Document Type : Research Paper

Authors

• Atoosa Ebrahimi Shah Abadi ¹
• Jahangir Yadollahi Farsi ²
• Niloofar Nobari ³

¹ Ph.D. student in Entrepreneurship, Faculty of Entrepreneurship, University of Tehran, Tehran, Iran

² Full Professor, Faculty of Entrepreneurship, University of Tehran, Tehran, Iran

³ Assistant Professor, Faculty of Entrepreneurship, University of Tehran, Tehran, Iran

Abstract

The rapid rise of AI-based platforms demands a clearer grasp of their technical and managerial aspects. To address this, this study compares three major types (transactional, innovation, and integrated) using the ADO (Antecedents-Decisions-Outcomes) framework. Through a meta-synthesis of 70 articles (2015–2025) and systematic coding, it shows that transactional platforms use microservices and open APIs for fast data exchange; innovation platforms rely on deep learning and open collaboration; and integrated platforms focus on unified data governance. Cloud infrastructure, security, and transparent governance link them all, but each type follows unique strategies shaping decisions and outcomes for firms, users, and ecosystems. This work clarifies how shared technological foundations diverge into distinct paths, offering insights for managers, policymakers, and future researchers to design robust, tailored governance for evolving AI ecosystems.
Introduction
Artificial Intelligence has become the beating heart of digital economies, redefining industries through machine learning, big data analytics, IoT integration, and cloud solutions. AI-driven platforms, ranging from data marketplaces to open innovation hubs, play a central role in reshaping value chains and amplifying agility and productivity. Despite this prominence, prior literature remains fragmented, focusing on technical specifications or isolated managerial aspects, without providing a comprehensive framework that interlinks their foundations, strategic choices, and impacts. The scholarly literature on AI-based digital platforms, despite rapid expansion, lacks an integrated, comparative framework that systematically distinguishes platform types and clarifies their technological and governance pathways. Building on this categorization, AI platforms can be grouped into three archetypes: Transactional platforms, which enable exchanges of goods, services, or data through modular architectures and open APIs; Innovation platforms, which nurture co-creation through open collaboration and advanced processing frameworks; and Integrated platforms, which fuse transactional and innovation capabilities with robust data governance and ecosystem-wide orchestration. Yet, much of the extant research remains confined to technological facets, such as machine learning pipelines, or to managerial dimensions like trust, compliance, and governance protocols. To address this gap, the present study employs the ADO framework to map how technological prerequisites, strategic design choices, and multi-level outcomes interlock across these platform types. By combining Cusumano’s typology with the ADO lens, this research proposes a structured comparative perspective: Transactional platforms optimize rapid, low-cost exchanges; Innovation platforms fuel open co-creation; and Integrated platforms consolidate efficiency through standardized, unified data governance. Through this integrated approach, the study illuminates the causal pathways that link technological foundations to managerial choices and, ultimately, to the societal and organizational impacts of AI-driven platforms.
Research Question
The overarching research question thus asks: How do the antecedents, key decisions, and outcomes of AI-based platforms differ and intersect across transactional, innovation, and integrated models?
Literature Review
The existing literature on AI-based platforms reveals that these platforms, through the fusion of machine learning, big data, and cloud computing, play a pivotal role in facilitating data exchange, open innovation, and data governance. According to the well-known classification by Cusumano et al. (2020), these platforms can be grouped into three main categories:
Transactional: They serve as intermediary infrastructures for the rapid exchange of data and services by leveraging modular architecture, open APIs, and cloud processing.
Innovation: They offer an open environment for co-creation of products and deep learning models, relying on GPU/TPU capabilities and developer networks.
Integrated (Hybrid): They represent a combined structure that integrates data governance while simultaneously managing transactions and innovation at an industrial scale.
In parallel with this classification, the ADO framework acts as a causal model linking three fundamental layers:
Antecedents: Core technologies such as cloud computing and machine learning, skilled human capital, and a regulatory playing field that aligns ethics with the law.
Decisions: Technical architecture, a seamless user experience, innovation policies, data regulations, and standards that shape the identity of these platforms.
Outcomes: Ranging from profitability and organizational efficiency to user satisfaction and engagement, and ultimately, far-reaching economic and ecosystem effects that redraw industrial boundaries.
Combining this framework with Cusumano’s model offers a fresh analytical pathway for comparative investigation of the structural, managerial, and outcome dimensions of AI platforms, providing a coherent foundation for future research in designing and governing this digital ecosystem.
Methodology
To address this question, a systematic meta-synthesis approach was employed. Relevant articles were sourced through comprehensive database searches (Google Scholar, Scopus, and Science Direct) covering 2015–2025. After rigorous screening, seventy high-quality articles were selected based on relevance to AI platforms, user-centric applications, and compatibility with the ADO structure. Data were coded thematically in three stages: open coding (identifying raw concepts per platform type), axial coding (categorizing into antecedents, decisions, and outcomes), and selective coding (developing the integrated conceptual framework). Inter-coder reliability was validated through Cohen’s Kappa, which reached a robust score of 0.80.
Results and Discussion
The meta-synthesis results reveal that AI-based platforms can be grouped into three distinct yet interconnected types: transactional, innovation, and integrated. Across seventy reviewed studies, common technological antecedents emerged—cloud computing, robust data governance, and open API frameworks are foundational for all three. Transactional platforms use microservice architectures and clear Service-Level Agreements to lower costs and build trust, especially in fast-paced data exchanges like financial transactions. Innovation platforms stand out for leveraging deep learning modules and open collaboration to co-create new products and expand complementary markets. Integrated platforms emphasize large-scale data orchestration and compliance, aligning diverse systems through unified governance to improve organizational efficiency and national data ecosystems. The combined findings and discussion show that, despite shared foundations, each platform type follows unique decision pathways: cost efficiency for transactional, collaborative agility for innovation, and cohesive governance for integrated. This validates the ADO framework’s capacity to trace how technological and institutional conditions shape strategic design and produce layered outcomes. These insights help managers align architecture, policies, and stakeholder roles to balance performance, security, and innovation in complex AI ecosystems.
Conclusion
In summary, this study clarifies how the intertwined dimensions of technology, governance, and user-centric design shape the trajectories of transactional, innovation, and integrated AI-based platforms. The proposed ADO-based synthesis does more than classify platform archetypes; it demonstrates a coherent causal pathway from technological and organizational prerequisites, through architectural and managerial decisions, toward tangible outcomes spanning operational, user, social, and economic domains. Practically, the findings deliver valuable guidance for managers, platform developers, and policymakers. Transactional platform managers should emphasize transparent contracts, modular microservice design, and robust security layers to build user trust and lower transaction costs. For innovation-driven ecosystems, fostering open collaborations, protecting intellectual property rights, and investing in scalable GPU/TPU infrastructure are key to sustaining rapid innovation cycles. Integrated platform leaders must adopt comprehensive data governance frameworks, aligning diverse standards and compliance policies to harmonize data flows across industries. From a policy standpoint, governments and regulators are urged to craft clear guidelines for secure data exchanges, promote open innovation incentives, and support national investments in cloud and AI infrastructure to maintain global competitiveness. Future research should extend this conceptual foundation through mixed-method studies, localized models, and robust quantitative testing to validate the ADO pathways across varying cultural and industrial contexts. By bridging theory and practice, this work contributes a systematic lens for understanding and steering the complex dynamics of AI-based platform ecosystems, charting a path toward more resilient, innovative, and ethically governed digital futures.

Keywords

• AI-based platforms
• ADO framework
• transactional platform
• innovation platform
• integrated platform
• comparative analysis

Main Subjects

• modeling and simulation