Latest / 1 More Paper / Generative artificial intelligence and organizational knowledge management: four alternative configurations — Hussinki et al. (2026)
Transcript
- You’re listening to “Generative artificial intelligence and organizational knowledge management: four alternative configurations,” by H. Hussinki, P. Mikalef, and P. Ritala. Published in 2026. ISSN: 1477-8238 (Print) 1477-8246 (Online) Journal homepage: the linked source Generative artificial intelligence and organizational knowledge management: four alternative configurations Henri Hussinki, Patrick Mikalef & Paavo Ritala To cite this article: Henri Hussinki, Patrick Mikalef & Paavo Ritala (17 Jun 2026): Generative artificial intelligence and organizational knowledge management: four alternative configurations, Knowledge Management Research & Practice, DOI: 10.1080/14778238.2026.2689386 © 2026 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. Published online: 17 Jun 2026. Submit your article to this journal Article views: 1856 View related articles View Crossmark data Citing articles: 1 View citing articles RESEARCH ARTICLE Generative artificial intelligence and organizational knowledge management: four alternative configurations a, Patrick Mikalef b,d and Paavo Ritala c,e Henri Hussinki aBusiness School, LUT University, Lahti, Finland; bDepartment of Computer Science, Norwegian University of Science and Technology, Trondheim, Norway; cBusiness School, LUT University, Lappeenranta, Finland; dSINTEF Digital, Department of Technology Management, Trondheim, Norway; eEntrepreneurship and Innovation, Luleå University of Technology, Luleå, Sweden ABSTRACT. This study examines the transformative potential of generative artificial intelligence (GAI) on the knowledge management (KM) systems and capabilities in organizations. GAI enables efficient processing and summarization of an organization’s proprietary data, generating actionable outputs and further insights (i.e. new data) for its knowledge workers. This improved visibility and leverage to organizational data has the potential to establish a better understanding of what an organization knows and to discover and integrate latent knowledge, leading to, e.g. enhanced KM and better-informed, more consistent, and quicker decisions. However, to reap the potential benefits of GAI, organizations must go beyond the mere adoption of this new digital technology. Organizations should first prepare and transform themselves to enhance their readiness for integrating GAI into their knowledge-related processes and work practices. We argue that KM, both its capability and system views, takes a key role in this transformation. Accordingly, we describe how GAI augments organizational KM capabilities and systems, and how these capabilities and systems act as catalysts for successful GAI use and value creation. In essence, this study outlines the steps for successful KM and GAI integration in organizations and suggests alternative starting points and configurations to achieve it. 1. Introduction. Recent industry surveys pinpoint organizations’ soaring investments in rapidly developing digital technologies, including artificial intelligence (AI). These investments enable organizations to be more data-driven, i.e., use insights extracted from data to make better, more consistent, and quicker decisions. As part of this progress, organizations have started taking initial steps to adopt and use generative artificial intelligence (GAI) to, for instance, increase the efficiency of their key processes and augment organizational capabilities, to facilitate ideation and innovation processes, and to reconfigure the knowledge work roles and tasks between humans and AI. GAI refers to new generation AI models (often using “transformer” architectures or image models) designed to process unstructured data, and generate novel outputs, such as text, images, or audio, making them distinct from previous-generation analytical AI models that predict or classify ARTICLE HISTORY Generative artificial intelligence; knowledge management; knowledge management systems; knowledge management capabilities; knowledge-based view; sociotechnical systems . GAI models, and most notably Large Language Models (LLMs, see Ganguli et al., 2022), had their “breakout year” in 2023, which saw organizations adopting GAI regularly in their business functions and placing strong investment plans for AI because of the rapid advances in GAI specifically. Following the initial excitement, organizations have gradually been moving from pilots, experiments, and upskilling towards integration of GAI into their workflows, practices, and products, which is integral for sustained GAI value creation over time. As a general-purpose technology, GAI has countless potential use cases across organizations, lowering barriers to adoption and distinguishing it from more specialized machine learning models. GAI is also increasingly used as an “agentic” technology, involving greater levels of autonomy and adaptation than previous-generation AI models. Consequently, extant literature posits that the emergence of GAI marks a paradigm shift ripe with opportunities for organizations and missing out on those opportunities may severely hurt an organization’s competitiveness or even its chances of survival . However, organizations might not be ready to seize these emerging opportunities due to a lack of suitable capabilities to leverage and create value with GAI. Even though organizations readily possess other information technology (IT) capabilities developed as part of past IT investments and activities, e.g., advanced data analytics and AI, and in big data analytics, they might not be helpful in the era of GAI. While machine learning-based analytical AI requires specialized teams, skills, and development projects, we expect that an efficient leveraging of GAI requires a more overarching transformation, touching upon a variety of aspects of knowledge work, and thus is deeply intertwined with an organization’s knowledge management (KM) capability and KM systems. In this study, we adopt a knowledge-based view of the firm as our theoretical backdrop, which means viewing the firm’s knowledge as the core differentiating aspect of its competitiveness. By doing so, we also follow scholars such as Alaimo and Kallinikos (2022) who have demonstrated how organizational (and inter-organizational) data is fundamentally tied to knowledge. With this data-and-knowledge nexus in mind, we further focus on KM capabilities, i.e., specific organizational capabilities focused on “identifying, capturing, and leveraging the collective knowledge in an organization to help the organization compete” (von Krogh, 1998), which will not be only significantly enhanced by the novel technological capabilities available through GAI, but these amplified organization-specific KM capabilities will enable organizations also to build tailored GAI models for enhanced outcomes and sustained competitive advantage. We augment this theoretical foundation with the sociotechnical systems lens to KM. Sociotechnical KM systems, defined as platforms that integrate technology and organizational infrastructure, organizational culture, intellectual capability, knowledge sharing, decision-making, and strategy crafting, will tremendously benefit from the introduction and integration of GAI and the related digital technologies for navigating data complexities. However, for GAI to have a substantial impact in organizations, it needs to be integrated with a solid foundational KM system that provides access to rich organizational data, thereby opening opportunities for knowledge-based value creation. In this study, we conceptualize a reciprocal association between KM and GAI in sustained knowledge-based value creation. Compared to earlier era KM systems and technologies, GAI stands out by excelling in processing and summarizing unstructured data (i.e., information, text, images) that contains human-bound characteristics (see, e.g., de Haan et al., 2024), helping organizations to tap into a diverse and rich source of knowledge with unprecedented efficiency. Further, GAI can also augment human capabilities in complex problem-solving by exhibiting unsupervised learning capabilities to find new patterns and relationships from the retrieved data, and by opening iterative cycles of conversational insights discovery. Yet, companies also struggle with GAI for several reasons, including challenges around standardization and reliability of outputs – also known as “hallucination”, in getting employees on board with GAI adoption, and in integrating GAI into workflows, products, and services. Ideally, if implemented and embedded in the organization’s KM architecture (both technically and organizationally), GAI can spur previously unnoticed, “latent”, and “hidden” insights for an organization’s knowledge workers, and when such benefits start to materialize, this can lead to further uptake and co-development between KM and GAI. In sum, the emergence of GAI should be seen as a promising opportunity for KM scholars and practitioners, yet this novel topic is currently under-researched. As a starting point for further scholarly and managerial investigation, this study contributes beyond the current literature in the intersection of KM and GAI by developing an integrative and agenda-setting framework that introduces four alternative configurations explaining the varying interplay among KM capabilities, KM systems, and GAI in organizational value creation. This is done by answering the following research question: How can the reciprocal integration of GAI, KM capabilities, and KM systems transform organizational knowledge-based value creation? We first explain the theoretical underpinnings based on a knowledge-based view of the firm and the sociotechnical systems perspective, then provide a theory-inspired elaboration on the potential benefits of GAI-KM integration in organizations. We suggest four alternative, yet not mutually exclusive, approaches as to how GAI and KM can augment each other. To conclude, a preliminary research agenda for GAI and KM is provided to inspire further avenues of work for scholars working in this domain. 2. Theoretical background. 2.1. Knowledge-based view and KM capabilities:. the emerging role of generative AI Knowledge-based view of the firm (KBV, e.g., Grant, 1996; Pereira & Bamel, 2021) provides useful foundations to examine how organizations create value with data and digital technologies. Fundamentally, creation of data, and utilization of data, together with AI, are processes that rely on knowledge, and help create new knowledge. As prior literature at the intersection of management and information systems emphasizes, organizations need human and other organizational knowledge resources to efficiently leverage their data and IT resources to create related value. Relatedly, KBV and its spinoff discussion on KM capabilities quite extensively elaborate on the key role of intangible knowledge resources and their management in organizational value creation, data-based value creation, and IT value creation. One of the main premises of the KBV is that resource heterogeneity and an organization’s competitive advantage are more likely to stem from knowledge than tangible resources, as the former are developed within an organization, whereas the latter nearly always originate outside the organization and are therefore available also to other organizations. In this regard, organizations should quickly proceed beyond mere purchasing of standard GAI technology to unlocking the full potential of their proprietary organizational knowledge through GAI-augmented KM capabilities for acquiring, organizing, and communicating both tacit and explicit knowledge. In practice, such integration involves tailoring GAI to an organization’s proprietary knowledge through approaches such as fine-tuning model parameters on internal corpora, connecting GAI to live organizational repositories via retrieval-augmented generation (RAG), or embedding domain-specific context and taxonomies into prompts. Such approaches differ in their technical complexity and the degree of KM system maturity required, but all share a common prerequisite: the organization must possess well-managed, accessible, and rich knowledge assets. On a broader scale, KM capabilities refer to higher-order routines that enable organizations to create, develop, and modify their knowledge, reorganizing the entire knowledge production system within the organization. Previous literature highlights knowledge processes such as knowledge acquisition, sharing, and application integral to the organization’s KM capability and organizational innovativeness. IT, including GAI, provides KM with novel, continuously improving capabilities for searching and retrieving information, expanding individual-level knowledge, and enabling its application across different organizational tasks. The emergence and anticipated transformative impact of GAI is therefore worth studying more extensively through the lenses of KM capabilities. 2.2. Knowledge management systems and. generative AI: sociotechnical view The literature on sociotechnical KM systems considers IT as a value-added element within such systems, comprising technology infrastructure, organizational infrastructure, organizational culture, knowledge, and people. The sociotechnical systems view emphasizes that strategic value for organizations does not sprout from technology infrastructure but from the dynamic interaction between people and knowledge within a well-developed organizational infrastructure wherein technology is a key enabler. This interaction between technological and social elements helps organizations enhance their KM activities, integrate their knowledge, and develop suitable organizational conditions for knowledge-based value creation. Because of the social embeddedness and complex interactions, KM systems enable competitive differentiation and sustained organizational value creation. Recent organizational scholarship that has adopted the “relational perspective” has highlighted the relevance of studying AI (and GAI) as part of the organization’s relational configurations and task structures that affect both human-AI interaction, but also how individuals work together. GAI unfolds novel opportunities for enhancing an organization’s KM systems and knowledge-based value creation through its data processing and inherent reasoning capabilities, enabling knowledge workers and organizations to gain a better understanding of themselves and their environment, leading to, for instance, improved problem-solving. However, when assessing the potential of AI, one should always understand the role of data and knowledge underlying the operation of those models. Thus, value-creation processes related to GAI are, first and foremost, conditioned by an organization’s KM system, which integrates all organizational knowledge and therefore serves as the primary data source for the GAI model. If the sociotechnical KM system – or parts of it – is challenged or lacking, merely implementing GAI tools will not solve the underlying problems. Organizations should instead focus on enabling and supporting the key interactions within the KM system to ensure acquiring, organizing, and communicating both tacit and explicit knowledge, captured in different forms of organization-specific data, relevant to GAI. Once the KM system-level support, including access to key data, for GAI has been established, the associated benefits may begin to reproduce as improvements in an organization’s KM system – a platform for the organization’s intellectual capability, knowledge sharing, decision-making, and strategy crafting – and eventually improved organizational performance. 3. Integrating generative AI and knowledge. management 3.1. GAI-amplified KM. In GAI-amplified KM (Figure 1), organizations create value with KM capabilities and systems by augmenting them through GAI’s novel technological capabilities, thereby increasing the degree of knowledge-based value creation. In this approach, the underlying logic is that an organization already has well-functioning KM capabilities and systems that deliver value, while the use of GAI enhances the impact of this KM on organizational value creation. Conversely, if the KM foundation is lacking, the introduction of GAI exposes key knowledge-related weaknesses, increasing organizations’ GAI-related risks. Consequently, this approach is topical and particularly suited for organizations with solid KM foundations and the intention to continue leveraging KM capabilities and systems as primary vehicles of value creation, while the function of GAI use is to keep up with technological progress and reap the low-hanging fruits. Think, for example, of the case of Duolingo, which adopted Glean’s GAI-powered search platform to enhance how employees access knowledge distributed across tools like Google Drive, Slack, and Confluence.1 Instead of replacing its KM systems, the solution builds upon them by aggregating and synthesizing existing content into fast, context-aware answers. This has enabled employees to locate relevant information more efficiently, reportedly saving hundreds of hours per month and improving productivity. As suggested in the extant literature, IT and digital technologies offer constantly evolving capabilities to amplify organizational KM. Similarly, organizations creating value with their KM capabilities and systems can now adopt GAI technology that increases their ability to augment, utilize, and draw from those capabilities and systems (Figure 1). Cutting-edge IT resources, such as GAI, can be used to amplify organizational KM capabilities for finding and retrieving knowledge, therefore boosting knowledge work and organizational value creation. However, different from past IT eras, GAI displays advanced traits, such as agentic, learning, and reasoning abilities, making it a promising technology to support the entire knowledge value chain of identifying, capturing, and leveraging the collective organizational knowledge, helping to maximize knowledge-based value creation and competitiveness. Relatedly, the potential benefits of GAI are widespread, materializing from the operational to strategic levels of the organization. The potential benefits of GAI include helping to automate and augment particularly repetitive and simple tasks. However, there is also evidence that GAI can augment humans’ creative and knowledge work processes from ideation to iteration to product and concept development. Further, KM systems might include complex, unstructured (especially text-based) data created by humans or information systems, which GAI can process and summarize, for instance, as presentations and reports. Therefore, GAI has the potential to revolutionize knowledge work by enabling personnel to obtain a quicker, more in-depth view of the organization’s hard-earned institutional knowledge residing in its KM systems. With the emergence of GAI, unstructured data is gaining momentum as a valuable organizational asset, as all accumulated data (especially text-based) can be processed, combined, summarized, and presented with this new technology. Given that GAI involves both “creational” and “conversational” affordances, the users of this technology can both use it to create new data, information, and knowledge (the creational affordance) but also augment the outputs through an interactive process (the conversational affordance). Therefore, GAI can be seen as a new layer in an organization’s KM system, augmenting the organization’s capability to efficiently process and utilize its knowledge that is captured in a variety of unstructured data. In fact, due to this ability, some scholars have suggested that GAI can help to even access organizational tacit knowledge better than previous technologies. Thus, knowledge synthesis, thematic analysis, pattern identification, and the extraction of meaning and understanding from organizational data enable organizations to completely redesign their KM approaches for enhanced value creation. While GAI can substantially amplify the utilization of existing KM capabilities, this mechanism is vulnerable to knowledge quality effects. Indeed, given its probabilistic nature, GAI does not distinguish between high-quality, outdated, biased, or contextually inappropriate knowledge unless such distinctions are encoded in the KM system. As a result, poorly curated repositories or legacy documents may be recombined and surfaced with high confidence, potentially misleading decision makers. From this perspective, GAI is ruthless, potentially widening the gap between those organizations that successfully create knowledge-based value and those that do not. A further risk concerns automation bias, whereby knowledge workers over-trust AI-generated summaries and reduce critical reflection, leading to superficial or homogenized interpretations of organizational knowledge. Addressing these risks requires specific KM mechanisms that function as safeguards against knowledge quality degradation. First, systematic knowledge curation and validation routines, including periodic audits of repository content, help ensure that the data GAI processes reflect the organization’s current state of knowledge rather than outdated or superseded information. Second, metadata governance practices, such as tagging organizational data with provenance indicators, timestamps, confidence levels, and contextual descriptors, enable GAI to weight and prioritize inputs more appropriately, reducing the likelihood of surfacing low-quality or contextually inappropriate knowledge with unwarranted confidence. Third, human-in-the-loop verification workflows, in which domain experts review and validate GAI-generated outputs before they inform organizational decisions, serve as a critical checkpoint against both hallucination and automation bias. Fourth, knowledge deprecation protocols prevent GAI from treating legacy information as current. These KM mechanisms constitute governance practices upon which reliable GAI amplification depends, and their maturity largely determines whether GAI enhances or distorts organizational knowledge work. 3.2. Knowledge management-enabled. customized GAI In KM-enabled customized GAI (Figure 2), the core idea is to draw from the organization’s knowledge assets and systems and develop bespoke GAI solutions. Indeed, KM has a key role in GAI customization, where organization-specific KM capabilities and systems are utilized to build organization-specific GAI, ensuring competitive differentiation and advanced value creation. This approach also suits organizations with a strong existing KM foundation, as it determines the degree of competitive differentiation that can be achieved through bespoke GAI. However, different from GAI-amplified KM (Figure 1), organizations selecting and executing this approach show more willingness to further explore GAI value creation opportunities, perceiving it as a potential central piece through which their KM efforts translate into enhanced organizational value. Think, for example, of the investment bank giant Morgan Stanley, which developed a bespoke GAI assistant for its wealth management division by grounding the system in its own curated internal knowledge base of research reports, policies, and advisory content.2 The tool is designed to generate answers exclusively from firm-approved materials, with direct links to source documents, ensuring accuracy and compliance. By embedding proprietary knowledge into the model’s functioning, the firm enhances the relevance and trustworthiness of AI outputs for financial advisors. Previous literature emphasizes that widely available technological resources (such as GAI) are neither strategic nor a source of competitive advantage, and they should be coupled with intangible knowledge resources to create meaningful value. AI systems are no different in that sense, as they are typically as good as their training data, with the quality and substance of the organizational data fed to AI models often determining the competitive value and differentiation that can be derived from them. Therefore, high-quality data – processed and managed in KM systems and realized as KM capabilities – can be conceptualized as a core driver to organization-specific advantages of GAI (Figure 2). While GAI can theoretically process all of an organization’s unstructured data, it will be more efficient when the input data is properly organized, indexed, and embedded, and when GAI models are pre-trained with this curated data. Indeed, when pre-trained and fine-tuned on organization-specific data and tasks, GAI is highly efficient at what it is designed to do: augmenting human and organizational capabilities for tasks involving natural language understanding and generation. Compared to the earlier technologies optimized for processing structured data and transforming it into information that people can understand, GAI goes beyond that by grasping hold of the unstructured data layer, utilizing its inherent reasoning abilities, and therefore providing organizations with a novel capability to process, report, and understand what they know based on their unstructured data. In other words, the value of GAI lies in its ability to effectively navigate the vast amount of data and synthesize, summarize, and develop relationships that would be impossible to do otherwise due to the sheer volume of data and the computational requirements to process it. In more concrete terms, organizations can customize GAI through several complementary approaches, each requiring different levels of KM system maturity. First, fine-tuning involves retraining a GAI model’s parameters on an organization’s proprietary data, such as internal reports, process documentation, client records, or codified best practices, so that the model internalizes organization-specific concepts and reasoning patterns. Second, retrieval-augmented generation (RAG) connects a GAI model to live organizational repositories at query time, allowing it to retrieve and synthesize relevant knowledge artifacts before generating a response. Unlike fine-tuning, RAG does not require retraining the model and can therefore accommodate frequently updated knowledge bases, but it presupposes that organizational data is well- structured and accessible through the KM system’s technical infrastructure. Third, contextual prompting involves embedding organizational metadata, taxonomies, or domain-specific instructions into GAI prompts, thereby guiding the model’s outputs toward organization-relevant interpretations without modifying the underlying model. In practice, organizations often combine these approaches enabling consultants to retrieve synthesized, context-aware insights from past engagements when advising new clients. What these approaches share is a fundamental dependence on the quality and accessibility of the organizational knowledge base, reinforcing the centrality of KM systems and capabilities in GAI customization. GAI can be viewed as a “general-purpose technology” – i.e., a technology platform that allows for continuous further innovation – that shows potential in many aspects, ranging from operational-level task support to creativity augmentation, to supporting strategic decision-making. However, the potential of GAI for organizations depends heavily on the data upon which it is built and customized. With large volumes of up-to-date data from reliable sources, GAI may reach its potential in organizations, but a lack of access to such data may cause it to “hallucinate”, i.e., to provide its users with misleading, non-existent, and potentially detrimental information. Therefore, to leverage GAI and achieve business value, it’s a good idea to “start with data” and build KM system that covers the key data sources and business processes, and then, build GAI solutions on top of that coherent KM foundation. However, building a KM system is not a plug-and-play effort, but rather a complex integration of knowledge, people, information systems, and organizational infrastructure. For instance, the knowledge required for efficient GAI utilization could be dispersed throughout information systems and humans, so organizations do not only need to integrate GAI with their information systems but also ensure that the knowledge embedded in humans is systematically captured and stored in a manner that enables its use by GAI. Indeed, GAI is not a silver bullet that magically keeps the organization’s data up-to-date and relevant – GAI models instead interpret and draw from the data that already exists and what is input to them during their use. Therefore, both data and GAI models require maintenance and ongoing resourcing to support their ongoing improvement and performance. When changes materialize in the organization or its external environment, they also manifest as changes in key data, prompting a need to address these dynamics in the organization’s GAI model. For instance, when new key data sources emerge, or old ones become obsolete, they should be filtered through the KM system and then either served for or removed from GAI’s disposal. This implies that a KM system supporting GAI implementations should extend across the organization and be collaborative, as the market dynamics and the required adjustments to data and GAI model are more efficiently identified and addressed as a joint effort between GAI experts and context experts. Overall, Figure 2 provides an impression of the potential, highly plausible positive effect of the KM system on GAI, and the related benefits an organization may reap from integrating GAI and the KM system. It implies that a well-functioning KM system establishes suitable conditions for GAI to flourish (Figure 2), while it may also be perceived as a moderator that influences the degree of benefits an organization may achieve from GAI use (as in Figure 1). However, customization of GAI through organizational KM systems introduces both epistemic and governance risks. Fragmented, siloed, or incomplete knowledge bases increase the likelihood of hallucinations, as the model attempts to probabilistically fill the gaps. Knowledge hiding or hoarding behaviors (see, e.g., Issac et al., 2023; Wang & Dong, 2022) become particularly harmful under this mechanism, as missing organizational knowledge is silently replaced by statistically plausible but incorrect outputs. Moreover, customized systems increase maintenance complexity and technical debt, making continuous updating and validation critical boundary conditions. To mitigate these risks, organizations should embed continuous KM practices into the lifecycle of customized GAI systems. These include version control mechanisms that track changes to the underlying knowledge base, periodic knowledge audits conducted by cross-functional review boards to verify that the data feeding GAI models remains accurate and representative, and structured feedback loops through which end-users report inaccuracies or gaps in GAI outputs back to KM custodians. Such practices help organizations detect and correct knowledge base deficiencies before they propagate through GAI-generated outputs, thereby reducing both the incidence and organizational impact of hallucinations in customized systems. 3.3. Knowledge management-enhanced GAI. In KM-enhanced GAI (Figure 3), organizations drive the degree of value derived from GAI by leveraging KM capabilities and systems to make GAI utilization more impactful. Different from first two approaches, the underlying logic here is that organizations have already adopted GAI and started creating value with it (then potentially faced challenges related to, e.g., data quality and/or availability) and are now making strives to improve the organizational conditions to maximize the benefits gained from GAI. Therefore, this approach is suitable for all organizations that have started their value creation journey with GAI and are eager to course correct and optimize their organizational setup for unlocking advanced GAI benefits. Think, for instance, of the increasing use of the new AI-native legal software Legora3 by different law firms. While Legora can provide very helpful assistance with legal cases, document preparation, and data analysis, the law firms themselves will be ultimately responsible to their customers. Therefore, a solid KM system and capabilities are massively important for ensuring compliance and delivery, while GAI systems would be helpful in providing “raw material” to the process. In essence, KM functions here as an organizational conditioning and governance layer for GAI use: it shapes which knowledge sources can be used, how outputs are contextualized and validated, and how organizations balance exploratory use with reliability, accountability, and risk control. GAI technologies can be utilized as general-purpose technologies that draw from the broad training data, available to all their users. The challenge with this approach is that it does not provide a specific differentiation or competitive advantage, as the abilities embedded in GAI models spread across organizations, and also their customers – a phenomenon labelled as technology-enabled democratization. Therefore, to move from generic productivity gains toward sustained and context- sensitive value creation, organizations need KM capabilities and practices, that guide how employees interpret, validate, share, and apply GAI-generated outputs in knowledge-intensive work, the extant literature proposes that KM practices such as KM-oriented supervisory work, knowledge-based human resource management, and employee recognition may help organizations develop a knowledge-friendly culture that supports ideation, knowledge sharing, and knowledge use. It is also highly plausible that within such culture also the use of digital technologies, such as GAI, is encouraged, as they boost knowledge work and organizational value creation. Furthermore, organizations should establish common practices for contextualizing their data to enable its efficacious use, e.g., by developing an organization-specific taxonomy, including descriptive metadata, that helps GAI better comprehend the specific contexts, what the retrieved data comprises, and what the end users’ queries refer to. In this mechanism, taxonomies and metadata are not primarily used to build bespoke GAI systems, but to guide everyday GAI use by helping employees and systems interpret outputs against shared organizational categories, quality standards, and decision contexts. This denotes that the mere cultural readiness for GAI use might not be sufficient for value creation purposes, and highlights the importance of KM system readiness, as has been previously proposed in information systems literature. Therefore, as part of effective management of their KM systems, organizations should develop, implement, and monitor strict, standardized practices for data contextualization to fully leverage the potential of GAI and avoid developing dysfunctional and siloed data structures. Overall, this mechanism highlights that KM capabilities and systems function as control mechanisms for GAI. Specifically, structured validation protocols, such as expert peer review of GAI-generated summaries and recommendations before they enter formal decision-making processes, help organizations maintain output reliability without forgoing GAI’s efficiency advantages. Organizational standards that specify acceptable data sources and minimum quality thresholds for knowledge inputs further contribute to stabilizing GAI outputs. Additionally, ongoing taxonomy maintenance and metadata enrichment ensure that GAI models accurately interpret the data’s organizational context, reducing the risk of contextually inappropriate outputs. However, the balance between control and flexibility is critical; excessive filtering or overly rigid governance structures may constrain exploratory knowledge discovery and reduce GAI’s creative potential, while weak validation routines allow unreliable knowledge to scale rapidly across the organization, eroding trust in both KM system and GAI. The effectiveness of this mechanism is therefore bounded by the organization’s ability to balance liberty and control. The expected outcome is therefore not simply more GAI use, but more reliable, contextually appropriate, and organizationally accountable GAI-enabled knowledge work. 3.4. Generative AI-driven KM systems. Finally, there is an approach we label as GAI-driven KM systems (Figure 4). Unlike in KM-enhanced GAI approach (Figure 3), wherein GAI value creation is merely improved with KM capabilities and systems, this approach gives GAI a more prominent role in building and transforming the organization’s KM systems. This is a suitable approach for organizations that pursue to organize their KM with an “AI-first” approach. Think, for instance, agent-based GAI systems that continuously retrieve external market data and insight, and feed those into the organizational KM system. Such approaches can now be built with the new agentic AI systems that are able to retrieve market and customer intelligence and thus provide novel entry points to how organizational KM systems are developed and maintained, and how they create value. The distinctiveness of this approach lies not merely in GAI adding processing or reasoning capabilities to existing KM activities, but in GAI becoming embedded as the initial architecture and operation logic of the KM system itself. In this mechanism, GAI is used to continuously structure, update, curate, and connect organizational knowledge, thereby reshaping how knowledge flows, repositories, and KM routines are organized. Ideally, such systems serve as sources of competitive differentiation and advantage due to their underlying knowledge complexity and organization-specific features. GAI exhibits the capability “to find patterns and relationships in the data on its own, without being [explicitly] told what to look for”. This ability provides the potential for GAI to make organizational KM systems more intelligent by acting as a continuous “feeder”, “filter”, and “curator” of unstructured knowledge, not only supporting individual knowledge work but also altering the structure, updating routines, and governance of the KM system itself. Thus, whereas GAI-amplified KM concerns the use of GAI to enhance existing KM activities, GAI for intelligent KM systems concerns the use of GAI to redesign the KM system through which organizational knowledge is captured, organized, maintained, and reused. This is possible through GAI’s ability to perform a wide array of language-related tasks, such as text completion, summarization, translation, and reasoning. In addition, GAI is quite easy to access and user-friendly, by virtue of its conversational affordances, as anyone who can type questions can also prompt GAI, at least to some extent. Therefore, GAI’s unsupervised (or semi-supervised) learning ability, coupled with its ease of use, makes it a technology that an entire organization can quite quickly engage with, resulting in a rapidly growing number of use cases, which allows an organization to learn quickly about how this new technology could or could not contribute to its goals. The above-discussed accessibility features of GAI are coupled with the autonomous and agentic features. This means that, ideally, GAI models can evolve to become broader search engines for intra- and extra-organizational knowledge, curated as part of KM systems. Among the approaches discussed thus far, this approach, where GAI fundamentally transforms KM systems and related value creation, is by far the least understood one. Yet, we believe that in the future this pathway will become increasingly prominent as GAI systems develop toward more intelligence, autonomy, error control, and coherence. However, when GAI actively reshapes KM systems, it can expose structural weaknesses such as data fragmentation, redundancy, and inconsistent governance. Poorly designed KM system architectures may see these deficiencies amplified rather than resolved. Additionally, uncontrolled experimentation with GAI can lead to shadow systems, compliance risks, and loss of institutional coherence. This mechanism is bounded by the organization’s ability to align AI experimentation with KM governance. 3.5. Summary of the GAI-KM approaches and an. agenda for future research In sum, we propose four interlinked yet analytically distinct value creation mechanisms where KM and GAI are integrated to bolster each other: GAI-Amplified Knowledge Management, where organizational KM capabilities and systems are augmented with the introduction of the novel technological capabilities of GAI, driving the degree of value derived from organizational knowledge resources. Knowledge Management-Enabled Customized GAI, where organization-specific KM capabilities and systems are utilized to build customized GAI, ensuring competitive differentiation and value creation through establishing suitable conditions for GAI to flourish. Knowledge Management-Enhanced GAI, where organizations creating value with GAI utilize KM capabilities and systems to drive the degree of benefits reaped from GAI. GAI-driven Knowledge Management Systems, where the technological features of GAI are embedded in the organization-specific KM systems, allowing competitive differentiation and advanced GAI value creation. The four approaches developed in the preceding subsections offer a coherent explanation of how, why, and under what conditions GAI and KM interact to shape organizational value creation. While analytically distinct, the approaches are mutually reinforcing and operate across different levels of analysis, ranging from individual knowledge work to system-level and organizational transformation. The approaches presented above highlight that GAI should not be understood merely as a new IT artifact, but as a sociotechnical phenomenon whose value depends critically on existing KM capabilities and systems, governance practices, and system/organizational architectures. They also emphasize that GAI-related benefits and risks are inseparable, or in other words, the same mechanisms that enable insight generation may also amplify epistemic risks and produce governance challenges if KM foundations are weak. Rather than providing an exhaustive list of future research opportunities, we outline a focused research agenda that builds directly on the causal mechanisms theorized in this study (Table 1). The agenda concentrates on research questions and theoretical perspectives that are most strongly substantiated by our framework and that can be pursued cumulatively in future work. The research agenda summarized above (Table 1) is organized around the four causal value creation mechanisms developed in this study, with each row representing a focused stream of inquiry that follows directly from the theoretical arguments advanced in this study. The first row, GAI-amplified KM, concentrates on how GAI enhances existing KM capabilities and systems by enabling more efficient synthesis and application of organizational knowledge. Research in this stream focuses on how GAI could reshape expert-level knowledge work and thereby unlock benefits beyond efficiency gains. The second row, KM-enabled customized GAI, highlights the role of organizational KM in shaping the quality and usefulness of organization-specific GAI applications. This stream focuses on how knowledge capture, structuring, organizing, and governance practices condition the performance and reliability of GAI systems, rather than treating GAI outcomes solely as technology driven. The third row, KM-enhanced GAI, perceives KM as a set of mechanisms through which organizations establish the balance between liberty and control in knowledge and GAI use, in order to maximize benefits and/or minimize risks. Research in this stream examines how, e.g., governance models and routines, validation practices, and knowledge protection influence the sustained use of GAI in knowledge-intensive work. The fourth row, GAI-driven KM systems, addresses how GAI contributes to the building and reconfiguration of KM system architectures and knowledge flows. This stream encourages research on how GAI can be used to build better, more adaptable, and potentially more agentic and autonomous KM systems. As elaborated in this study, the nexus between GAI and KM is theoretically promising, but the practical implications should also be considered. To this end, we emphasize that at this stage qualitative research, such as in-depth case studies, should be conducted to find out and determine whether the four value creation mechanisms materialize in practice. Multiple-case studies could also embark to explore why similar GAI technologies can yield vastly different outcomes across organizations and help surface how different organizational or cultural contexts might affect successful KM and GAI integration in organizations. Furthermore, as different aspects of KM, GAI, and organizational value creation can be (or have already been) operationalized as survey instruments, future research should seek also for more generalizable evidence on the association between KM, GAI, and organizational value creation through quantitative and statistical methods, such as PLS-SEM. Moreover, configurational methods, such as Fuzzy-set Qualitative Comparative Analysis (fsQCA) could be employed to explore the best KM-GAI configurations for value creation in different contexts, and longitudinal studies could help explain the effect of, e.g., knowledge governance practices on GAI output, evolving trust dynamics, and organizational value creation over time. 4. Discussion and conclusion. The objective of this study was to explore how GAI could transform organizational KM and knowledge work, and what configurations could enable this synergy and integration. Through the lenses of the knowledge-based view and sociotechnical systems, we developed an integrative framework, theorizing the amplifying role of GAI in the association between KM and organizational value creation, as well as the foundational impact of KM on organizational GAI value creation. Going beyond the current academic discussion that has examined, e.g., the impact of GAI on organizational knowledge creation, storage, transfer, and application (i.e., KM processes, see Alavi et al., 2024) this study introduced four alternative configurations explaining how the interplay between KM capabilities, KM systems, and GAI could translate into enhanced organizational values. On the one hand, technological affordances available through GAI enable more comprehensive and efficient leveraging of organizational knowledge to enhance KM-based value creation (Approach 1) or help capture organization-specific knowledge within proprietary GAI models (Approach 2). On the other hand, the roles of KM capability and KM systems are pronounced in reaping the potential benefits of GAI, either by ensuring feasible knowledge is available to contextualize GAI use (Approach 3), or by providing the interface for organizational use of more intelligent KM systems driven by novel affordances of GAI (Approach 4). We therefore propose that there is a nuanced, reciprocal, and multifaceted relationship between GAI and KM, providing organizations with long-term opportunities that are only partially emerging. 4.1. Research implications. Our main argument is that to reap substantial benefits from GAI, organizations must align this rapidly evolving technology with a coherent KM foundation that encompasses both KM capabilities and systems. As suggested across our four approaches (see Figures 1–4), we argue that the GAI-KM integration can occur through multiple pathways. Our main findings provide several implications for research, which we discuss next. First, along the lines of the KBV, organization-specific combinations of knowledge resources lay the foundations for competitive advantage. Fundamentally, since the technology alone cannot provide a lasting advantage due to its wide availability and general-purpose features, and since data, in itself, is not valuable in isolation to the knowledge around its creation and use, we argue that focusing on KM capabilities is of particular relevance in the context of GAI-based competitive advantages. This opens up an important avenue for KM scholars for further conceptual and empirical inquiry, examining how KM capabilities interact with GAI in different ways. The four configurations identified in this paper provide a useful starting point for such inquiry. Second, the sociotechnical perspective provides a further perspective on how to build organizational advantages around data and algorithms that are both useful and unique (see also, Kemp, 2024). Given the many potential synergies between KM (systems and capabilities) and GAI, organizations need to establish favorable conditions for knowledge flows, especially between their GAI experts and domain experts. In essence, domain experts know which sources contain relevant and valuable data, and GAI experts (and data scientists, for instance) understand how to establish a KM system infrastructure that enables this data to be served for GAI, both playing key roles in successful GAI adoption and use. Conversely, a flawed KM system (architecture) can completely undermine the efforts to leverage GAI for value creation, emphasizing the underlying problems rather than providing the crucial foundation. All this suggests new types of thinking and theorizing over what constitutes a “KM system” and a “KM capability”, and to what extent these are sociotechnical. Indeed, given the increasingly relational role of digital technologies, we need to understand better organizational design and practice as it comes to agency, actorhood, task structures, and incentives, to name a few (for discussion, see, e.g., Huysman, 2025; Ramaul et al., 2026). Thus, we call for better integration of the KM literature with the broader organizational literature and information systems scholarship. Third, given the increasing integration of GAI technologies with organizational knowledge, our study calls for a better understanding of the specific affordances and limitations of GAI technologies. The conversational and creational affordances have demonstrated the increasing malleability and capacity of GAI technologies in organizational contexts, especially in knowledge work. However, knowledge work also involves new challenges, such as epistemic challenges regarding the validity and truthfulness of GAI outputs (see, e.g., Hannigan et al., 2024). Therefore, utilizing any of the four approaches discussed in the current study not only provide new opportunities, but they also insert new complexities and uncertainties that call for better knowledge, data, and AI governance. For instance, there is increasing discussion about “human-in-the-loop” systems and roles as GAI becomes more pervasive. From another perspective, GAI systems are becoming more agentic and autonomous, requiring new types of governance, guardrails, and interfaces within organizations. 4.2. Practical recommendations. Aligning KM with GAI is an urgent and growing concern for organizations of different types and sizes. When approached deliberately, taking into account strategic objectives, existing KM capabilities and systems, and the distinctive affordances and risks of GAI, organizations may unlock significant value in knowledge work and decision-making. However, these benefits are unlikely to materialize if organizations adopt a technology-first logic, assuming that procuring GAI tools alone will automatically transform KM and organizational value creation. Instead, successful integration requires a staged, sociotechnical approach that places KM foundations, experimentation, and evaluation at its core. 4.2.1. First: start with the knowledge management. foundation As a first step, organizations should systematically assess the maturity of their KM capabilities and the current state of their KM system. This includes evaluating how knowledge is captured, stored, curated, shared, and governed across the organization. Access to skilled KM experts is essential at this stage, as they play a critical role in establishing human-centered data pipelines that ensure the accuracy, relevance, and contextual richness of the knowledge made available to GAI models. Importantly, organizations must account for human behaviors that directly shape knowledge quality, such as inconsistent data entry practices, reliance on informal storage solutions, and knowledge hoarding or hiding. Understanding these tendencies enables targeted interventions – ranging from revised workflows and incentives to clearer accountability – that reduce bottlenecks and support change management. Organization-wide training on data and knowledge value creation is therefore highly recommended, emphasizing employees’ dual role as both users and contributors of organizational knowledge. Without these foundations, GAI risks amplifying existing deficiencies rather than generating meaningful improvements. This recommendation follows directly from Mechanisms 2 and 3, but in different ways. For Mechanism 2, KM foundations provide the structured and proprietary knowledge base needed for customized GAI. For Mechanism 3, they provide the validation, contextualization, and governance routines needed to make already adopted GAI reliable in organizational use. Without a mature KM foundation, the data inputs that drive GAI performance remain fragmented and ungoverned, limiting the organization’s capacity to move beyond generic productivity gains toward differentiated, knowledge-based value creation. 4.2.2. Second: experiment bottom-up and identify. the “jagged frontier” GAI differs from many earlier KM-related technologies in its experimental nature and low barriers to initial use. Unlike expert systems or traditional machine learning applications, which typically require extensive upfront modeling and development, GAI allows organizations to explore potential value through trial-and-error and incremental scaling. Organizations should actively encourage bottom-up experimentation within business units to identify where GAI adds real value in knowledge-intensive tasks. Such experimentation enables organizations to locate the so-called “jagged frontier”, where performance gains vary substantially across tasks and contexts. A decentralized IT and GAI governance approach is therefore advisable in early stages, as learning and innovation often emerge faster within business units than through centralized control. At the same time, organizations should avoid fragmentation by maintaining shared KM standards and coordination mechanisms. Successful experiments can then be translated into scalable business cases, generating short- or long-term competitive advantages. This recommendation aligns with Mechanism 1 (Figure 1), which theorizes that GAI amplifies existing KM capabilities and systems as well as knowledge work. The emphasis here is on discovering where GAI most strongly amplifies existing knowledge work, rather than on customizing GAI systems or redesigning KM architectures. Bottom-up experimentation enables organizations to discover empirically where this amplification effect is strongest, thereby informing subsequent decisions about scaling and governance. 4.2.3. Third: build the information architecture and. secure technological expertise To move beyond experimentation, organizations must invest in information architecture that reliably supports GAI utilization. This includes well-defined data pipelines, commonly agreed storage locations, and continuously maintained taxonomies and metadata structures. Such infrastructure helps ensure that customized GAI systems are trained, grounded, or queried using relevant, up-to-date, and contextually meaningful organizational knowledge, rather than recycling outdated, incomplete, or misleading information. Sufficient technological expertise is required to tailor GAI solutions to organizational needs and to integrate them with existing KM system. While core KM architecture and governance capabilities should typically be developed in-house, given their organization-specific and long-term relevance, more specialized GAI development tasks can, where appropriate, be partially outsourced. The key managerial challenge lies in coordinating KM and AI expertise rather than treating them as separate domains. This recommendation operationalizes the KM system foundations theorized in Mechanism 2 (Figure 2), which argues that well-structured, governed, and accessible organizational knowledge is a prerequisite for building customized GAI that delivers competitive differentiation. Investing in information architecture is, in essence, investing in the mediating pathway through which KM enables organization-specific GAI value creation. 4.2.4. Fourth: leverage agentic AI to maintain and. enrich the knowledge base As KM – GAI integration matures, organizations can increasingly leverage AI agents to assist in maintaining and enriching the organizational knowledge base. These AI agents can participate in general-purpose work nearly unsupervised and accomplish various types of tasks end-to-end, including supporting knowledge classification, summarization, synthesis, and detection of gaps or redundancies in existing repositories. When appropriately governed, AI agents can help keep knowledge assets current and actionable, reducing manual maintenance burdens while enhancing system intelligence. However, autonomy should be introduced cautiously. Clear oversight mechanisms are required to ensure that AI-supported knowledge updates remain aligned with organizational standards, strategic priorities, and epistemic safeguards. Unlike Mechanism 3, where KM improves the organizational use of GAI, this recommendation concerns the use of GAI to actively reshape, maintain, and enrich the KM system itself. This recommendation corresponds to Mechanism 4 (Figure 4), which conceptualizes GAI as a driver of more intelligent KM systems. By deploying AI agents that continuously classify, curate, and enrich organizational knowledge, organizations enact the feedback loop through which GAI contributes to the long-term reconfiguration of KM system architectures and knowledge flows. 4.2.5. Finally: measure and concretize outcomes. Organizations should explicitly measure and monitor the outcomes of KM – GAI integration. While financial metrics such as return on investment (ROI) may be useful, they should not be the sole indicators of success. Complementary measures, such as improvements in decision quality, time savings, knowledge reuse rates, user confidence, and error reduction, are often more informative in knowledge-intensive contexts. Making these outcomes visible helps sustain managerial commitment, guide further investment decisions, and support continuous learning and adjustment. As the theoretical framework developed in this study suggests, the benefits of GAI – KM integration are mechanism-dependent and context-sensitive. Amplification effects (Mechanism 1) may manifest as efficiency gains, while customization benefits (Mechanism 2) may be better captured through quality and differentiation metrics. For Mechanism 3, relevant indicators include output reliability, reduced errors, improved compliance, user trust, and the quality of decisions informed by GAI outputs. For Mechanism 4, indicators should capture KM system adaptability, knowledge-base currency, and reductions in fragmentation or redundancy. Systematic measurement is therefore essential not only for justifying investment but for understanding which integration pathways generate the most value in a given organizational context. 5. Conclusion. This study explored the potential key role of an organization’s KM foundation for leveraging GAI for value creation, as well as the transformative potential GAI entails for KM capabilities and systems. GAI models have started a new era where organizational knowledge, especially unstructured data, is processed, combined, summarized, and utilized with unprecedented efficiency and novel reasoning abilities to drive key processes and decision-making. This underscores the need for organizations to reassess the robustness and applicability of their KM capabilities and systems for successful GAI integration. We propose four theoretically separate but synergistic value creation mechanisms to demonstrate and explain how KM and GAI may interact to generate organizational outcomes. Coupled with a research agenda and practical recommendations, this study facilitates further scientific exploration of this topic and drives organizational change and success. Notes 1. the linked source. /duolingo. 2. the linked source. milestone-in-innovation-journey-with-openai. 3. the linked source. CRediT: Henri Hussinki: Conceptualization, Investigation, Project administration, Supervision, Visualization, Writing – original draft, Writing – review & editing; Patrick Mikalef: Conceptualization, Investigation, Writing – original draft, Writing – review & editing; Paavo Ritala: Conceptualization, Investigation, Writing – original draft, Writing – review & editing. Disclosure statement No potential conflict of interest was reported by the author(s). Funding The author(s) reported that there is no funding associated with the work featured in this article. Data availability statement Data sharing is not applicable to this article as no new data were created or analyzed in this study. Author contributions