Learning and using technology in the workplace are essential for a company’s commitment to the sustainable development of its resources. Finding competent engineers who can handle information communication technologies (ICTs) is a challenge for companies. Currently, however, the ability to use these technologies is limited to technicians with specialized training, and not everyone can engage in development. Therefore, it is safe to conclude that equity in the use of technology has not yet been realized. This study aims to analyze, based on actual cases, the necessary conditions and mechanisms for people with diverse experiences and circumstances, not limited to engineers, to participate in ICT development to address human resource diversity. The use of technology such as low-code platforms (LCPs) that have recently emerged on the market has shown that nonprofessional engineers without programming training can participate in development projects. This research will be useful to managers in advancing Diversity, Equity, and Inclusion (DEI) strategies in their workplaces and contribute to organizational research regarding new trends in technology use by individuals: low codability. The findings of this study are of significant relevance to the Sustainable Development Goals (SDGs) of decent work and economic growth, as well as gender equality.

Work automation is shifting from centrally built systems to decentralized, frontline, self-service tools refined in use. Agentic AI extends this shift by enabling large language model (LLM)-based agents to interpret instructions, plan multi-step tasks and act via tools. However, value often fails to scale when similar agents proliferate across projects without coordination. Drawing on a comparative analysis of five AI projects conducted in a Japanese industrial conglomerate, this study examines how governing a System of Agents (SoA) as an organizational portfolio of reusable assets can support sustained value creation. Using traceable project artefacts, weekly review records, development logs and operational feedback traces, we identify four configurations, from model-based AI to portfolio-governed SoA. Cross-case analysis suggests that SoA adoption alone is insufficient: sustained value creation only became observable when governance shifted towards assetization, standardized logging and human intervention routines, supported by portfolio-level decision rights that facilitate reuse and learning across deployments. The article derives design implications for the governance of Agentic AI as reusable organizational capabilities, treating return on investment as one observable dimension of value.