News

Subject: Information System Science

Title: AI-Augmented Data Stewardship: Evaluating a Supervised GenAI Workflow and Diagnosing Master Data Inconsistencies in the FMCG context

Abstract: 

Multinational enterprises in the Fast-Moving Consumer Goods (FMCG) sector increasingly manage product master data across heterogeneous enterprise systems. Structural differences across these fragmented environments produce incompatible records, creating a manual reconciliation burden that is slow and difficult to scale. Recurring inconsistencies persist because the underlying organizational and process conditions remain structurally unchanged.

This study investigates whether a supervised Generative AI (GenAI)-assisted workflow improves reconciliation efficiency and representational consistency over manual processes, and examines the root factors causing recurring data errors. Task-Technology Fit (TTF) and data quality research and frameworks provide the interpretive lens.

Using an embedded mixed-method single-case design, the quantitative component applies a quasi-experimental comparison of two workflow conditions across a dataset of product records from a real operational harmonization project. The qualitative component draws on eight semi-structured stakeholder interviews analyzed through thematic analysis.

The findings reveal that the supervised GenAI workflow substantially reduced total processing time and improved structured output quality, while achieving near-equivalent attribute-level correctness to manual methods. Human review remained central to the design, with the majority of model outputs accepted without modification. The qualitative component identified four upstream themes, covering governance fragmentation, manual transfer dependencies, data entry disconnection from downstream consequences, and structural inheritance from legacy system decisions, that explain why similar inconsistencies reoccur regardless of downstream efficiency gains.

Taken together, the study demonstrates that supervised GenAI augmentation improves the efficiency and structural consistency of stewardship work, but operates entirely at the correction layer. Addressing recurrence requires upstream governance intervention that the workflow alone cannot provide. The study extends TTF to a performance-diagnostic application in governance-constrained enterprise stewardship and offers transferable insights for FMCG organizations operating in comparable multi-system environments.

Keywords: master data management, data stewardship, generative AI, Task-Technology Fit, human-in-the-loop, data quality, prompt engineering, FMCG, information quality, data governance.

Subject: Information System Science

Title: Transitioning to Zero-Touch Edge Networks: An Architectural and Strategic Roadmap for a Telecom Connectivity Management Platform

Abstract: 

Telecommunications operators managing the residential network edge face a structural mismatch: the proliferation of connected devices has made reactive, manual troubleshooting economically unsustainable, while incumbent Cloud Monitoring Solutions remain observational by design — able to surface service degradations but not to resolve them autonomously. This thesis asks what architectural, data, and operational capabilities such a platform must acquire to move from reactive, human-in-the-loop troubleshooting toward proactive, Zero-Touch management aligned with the ETSI Zero-touch network and Service Management (ZSM) Level 4 autonomy criteria.

The research adopts a hybrid Design Science strategy that combines an embedded single-case study of a commercial connectivity-management platform with the design and evaluation of a concrete artefact. Its empirical basis comprises two interview phases with five expert informants, a technical analysis of the platform’s telemetry pipeline, and a supervised-classification feasibility probe testing whether fault detection is extractable from the constrained data substrate the platform currently retains.

The probe established that fault detection is technically feasible from the existing telemetry, at performance consistent with operational use. On this basis the thesis designs a six-block Target Architecture, traced to ETSI GS ZSM 002, and sequences it into a three-horizon Maturity Roadmap evaluated by the same informants.

The central finding is that the binding constraints on Level 4 autonomy are organisational and regulatory rather than technical, and that the realised autonomy ceiling is co-determined by operators — and even by client devices — outside the platform’s control. The architecture reconciles this through a supervised-autonomy posture: it provides full closed-loop capability while exposing the closing of the loop as a policy-governed, progressively widening authority. Level 4 is thus reached as a capability and approached as a behaviour — a qualification offered as a transferable design pattern for autonomy adoption under trust and accountability constraints.

Keywords: Cloud Monitoring System, AIOps, ETSI ZSM, Zero-Touch Networks, TR-369 USP, Design Science Research, Operational Maturity, Cloud Deployment Models

Subject: Information System Science

Title: Skin deep: generative AI vs. rule-based personalization in beauty E-Commerce

Abstract: 

The rapid diffusion of Generative AI across digital commerce has created new imperatives for beauty brands seeking to deliver genuinely individualized consumer experiences. This thesis examines how the adoption of Generative AI-powered personalization tools impacts customer experience, perceived personalization, conversion potential, and brand value in beauty e-commerce, with a particular focus on Erborian, a FrancoKorean prestige cosmetics brand and part of the L’Occitane Group, as an empirical case. The study is motivated by a significant gap in the existing literature: while the theoretical foundations of AI-driven personalization, consumer trust, and digital customer experience are well established, empirical research comparing rule-based and GenAI diagnostic tools in a real-world, consumer-facing beauty e-commerce context remains scarce.

The study adopts a qualitative research design grounded in an interpretivist epistemological stance. Primary data was collected through nine protocol-based semi-structured interviews conducted in May 2026 with female cosmetic consumers aged 18 to 30. Each session combined embedded live user testing of two skin diagnostic tools, Erborian’s rule-based text quiz and Yepoda’s Vision AI-powered Skin Analyzer (operated by Haut.AI, trained on three million facial images to assess 150+ skin biomarkers), with a post-experience semi-structured interview exploring six theoretically-derived themes. Data were analysed using a theory-driven thematic analysis (Braun & Clarke, 2006) anchored in four theoretical pillars: the Technology Acceptance Model (Davis, 1989), consumer trust (McKnight et al., 2002; Komiak & Benbasat, 2006), the Personalization-Privacy Paradox (Awad & Krishnan, 2006), and customer experience quality (Verhoef et al., 2009).

The findings reveal a central paradox: Yepoda’s Vision AI diagnostic is perceived as more objectively precise, yet Erborian’s rule-based quiz produces recommendations experienced as more personally relevant. This precision-relevance gap arises from a fundamental asymmetry in information architecture, the quiz captures the consumer’s subjective goal-state, while the Vision AI captures her objective skin-state, and suggests that genuine hyper-personalization requires the integration of both. All nine participants independently converged on the same consumer-derived ideal: a hybrid model combining optional Vision AI facial analysis with structured contextual questioning. The study further finds that brand trust remains the dominant mediator of purchase intention, overriding diagnostic quality in the majority of cases, and that AI disclosure generates a sharply segmented response, 5/9 participants indifferent or positively impacted, 4/9 cautious or resistant, with important implications for transparency strategy design. The Personalization-Privacy Paradox is empirically confirmed and theoretically refined: participants distinguish between AI technology disclosure, which can reduce adoption, and data governance disclosure, which is universally expected and reassuring.

The study makes four theoretical contributions: it introduces the distinction between objective skin-state and subjective goal-state personalization as a refinement of the perceived personalization construct; it demonstrates that brand equity moderates the conversion impact of AI diagnostic tools; it proposes a calibrated transparency model distinguishing between two functionally distinct types of disclosure; and it provides an empirically grounded benchmark for hyper-personalization architecture in beauty e-commerce. For Erborian, the findings yield three strategic recommendations: develop a hybrid diagnostic architecture with an opt-in Vision AI module; anchor the AI narrative in the brand’s authentic K-Beauty tech heritage; and implement a calibrated transparency strategy that foregrounds data governance while contextualising AI technology. The study’s limitations, including sample homogeneity, constant tool exposure order, and researcher positionality, point toward a rich agenda for future quantitative and longitudinal research.

Keywords: Generative AI, beauty e-commerce, skin diagnostics, personalization, customer experience, brand trust, Vision AI, Personalization-Privacy Paradox, hyper-personalization, Erborian, K-Beauty.

Subject: Information System Science

Title: Designing and Deploying Generative AI Agents for Sustainable Performance Gains in B2B Knowledge-Intensive Service Firms

Abstract: 

By 2026, generative artificial intelligence has become organizational infrastructure in most large firms, yet only a small minority of organizations record net-positive outcomes from their deployments. This thesis addresses the resulting gap between robust individual-level productivity gains and weak organizational performance, asking how those individual gains can be converted into durable organizational value in B2B knowledge-intensive service firms.

The central research question asks how generative AI agents can be systematically designed and deployed to generate sustainable performance gains, and what organizational conditions determine their effectiveness. It is operationalized through three sub-questions concerning design principles, measurable outcomes, and the organizational conditions that moderate effectiveness across tasks and employee profiles.

The research follows an action research design conducted within Praxedo, a B2B SaaS firm, across three phases: a qualitative diagnostic, the design and deployment of six purpose-built AI agents, and a before-andafter quantitative evaluation drawing on historical Salesforce baselines and matched consultant samples.

The interventions produced substantial gains, including task-time reductions of roughly 53 to 72 percent, an 85 percent reduction in CR filing delay for junior consultants, a 30.6 percent reduction in time-to-value, and a 2.4-point NPS improvement. Gains were consistently larger for junior than for experienced consultants, confirming a leveling-up effect. The findings indicate that sustainable AI value depends less on the technology itself than on the surrounding governance ecosystem: task-specific design, a Generate / Verify / Refine / Validate governance cycle, progressive deployment, and deliberate investment in human capability.

Keywords: generative AI, AI agents, organizational performance, B2B SaaS, action research, AI governance, knowledge transfer, productivity, customer success, leveling-up effect

Title: Linking IT Investment Benefits to the Profit & Loss Statement at Allianz SE

Abstract: 

This thesis investigates the current benefit governance practices at Allianz SE and how profit and loss statements (P/L) can be used in the steering of IT investments. It also analyses how a structured linkage framework can address gaps which hinder the aforementioned relationship. The thesis builds on the IT productivity paradox, as well as Benefit Realization Management literature, and therefore positions financial traceability at the intersection of both; arguing that it is the missing link between IT spend and IT value capture. To illustrate this, an explanatory case study design was applied to Allianz SE internal IT portfolio, for projects between 2018 and 2024, drawing on the firm’s internal IT portfolio management systems, financial planning platform, and IT economics reporting infrastructure. For this, the thesis combines descriptive portfolio diagnostics, with KPI delta computations with data hygiene conditions, structured mapping of KPIs to their intended P/L line landing zones, and a case study analysis using an Allianz internal project to test the proposed framework. The results of this paper show that the traceability gap is a structural gap rather than a data quality issue. The main contribution of this thesis is conceptualisation of a framework that deploys a five-layered benefits to P/L model, which specifies category hygiene standards, an automated 5 year benefit window, an approach to treat outliers, a mandatory P/L flagging rule, and a list of the 10 monetary KPIs which are mapped to “Growth”, “Productivity”, and “Losses”. This is followed by 9 recommendations for a successful implementation, within an 18-month time frame, which allows data model enhancements, as well as adjustments to the internal portfolio steering processes. This approach provides a preliminary framework for IT portfolio steering, and the basis for further empirical research, which is needed to validate the applicability in large corporate environments. 

Keywords: IT investment governance, Benefit Realization Management, IT Business Value, Economic Benefit Factor, KPI to P/L linkage, Value Based Portfolio Steering, Insurance IT Transformation, IFRS 17, Measurement Hygiene, Benefit Concentration

Subject: Information System Science

Title: The Successful Failure: How Governance Asymmetry and Sensemaking Processes Transform Technically Sound IT Projects into Organisational Failures

Abstract: 

Information technology projects within large organizations are continuously in improvement, and research new underexplored areas. Projects that meet their technical requirements are labelled as failures by the managerial structures that lead them. This phenomenon, not choose but installed deeply in the structure of the team, generates a divergence between project owners and managers on the definition of the cohesion, the risk aversion, and the systematic failure to learn from experience. Existing literature has documented the limitations of the Iron triangle (Time, cost, scope) as the dominant framework for IT projects and proposed multidimensional framework instead. However, the precise mechanisms that will change the failure projects into technically successful projects. And change the governance conditions that produce this outcome. Proposition that remains insufficiently theorized. This study addresses that gap. 

The central research question guiding this thesis is: How do organizations make sense of IT projects that are technically successful but perceived as managerial failures? 

Based mainly on Karl Weick’s (1995) theory of organizational sensemaking, and additionally the concepts of sensegiving and sensehiding theorized by Gioia & Chittipeddi (1991) and Whitney & Daniels (2013), the study will develops a six stage conceptual framework process explaining how the complexity driven technical adaptation can be transformed into an institutionalized failure narrative through governance, asymmetry and lack of retrospective.

The study adopts an interpretive approach, thanks to a qualitative multi-case study design conducted within the IT department of the BEL group, a large international food and consumer goods company. Four IT projects were selected for this study, trying to have a full vision of the scope of the department. They were specifically chosen due to a lack of visibility, communication, or failure that happens during or after the process. Data was collected through nine semi-structured interviews with both project owners and managers. The results were analysed using an abductive analysis framework across six phases. Four major themes emerged: divergent definitions of success, structural invisibility of technical work, the role of the manager project owner relationship, and the universal absence of formal post-mortem processes.

The findings confirm the six-stage framework while introducing two significant changesets. First, the evaluative difference between project owners and managers is not entirely a divide of perception; participants assumed that multi-dimensional success was important and that governance design was the real problem. The information available to each layer determines what can be seen, shared, and remembered. Second, the study identifies the absence of post-mortem reviews as a sensemaking void: a structural organizational condition in which the technical knowledge of a project is never formally recorded, leaving a void of knowledge that emphasizes the Iron triangle deviation rather than adaptive technical achievements. The result is a selfreinforcing cycle in which governance identifies the wrong weaknesses that produce invisible complexity and perception gaps. This gap produces failure narratives that shape the governance of future projects. 

The study concludes with recommendations for the institutionalization of postmortem processes, real-time complexity mechanisms, and governance orientation toward stewardship and not control.

Keywords: IT project management, Sensemaking, Iron Triangle, managerial failure, technical success, perception gap, post-mortem, organizational learning, governance, sensegiving, sensehiding, qualitative research, multi-case study

Subject: Information System Science

Title: Change Management and the Integration of Generative AI in Organizational CRM Projects: A Qualitative Study in the Context of Salesforce Implementation

Abstract: 

This thesis examines how change management (CM) practices shape the adoption of generative AI in customer relationship management (CRM) implementation projects. Using a mixed-methods design combining semi-structured interviews (N=5) with a pilot survey (N=89), we study two Salesforce CRM projects conducted by Accenture for two clients, a major French energy and services group and a French water utility. The findings reveal that AI readiness is highly uneven across populations, that resistance takes primarily epistemic rather than affective forms, and that change management is systematically deprioritized despite its measurable impact on readiness. We introduce two theoretical concepts: epistemic resistance, resistance rooted in the inability to evaluate AI output reliability, and augmentation drift, the gradual shift from augmentation to de facto automation when users stop critically engaging with AI outputs. The pilot data supports epistemic resistance as a stronger predictor of non-adoption (β=-0.289) than affective resistance (β=-0.179). Practical implications include investing in AI literacy over traditional communication, diagnosing digital maturity before deployment, and using regulatory compliance as a trust-building mechanism.

Keywords: Change Management, Generative AI, CRM, Salesforce, Epistemic Resistance, Augmentation Drift, Mixed Methods