Software-defined vehicles (SDVs) rely on tightly coupled software ecosystems spanning in-vehicle controllers, edge platforms, cloud services, Over-the-air (OTA) infrastructure, cybersecurity services and supplier-managed components. This architectural shift has fundamentally altered defect characteristics, creating cross-domain failure modes that exceed the capabilities of traditional automotive defect triage and resolution processes. Technology managers must address fragmented incident management systems, component-centric analysis, and severity-based prioritization models that limit organizations' ability to rapidly identify, correlate, and resolve safety-critical defects at fleet scale. This article proposes a unified, cloud-native framework for intelligent defect triage and resolution tailored to SDV environments. The framework is designed to integrate signals from vehicle diagnostics, OTA update systems, cloud observability platforms, and supplier incidents that feed into a unified, normalized, safety-aware defect model. Machine learning–driven correlation, automotive safety classification, and fleet-level impact assessment enable automated prioritization and resolution orchestration while preserving regulatory traceability. Drawing on lessons from industrial implementations of automated incident management, observability, and bug triage systems, the framework outlines how organizations can modernize defect management practices while preserving human oversight and regulatory traceability. The article also discusses implementation considerations, governance requirements, anticipated business benefits, and adoption challenges relevant to technology managers responsibilities. The framework illustrates how cloud native architectures can support both operational scalability and safety assurance in modern automotive software ecosystems essentially benefitting and encouraging technology managers into its adoption.
Traditional supply chain network design (SCND) and optimization projects prioritize mathematically optimal solutions but do not consider the friction of implementation in the business environment. As a result, there exists a gap between the estimated savings and the actual ones. The size of the gap varies by company, project and industry. The existence of this gap makes it difficult for the management to choose between alternative projects and undermines the credibility of the analytics behind the optimization work. Furthermore, as modern networks grow increasingly intertwined due to global disruptions and digital integration, the penalty for underestimating this transition friction has grown exponentially. This paper addresses this gap between theoretical savings and realizable value by introducing an Implementation-Weighted Framework for decision-making in supply chain optimization projects. The framework quantifies implementation difficulty through a highly customizable Complexity Index (CI). This index is calculated by deconstructing proposed scenarios into core structural levers such as physical nodes, product-plant allocations, and transportation modes and assigning organization-specific weights to each. By shifting the primary decision metric from absolute savings to Return on Change (ROC), technology managers can adequately access the potential of every initiative and make data-driven decisions.
Digital business models are becoming increasingly important as digitization reshapes both private and business sectors. Yet many companies struggle to select a suitable method for developing and optimizing digital business models. This study addresses this challenge by evaluating six methods: the St. Gallen Business Model Navigator, Business Model Canvas, Digital Canvas, IoT Business Model Builder, Business Model Innovation Framework, and D³ Model. The methods are assessed using six criteria: innovation focus, flexibility, practical applicability, ease of use, scalability, and market orientation. The evaluation combines strengths and weaknesses analysis, utility analysis, and a decision matrix. In addition, a case study with a traditional manufacturing SME at medium digital maturity demonstrates the practical application of the D³ Model. For the context studied, the D³ Model emerges as the most suitable approach, offering a systematic, customer-centric, and data-driven way to support digital business model development. This result is interpreted as context-dependent decision guidance rather than evidence of general methodological superiority. By highlighting the distinguishing features and trade-offs of the evaluated methods, the study supports managers in selecting methods that fit their specific organizational context, strategic priorities, and resource constraints. The findings further emphasize the managerial value of structured and data-driven approaches to business model innovation in maintaining competitiveness in an increasingly digital environment.
Small and medium-sized enterprises (SMEs) often face significant obstacles in digital transformation (DT) due to fragmented operations, disconnected systems, limited resources, and the lack of execution-ready roadmaps tailored to their operational constraints. This study proposes the IDEAL–ACT model, a dual-domain, modular framework designed to support SMEs in progressing from basic workflow digitization toward integrated, data-enabled, and analytics-ready operations. The vertical domain, IDEAL (Identify, Digitize, Establish, Accumulate, Learn), focuses on internal process stabilization, workflow digitization, systems integration, data accumulation, and analytical learning. The horizontal domain, ACT (Aggregate, Contextualize, Transform), extends this foundation toward external connectivity, contextual analytics, and future AI-oriented applications. A central premise of the framework is that Cyber Physical Systems (CPS) should not be treated as an initial prerequisite, but as a capability that is progressively developed through structured IT–OT integration. The framework was applied in a mid-sized construction chemicals manufacturer to assess its practical feasibility. The case demonstrated improvements in traceability, synchronization, data continuity, and selected operational performance indicators, while also clarifying the staged nature of analytics adoption. Rather than prescribing technology-first transformation, it provides managers with a sequence aligned with organizational and infrastructure readiness. The IDEAL–ACT model contributes to DT theory and practice by bridging the strategy–execution gap, positioning CPS as an enabling capability for analytics readiness, and emphasizing that, particularly in SME contexts, intelligence should follow integration.
This study examines how Top Management Team (TMT) cultural diversity and organizational mechanisms influence the adoption and use of information and communication technologies (ICT) in industrial and commercial companies. ICT adoption is considered an organizational capability that can facilitate the subsequent implementation of artificial intelligence (AI). Using data collected from 480 companies, the hypothesized relationships were tested through partial least squares structural equation modeling (PLS SEM). The measurement model demonstrated adequate reliability and validity. The results indicate that TMT cultural diversity has a positive and significant effect on ICT adoption and use. Formalization also positively influences ICT adoption, whereas decision-making centralization has a negative effect, suggesting that highly centralized structures may hinder technological implementation. These findings provide empirical evidence of how TMT characteristics and organizational mechanisms shape digital transformation and create conditions that may support the transition toward AI adoption. Future research should examine the roles of leadership and innovation culture in this process and conduct comparative studies across sectors and regions.
Persistent rework and low first-pass yield remain common in precision, regulated manufacturing despite years of improvement efforts. This study proposes and validates an Integrated Quality Control Framework that combines human oversight with AI-based measurement to address socio-technical barriers in human–robot collaboration. We created a four-layer structured framework: Assessment, Design, Implementation, and Monitoring, and compared three quality control solutions: human-based, AI-based, and integrated in an experimental way. An established regulated manufacturer, referred to here as “Alpha” for confidentiality, experienced rework rates between 85 and 95% during the ten-year period 2016 to 2025, with a First Pass Yield below 4%. This constituted a major quality problem that was not resolved through traditional improvement efforts. A measurement test with five human inspectors individually with more than 5 years of experience was carried out to compare these approaches. This human-in-the-loop approach confirmed significant variations between the humans' coefficient of variation CV: 1.4 to 9.3%, while an AI solution had 4-12× higher consistency CV: 0.2-2.0% in the measurement result for geometrically complex features. The integrated approach is expected to benefit a 15–30% reduction in rework rates, an increase in First Pass Yield from
Dynamic economic dispatch (DED) is a recurring operational decision that directly shapes the fuel cost, reliability and emission footprint of any electricity-producing organization—from large utilities to small and medium-sized microgrid operators and emerging energy startups. Classical repeated optimization can become difficult to operate under fast-changing demand, ramping limits and growing decision horizons. This paper proposes a deep reinforcement-learning (RL) decision-support framework that learns a feedback dispatch policy from a simulated power-system environment and provides fast advisory recommendations to human operators. RL is positioned as a sampling-based approximate dynamic programming approach, not as a replacement for classical optimization. Six RL algorithms—Soft Actor-Critic (SAC), Proximal Policy Optimization (PPO), TD3, DDPG, A2C and REINFORCE—are evaluated on four test systems (5, 10, 15 and 20 generating units) using the same environment, reward design, feasibility-repair layer and evaluation protocol. The results do not support a universal winner: performance depends on the joint assessment of cost, after-repair feasibility, inference time and robustness across case sizes. SAC is the most robust feasibility-first method in the reported experiments, while PPO, DDPG and TD3 show case-dependent cost advantages under specific system sizes. The framework is therefore interpreted as an RL-based decision-support architecture rather than as a single-algorithm prescription. Model inference requires approximately 0.36–1.10 ms per dispatch decision, supporting low-latency advisory computation; end-to-end deployment latency is not measured. Managerial implications, a seven-stage deployment roadmap, and use cases for utilities, SMEs, and startups are provided. RL is recommended as an operator-in-the-loop advisor, not an autonomous controller.
Retaining skilled personnel in Research & Development (R&D) is vital for maintaining an organization's innovative capacity and long-term sustainability. However, quantifying the specific impact of leadership on staff turnover using standard Human Resources (HR) data remains a persistent challenge for engineering managers. This study addresses this gap by proposing a pragmatic Explainable AI (XAI) framework to evaluate “managerial-proxy” variables workplace conditions heavily influenced by leadership actions—that are most associated with attrition risk. Utilizing the publicly available IBM HR Attrition dataset, representing a cohort of 961 R&D employees, we developed a high-performance XGBoost machine learning model that achieved a ROC AUC of 0.82 and an accuracy of 0.8672. Subsequent interpretative analysis via SHAP (SHapley Additive exPlanations) identified that working overtime, low job satisfaction, lower monthly income, and limited tenure with the current manager are the primary drivers increasing an employee's likelihood of leaving. Unlike traditional descriptive reports, this framework provides engineering organizations with a data-driven diagnostic toolkit and a retention checklist to move from reactive data tracking to proactive, targeted leadership interventions. The proposed methodology offers both large global corporations and agile startups a replicable roadmap for building more resilient, human-centric, and sustainable technical workforces. By providing transparency into the “black-box” logic of machine learning, this research fosters the trust required for the successful integration of AI-augmented decision-making into modern engineering leadership workflows.
Training is a mandatory requirement across industries, yet its sustainability hinges not only on effective delivery but also on efficient administration. While previous research has largely focused on optimizing training content and methods, the administrative workload remains an underexplored area despite its significant impact on time and cost. This project applies the Lean Six Sigma (LSS) framework, using the DMAIC cycle to analyse and improve the administrative processes involved in organizing Internal and External Instructor-Led Training (ILT). By mapping the phases, administrative, pre-training, during training, and post-training, the study identifies inefficiencies and explores how digitalization and automation can streamline operations. The implementation of digital tools led to a reduction of administrative steps by 33% for Internal ILT and 27% for External ILT, contributing to waste elimination, cost savings, and enhanced sustainability. Although the findings are context-specific and dependent on technological readiness, the approach offers a replicable model for improving training administration across sectors, adding value through operational efficiency and long-term strategic impact.
Engineering managers in capital-intensive industries face a critical challenge: how to select high-risk, high-reward innovation projects when traditional tools like Net Present Value (NPV) are unreliable. NPV encourages anchoring on financial forecasts that are often the least reliable part of a proposal, pushing managers to either reject transformative projects because their value is hard to quantify, or fund weak projects based on overly optimistic numbers. This article introduces a practical, four-step decision framework—integrating DEMATEL, the Analytic Network Process (ANP), and Fuzzy TOPSIS—designed for corporate venture capital (CVC) and strategic technology investments. The framework first maps causal relationships among investment criteria to separate root causes from downstream effects, then derives interdependency-based weights, ranks candidate portfolios under linguistic uncertainty, and produces a transparent, defensible recommendation. Through a 48-month longitudinal case study at a ${\$}$100 million CVC fund, validated by a 27-member expert panel spanning financial, technical, and strategic perspectives, we demonstrate the framework's real-world efficacy. A key insight for managers is that upstream risks, such as technology disruption and regulatory compliance, are far more influential on project success than downstream financial projections, which the analysis confirmed account for only a small share of the overall decision weight. The portfolio selected using this framework achieved a 23% higher success rate, together with a higher average ROI, stronger strategic score, and faster time to value, compared to the firm's historical baseline. These results provide a validated methodology to de-risk strategic innovation, offering practical value to large enterprises and resource-constrained SMEs alike.
The COVID-19 pandemic served as both a crisis and a catalyst for transformation across Latin American business landscapes. This editorial article synthesizes key insights from eight empirical studies featured in this special issue. These studies span diverse yet interlinked themes, including workplace transformation, entrepreneurial resilience, environmental sustainability, digitalization, emotional well-being, and inclusive leadership. Collectively, they reveal that Latin American firms—especially small- and medium-sized enterprises and women-led ventures—faced not only operational disruptions but also deep social and psychological challenges. Yet, many responded with remarkable adaptability, leveraging informal networks, digital tools, and people-centered management approaches. The papers highlight how flexible work models, emotional salary, and virtual platforms reshaped organizational resilience and worker engagement, while also cautioning against the unintended consequences of digital acceleration, such as mental health deterioration and social exclusion. Moreover, the findings emphasize the need for context-sensitive, inclusive strategies—whether through green finance reforms, gender-responsive entrepreneurship support, or institutional mechanisms for well-being. By drawing connections across sectors and methodological approaches, this article distills critical lessons for businesses, educators, and policymakers seeking to build more equitable, innovative, and resilient economies in the post-pandemic era. This article also outlines future research opportunities aimed at deepening our understanding of leadership, sustainability, and digital transformation in Latin America’s evolving and often turbulent socioeconomic environment.
Peer review rests on three foundational commitments: confidentiality, reviewer accountability, and protection of the author's intellectual property. The widespread availability of artificial intelligence (AI) tools, such as ChatGPT, Gemini, and Claude, puts all three at risk in a single action. This editorial outlines three reasons reviewers must not use AI to generate or draft reviews. First, sharing a manuscript with any AI provider breaches the confidentiality obligations every reviewer accepts, regardless of the provider's data practices. Second, it exposes the author's ideas, data, and methods to opaque third-party systems, thereby violating the intellectual property protections they are entitled to until publication. Third, it substitutes a probabilistic output for the reviewer's own accountable judgment, placing the reviewer's reputation and integrity at risk. We describe recent cases of AI-generated reviews at a major conference and adversarial prompt injection in submitted manuscripts, and propose that IEEE itself consider building a compliant AI tool that augments reviewers on narrow, verifiable tasks without replacing their judgment or accountability. Until such a tool exists, reviewers must not use third-party AI tools for any part of the review process.
While Blockchain Technology (BCT) is widely touted as an enabler for the Circular Economy (CE), engineering managers lack structured tools to evaluate which technical features align with specific circular strategies. This article presents a practical decision-support framework using the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS). Integrating the Technology-Organization-Environment (TOE) framework with the Technology Acceptance Model (TAM), we prioritize BCT characteristics such as immutability, transparency, and smart contracts across six key circular practices. Results from a panel of domain experts indicate that while transparency is universally critical, specific features like cryptography are nonnegotiable for Digital Product Passports (DPP) but less relevant for material recovery. Furthermore, sensitivity analysis reveals that deploying blockchain as a Decision Support System offers higher organizational utility than purely financial use cases. This study provides a quantifiable roadmap for technology leaders to de-risk blockchain adoption in sustainable supply chains.
In an era of rapid technological change driven by artificial intelligence (AI), engineering education plays a pivotal role in preparing future professionals for sustainable industrial management. Management and Production Engineering (M&PE) programs offer an opportunity to develop skills and competencies to lead project teams, implement innovations, and engage in research and development aligned with the principles of Industry 4.0 and 5.0. This article includes a literature review that identifies key AI-related skills expected of managers and production engineers, as well as thematic trends in AI integration within M&PE education. It examines M&PE curricula at Polish universities, focusing on the extent to which they incorporate AI-related content. Expert insights are presented within a SWOT-AHP framework to assess the feasibility and implications of embedding AI into M&PE education for responsible and sustainable industrial leadership in AI-driven environments. The identified strengths and weaknesses provide guidance for designing educational programs. This enables assessment of graduates' skills and strategic alignment of their roles within project teams, ensuring better integration of AI tools into industrial management processes.
This article describes the deployment of Amazon QuickSight—Amazon Web Services' AI-powered business intelligence platform—to analyze transaction and position data reconciliation reports in a financial services environment. In a dataset of approximately 84,000 transactions over a six-week reporting cycle, the tool flagged 203 raw discrepancies compared with 147 from the manual process, a 38% increase in raw detection. Following analyst-led reclassification of timing artifacts, 172 errors were validated, representing a net increase of approximately 17% over the manual baseline. Root-cause analysis further revealed that 71% of validated errors originated from two upstream system integration points that the manual process had never isolated. The investigation that followed, and the organizational resistance it encountered, reflects patterns well-documented in algorithm aversion research. Experienced analysts, whose professional identity was tied to manual reconciliation expertise, initially rejected the tool's findings—consistent with Dietvorst, Simmons, and Massey's demonstration that people lose confidence in algorithms more rapidly than in human forecasters after observing the same error. This article reports what happened, what was learned, and translates those lessons into practical guidance for engineering and technology managers across organizations of varying size: large financial institutions, small- and medium-sized enterprises, and fintech startups. The central argument is that AI-assisted reconciliation tools change not only the speed of reporting but the nature of insight that becomes available—and capturing that change requires deliberate management action, not just technology deployment.
Manufacturing enterprises are increasingly embedding artificial intelligence (AI) inside Enterprise Resource Planning (ERP) transactions for day-to-day operations. The promise is simple: better decisions, fewer exceptions, and less wasted motion. The underestimated challenges are managerial and operational: defining what “best” means when objectives conflict; meeting transaction-time expectations in high-volume work; enforcing hard constraints outside the model; designing for rugged task worker devices; and building a control loop where overrides and feedback become signals for improvement rather than friction. This article presents a manager-ready playbook for deploying AI decision support inside live ERP execution systems, emphasizing speed, governance, and operator trust over model sophistication. Eight deployment areas are covered, each drawn from hands-on work in high-volume manufacturing environments. The first two deal with getting objectives and constraints right before anything is built. The next four cover workflow placement, device limitations, missing recommendations, and override handling during rollout. The final two address how to measure what matters and how to build lasting adoption. Each area comes with a concrete management pattern. The discussion references published work on deployment failures, automation trust, and AI governance where relevant. Together these eight areas give engineering managers a practical structure for bringing AI into live ERP execution without losing operational control.
The source and quantity of energy use yield the greenhouse gas footprint for operations. Traded open markets for energy and emissions enable applying these parameters for decision-making. This paper analyses the dynamics of energy use in operations where energy price and carbon emission from the grid are hourly fluctuating and vary at different times of the day. Based on the analyses, we show how alternative energy storage capacity strategies may impact the cost of energy consumed and energy-related emissions. These objectives may be conflicting with each other. A simulator software is presented to analyze the scenarios.
As organizations scale artificial intelligence (AI), responsible AI has shifted from a technical checklist to an enterprise leadership and governance challenge. This study examines how senior digital leaders operationalize trustworthy AI through interviews with executives from retail, biotechnology, fintech, cybersecurity, and consumer services. The study identifies six organizational components—leadership, talent, data, storytelling, standards, and enforcement—that shape ethical AI governance. The findings reveal recurring implementation mechanisms used to reduce ethical risk during AI development and deployment, including leadership-led governance forums that adjudicate use cases, boundary-spanning roles that translate between product, legal, and technical teams, ethical review checkpoints embedded in the AI lifecycle, data-quality ownership and reliability controls, and narrative practices that make model behavior interpretable for decision makers and stakeholders. Across cases, leadership functions as the integrator that aligns culture, accountability, and cross-functional coordination so these components operate as a coherent system. The study provides practice-oriented guidance for managers responsible for AI products, analytics, operations, compliance, and digital transformation by detailing governance and decision-process patterns that innovation, compliance, and accountability.
The digital transformation of construction value chains must drive the development and integration of innovative components and systems in the future built environment. This would support the transition from today resource-intensive practices towards a smart and regenerative circular economy, addressing increasing regulatory complexity and customized end-user requirements. Still its implementation faces challenges to enhance interaction and co-creation among stakeholders and to manage growing data volumes over the lifespan of construction products and buildings without unduly burden innovation stakeholders and especially SMEs. This article presents an exploratory case study of a digital clustering service based platform to help the European construction value-chains and their SMEs dealing with innovation development: clustering enhances knowledge sharing, opportunity matchmaking, and collaboration across the innovation lifecycle, facilitated by a digital platform providing access to value-added services. Two services are introduced as lighthouse examples: the Digital Product Passport, mandated by EU legislation to improve product traceability and transparency in support of the circular economy; and a Retrieval-Augmented Generation architecture based on a Large Language Model toolkit to accelerate document processing and navigation of complex construction regulations. This digital transformation case study, based on field tests and early observations, aims at assessing, from a managerial perspective, coordination of expertise and innovation leadership by value chains and SMEs. It stresses the role of systematic access to evolving regulations for better informed decision-making aligned with circular economy objectives. The main expected impact for construction SMEs is a transition from a ‘Business-as-usual’ innovation creation toward a co-development process by all holistic value-chain stakeholders over the innovation life-cycle stages, based on Innovation 2.0 principles and powered by a clustering platform.
