarios for intelligent digital human solutions.”
		Industry-Academia Collaboration	“In April 2023, Quwan Technology and Hong Kong University of Science and Technology (Guangzhou) jointly established an AI laboratory.”
	Innovation Platforms	Innovation Platform Construction	“Quwan Technology’s digital human technology platform offers a comprehensive, one-stop solution that includes scenario development, digital human avatars, and live streaming with digital humans.”
		Emergence of New Business Scenarios	“Application areas include science education, live-streaming retail, and animation gaming. These services have been deployed across various local services on digital network platforms.”
AI-Integrated Capabilities	Human-AI Collaboration	Combination of Human Creativity and AI	“Quwan Technology has developed AI-based music composition technology integrating audio processing, deep learning, big data analytics, and music production.”
		Complementarity of Human Experience and AI Prediction	“To effectively implement large language models (LLMs) within enterprises, these models must first be supplied with high-quality contextual data. Without knowledge of the organization’s operations, processes, or users, the LLM functions as an outsider.”
	Super modular Development	Complementary Technology Modules	“Quwan Technology provides modular voice interaction components on its digital platform. Developers and users can freely combine these components to build personalized services.”
		Integrated Innovation Modules	“Quwan Technology has developed an AI-integrated digital human machine with real-time intelligent recognition, dialogue, perception, and generation capabilities. This includes highly realistic facial expressions, voices, and gestures, supporting multilingual, real-time Q&A.”
	Ecological Governance	AI Resource Pool Construction	“The company has built a comprehensive AI resource pool by integrating various technologies and data assets. This includes short video creation, digital human production, and mobile AI voice systems, offering technical support across sectors such as livestreaming, media, government services, and more.”
		AI Incentive Mechanisms	“The platform incentivizes creators to engage in AI-generated content creation through revenue-sharing models, point-based rewards, and exclusive privileges.”
Synergistic Ecological Business Model Innovation	Differentiated Network Proposition	Smart Technology Inclusivity	“Quwan Technology follows a people-oriented and digitally inclusive philosophy, leveraging its advantages in technology and platform resources to solve social challenges through digital tools.”
	Empowerment Value Creation	Virtual Digital Human Technology Platform	“The company has developed its own virtual digital human generation technology platform, serving as a technical engine to power a diverse matrix of interactive products.”
		Metaverse Applications	“Quwan Technology continues to apply its technologies across multiple metaverse-related industries, including immersive education, virtual healthcare companionship, intelligent manufacturing, entertainment, and e-commerce.”
	Scene-Driven Value Capture	Multi-Scenario Solutions	“Quwan Technology is tailoring digital human solutions to different industries, accelerating deployment in science education, livestreaming retail, and animation gaming.”

.

At this stage, Quwan Technology adopts a strategy of building a loosely coupled ecosystem. Specifically, it strengthens AI-integrated capabilities by forming strategic alliances, developing intelligent innovation platforms, and applying coordination-oriented resource orchestration (see Figure 6).

4.3.1. Model Bottlenecks

Driven by rapid iterations in large-model technologies and the reconfiguration of consumer demand, Quwan Technology now faces a set of emerging innovation bottlenecks.

First, the exponential advancement of large-model technologies has led to frequent shifts in technological paradigms. Industry competition has moved beyond isolated technological breakthroughs toward ecosystem-level rivalry, requiring firms to rapidly integrate heterogeneous technology modules.

Second, the demand structure is undergoing dynamic evolution. User preferences are increasingly oriented toward personalized and scenario-based digital services, making it difficult for traditional standardized product supply models to meet rapidly changing market expectations.

Third, there exists a capability–demand adaptation gap. The current technology architecture lacks sufficient modularity, impeding the efficient allocation of innovation resources among ecosystem participants and restricting the scalable expansion of the value co-creation network.

4.3.2. Coordination-Oriented Resource Orchestration Strategy

To address the innovation challenges arising from paradigm shifts in technology and evolving demand structures, Quwan Technology adopts a coordination-oriented resource orchestration strategy. It promotes enterprise innovation through cross-industry convergence, industry–academia collaboration, innovation platform development, and the incubation of new business models.

First, Quwan Technology has established a three-dimensional strategic alliance by constructing a closed-loop collaboration mechanism that spans “demand orientation—technological research—result transformation.” Leveraging its technological strengths in AI, audio, and big data, Quwan actively forms strategic partnerships with leading enterprises across industries. These collaborations not only deepen its understanding of industry-specific needs but also enable the co-creation of innovative solutions by aligning Quwan’s technological capabilities with real-world business scenarios. For example, Quwan Technology partnered with the Hong Kong University of Science and Technology (Guangzhou) to establish a Joint Artificial Intelligence Laboratory. The lab focuses on cutting-edge areas such as multimodal AIGC and intelligent 3D generation, undertaking multi-level, systematic research. This university–industry collaboration leverages the academic research capabilities and talent of universities to accelerate technological development and application.

Second, by constructing open innovation platforms, Quwan meets the diversified needs of model application while attracting developers and partners to co-create a vibrant intelligent ecosystem. The company has developed a comprehensive digital human technology platform offering end-to-end solutions—including scenario design, digital human modeling, and digital human live broadcasting. This platform integrates Quwan’s expertise in AI, audio processing, and image rendering to deliver full-process services, from digital human creation to application deployment. Guided by its coordination-oriented orchestration strategy, Quwan continues to explore new business domains and expand the application scenarios of its digital human technology. Currently, Quwan’s business spans a range of sectors including science education, live-stream retail, and animation and gaming.

4.3.3. Building AI-Integrated Capabilities

Quwan Technology has established a technically distinctive organizational capability system structured around three core dimensions: human–AI collaboration, hyper-modular development, and ecosystem governance.

• Human–AI Collaboration

At this stage, AI-driven intelligent interaction facilitates the construction of collaborative value networks through effective human–AI collaboration, which plays a pivotal role in enhancing innovation capacity. The first dimension involves the fusion of human creativity with AI capabilities. In content creation and marketing, Quwan has developed AI-based automated composition technologies that integrate audio processing, deep learning, data analytics, and music production. Second, human cognitive abilities complement AI decision-making. Contextual awareness and domain expertise are utilized to interpret and refine AI-generated decisions, thereby improving their reliability and relevance. As Quwan’s Head of R&D observed, the internal deployment of large language models necessitates high-quality contextual input; without a nuanced understanding of enterprise processes and user needs, the model risks behaving like “an outsider.” Thus, only through dynamic interaction and iterative validation between human expertise and AI systems can deep integration be realized, enabling the formation of collaborative innovation networks.

• Hyper-Modular Development

Quwan Technology has adopted a hyper-modular innovation framework, emphasizing the deployment of optimized complementary technology modules and integrated innovation units to support ecosystem enablement. This strategy lowers development thresholds and fosters a self-reinforcing, symbiotic value-creation system. First, given the central role of AI voice and digital human technologies in the enterprise’s technical architecture, modularity is essential for both general-purpose and industry-specific applications. Quwan’s digital platform provides customizable voice interaction modules, enabling developers and users to configure and combine functionalities based on specific requirements, thereby delivering personalized services. Second, integrated innovation modules serve as a cornerstone of Quwan’s architecture, facilitating seamless interoperability across multiple technical domains. By incorporating real-time recognition, smart dialogue systems, intelligent sensing, and generative AI, Quwan enables the holistic development of intelligent solutions. For example, the company developed an AI-powered digital human terminal for China Telecom that integrates facial expression modeling, speech synthesis, motion rendering, and multilingual real-time Q&A capabilities. Through these modular systems, Quwan has built a highly flexible and scalable technological ecosystem.

• Ecosystem Governance Capabilities

In today’s environment—characterized by the wide reach and rapid dissemination capacity of digital platforms—strong ecosystem governance has become increasingly vital. Platform enterprises generate value by empowering ecosystem stakeholders. First, Quwan has constructed a robust AI resource pool that integrates technological components and data assets. This resource pool includes modules for short video generation, intelligent digital humans, and mobile voice interaction, providing comprehensive technical support for global clients in live streaming, media, and public service sectors. Second, the company has established dynamic incentive mechanisms to encourage active participation from diverse ecosystem actors. Quwan customizes rewards based on client type and creator profile, promoting engagement and content contribution. Through smart assessments, adaptive rewards, and gamified point systems, the platform motivates participants to contribute consistently to the generation and curation of AI-driven voice and visual content, thereby sustaining ecosystem vitality.

4.3.4. Collaborative Ecosystem Business Model Innovation

Quwan Technology has advanced its collaborative ecosystem business model by establishing a loosely coupled innovation ecosystem and leveraging hyper-modular technological outputs to reduce entry barriers for industrial applications. This strategy enables small and medium-sized enterprises (SMEs) to access customized AI services at standardized costs, facilitating the cross-level diffusion of technological value and maximizing the advantages of cutting-edge AI models. By promoting inclusive access to intelligent technologies, Quwan empowers ecosystem participants to co-create value and has launched a range of innovative services, including virtual digital human platforms and metaverse applications. Moreover, through the development of scenario-specific solutions tailored to diverse application contexts, the company has effectively identified emerging value opportunities and achieved a strategic transition toward a collaborative, ecosystem-oriented business model.

5. Conclusions

This study takes Quwan Technology as a representative case to explore the dynamic process of business model innovation of digital platform enterprises under the influence of AI. As shown in Figure 7, the main findings are summarized below:

(1)

AI serves as a critical enabler for overcoming stage-specific challenges in digital platform enterprises. At the startup stage, enterprises face the paradox of abundant market opportunities but limited internal resources. AI contributes internally by supporting the development of user-centered tools and enabling freemium or partial-payment strategies that attract early users and facilitate initial platform growth. In the growth stage, as competition intensifies and user demands grow increasingly complex, AI-augmented capabilities help address social complexity and support the shift toward a social platform business model. In the transformation stage, driven by advances in technology and evolving user expectations, AI-integrated capabilities become central to achieving collaborative ecosystem-based business model innovation.

(2)

Resource integration strategies must be dynamically adapted to the developmental stage. During the startup stage, an entry-oriented bricolage strategy is employed, combining available internal and external resources to rapidly meet market demands and develop core functions. In the growth stage, a deep integration strategy is adopted, emphasizing cross-departmental collaboration and strategic partnerships to reconstruct organizational capabilities and address diversified user needs. In the ecosystem empowerment stage, a coordination-oriented strategy is implemented to build cross-industry alliances and open innovation platforms, thereby enabling technological convergence and co-creation of value.

(3)

The deep integration of AI and organizational capabilities is essential to driving business model innovation. In the startup stage, AI-assisted capabilities are developed to enhance real-time responsiveness, facilitate technical visualization, and support relational governance—collectively improving user engagement and system efficiency. During the growth stage, AI-augmented capabilities enhance interaction experience, functional iteration, and compliance governance, contributing to better user experiences and regulatory alignment. In the ecosystem empowerment stage, AI-integrated capabilities span human–AI collaboration, hyper-modular development, and ecosystem governance, laying the foundation for a collaborative innovation ecosystem.

(4)

Business model innovation follows an evolutionary path from tools to platforms to ecosystems. In the startup stage, the business model is characterized by the trajectory of “differentiated functional experience–functional value creation–free value acquisition,” with AI enabling the development of tool-based products. In the growth stage, innovation centers on enhancing community value through a differentiated experience system that integrates technology and emotional engagement. In the ecosystem empowerment stage, the establishment of a loosely coupled innovation ecosystem facilitates a shift toward a collaborative ecosystem business model. This model enables cross-level diffusion of technological value and the realization of multi-stakeholder benefits.

6. Discussion

First, compared with existing studies on AI and business model innovation, this research moves beyond the limitations of one-dimensional classification logics that often characterize analyses of AI-enabled digital platform enterprises [55,100,101]. Specifically, it proposes a four-dimensional framework, ‘context–resource action–organizational capability reconstruction–business model innovation’, to systematically examine this dynamic process. Furthermore, it proposes a coupling strategy model of AI-driven resource action and organizational capability development, revealing how digital platform enterprises leverage AI to drive innovation across different stages of development.

Second, while prior research has primarily adopted singular theoretical perspectives, such as institutional logic, data network effects, or human–machine interaction [23,25,102], this study introduces a stage-based analytical framework encompassing entrepreneurship, growth, and ecosystem empowerment. It identifies a progressive pattern of innovation, challenges each corresponding to distinct resource strategies. These strategies enable the evolution of AI capabilities from auxiliary tools to augmented capabilities, and ultimately to integrative capabilities. This approach breaks away from classification schemes based solely on AI technological maturity or business model types, embedding the interaction between AI and organizational processes into the trajectory of capability building.

Third, by using Quwan Technology, a representative Chinese enterprise, as a case study, the research sheds light on the unique innovation pathways of emerging market digital platform firms operating under regulatory constraints and technological catch-up pressures. This provides a differentiated and context-sensitive perspective that enriches global discourse on digital platform innovation.

However, the study has several limitations. First, as a single-case study, it reveals the dynamic logic of AI-driven business model innovation through a longitudinal analysis of a Chinese digital platform enterprise. The generalizability of its findings may be limited by the industry-specific characteristics of digital social platforms and esports, as well as cultural contextual differences. At the cultural level, Chinese users exhibit a pronounced demand for “emotional resonance” and “group belonging” in social interactions—patterns that may require reconfiguration in Western markets. For instance, the transition from a tool-oriented platform to a socially embedded platform depends heavily on understanding how users construct social capital. While Western users may favor weak-tie relationships grounded in interest-based networks, Asian users often develop strong-tie connections through collaborative, task-oriented interactions. Future research could benefit from comparative multi-case studies across diverse industries (e.g., manufacturing, e-commerce, mobility) and geographic contexts to test and extend the applicability of these findings.

Second, although this study centers on internal organizational capability development and resource coordination, it does not sufficiently explore how the macro-institutional environment—such as national differences in data governance, privacy regulation, and AI ethics—systematically shapes the AI application strategies of digital platform enterprises. For example, Chinese platform firms operate under unique compliance mandates, offering valuable insights for enterprises in highly regulated environments. Conversely, in regions with more relaxed regulatory regimes concerning data sovereignty and algorithmic transparency, innovation trajectories may be more heavily influenced by market forces than by regulatory pressures. Future research could incorporate an institutional theory perspective to better understand how variations in regulatory contexts influence the strategic deployment of AI and the pathways of innovation.

Third, this study does not address the long-term implications of AI technological evolution for the ecosystem governance capabilities of platform enterprises. As AI technologies continue to evolve and iterate, platform ecosystems are simultaneously undergoing dynamic transformation. Future research could adopt technology foresight methodologies to investigate the co-evolution of emerging AI technologies and business model innovation, thereby uncovering how long-term technological trajectories shape platform governance structures and ecosystem dynamics over time.

7. Implications for Theory and Practice

This study makes several key theoretical contributions to the literature on AI, resource orchestration, and business model innovation in digital platform enterprises. First, the study extends and contextualizes resource orchestration theory and organizational capability theory by constructing an integrated theoretical framework that links AI with staged business model innovation. Through the development of a capability-building model and a four-dimensional interactive analytical framework, the research advances our understanding of how digital platform enterprises evolve under the influence of AI. While prior grounded theory research has largely focused on AI’s impact on discrete components of business models, such as value proposition redesign or revenue model adaptation [103,104,105], this study provides a longitudinal, holistic account of Quwan Technology’s transformation from tool-based innovation to collaborative ecosystem co-creation. It thus contributes to unpacking the “black box” of the innovation process—an aspect often overlooked in existing studies.

Second, this research develops a coupling action framework—“resource orchestration strategy–organizational capability building–business model innovation”—that concretizes the interaction between resource orchestration and capability-building processes. This framework positions stage-specific challenges as triggers for strategic adjustment and capability upgrading, revealing a sequential pattern through which firms apply resource bricolage, deep integration, and coordinated orchestration strategies. In doing so, the study goes beyond the conventional binary logic of “technology application → business model transformation” [106], and instead proposes a dynamic, stage-based evolution model for AI capability development.

Third, this study contributes a technology–organization–environment co-evolution perspective, showing how AI capabilities transition from auxiliary tools to core organizational competencies through systemic integration. The proposed three-tiered capability model—comprising auxiliary, augmented, and integrative AI capabilities—offers a refined and operationalized definition of “AI capability building,” a concept that has remained under-specified in prior research. By breaking down capability development into distinct stages and measurable dimensions, the model enables more systematic and predictive research on AI-driven innovation.

Apart from its research implications, this study also provides some important practical contributions by guiding practitioners to effectively implement AI within their digital platform organization and extract value from their business model innovation. First, AI-driven innovation in digital platform enterprises should be understood as a dynamic process of adaptation and co-evolution among dilemmas, resources, capabilities, and business models. At its core, this process requires the deep embedding of AI technologies into the iterative cycles of resource allocation, organizational capability building, and business model evolution. By adopting stage-specific strategic focuses and upgrading capabilities accordingly, platform enterprises can progress from tool-based innovation to ecosystem-level collaboration. Managers of platform enterprises are thus advised to focus on the evolving and localized needs of users, assess the strategic value of AI in context, and formulate differentiated innovation strategies that align AI applications with specific business model trajectories.

Second, resource orchestration strategies must be dynamically tailored to the developmental stage of the platform enterprise. A uniform approach to resource integration is insufficient given the complexity and variability of enterprise growth trajectories. Instead, platforms should construct stage-appropriate, AI-enabled resource strategies that match the firm’s internal capabilities and external opportunities. For policymakers, this suggests the need for targeted policy interventions that support firms based on their growth phase. In addition, governments should encourage ecosystem responsibility among leading enterprises, fostering collaborative governance structures that ensure the inclusive and sustainable development of digital innovation ecosystems.