South Korea Charts Talent Policy for Agentic AI in Software Industry
Korea is witnessing a renewed national debate over agentic artificial intelligence, a class of AI that can autonomously plan, execute tasks, and even write software. In industry circles, the term SaaS apocalypse, or “SaaSpocalypse,” has circulated as AI begins to redefine how software is built and delivered. Analysts say AI that can design and code could shrink traditional software projects from years to weeks or even days, reshaping both the software industry and the job market.
On March 13, the Ministry of Science and ICT (MSIT) convened a policy briefing in Seoul’s Gwanghwamun district, gathering experts at the National Council for Science and Technology Advisory’s main conference room. The session, titled “Agentic AI Era: Responses for the Software Industry and Talent Development,” followed six prior government-hosted colloquia as the ministry seeks a concrete plan to shift Korea’s AI and software policies.
In total, the ministry has held six “innovation colloquiums” focused on software, AI, and talent development over the past month, drawing about 70 representatives from industry, academia, and research labs. The March 13 meeting aimed to translate those discussions into actionable policy directions for talent cultivation and AI-enabled software industries.
The discussion sits against a broader global context. In January, Anthropic released Claude Co-Work, a development that sparked forecasts that enterprise software could be replaced by AI-assisted workflows. Analysts linked the release to a sharp reevaluation in the U.S. software sector, noting a substantial decline in market capitalization for some leading software firms.
Industry voices in Korea warned that the center of gravity in the software ecosystem is shifting from suppliers to users. Cho Joon-hee, chairman of the Korea AI and SW Industry Association (KOSA), said customers are increasingly using AI to build only the features they need, pressuring policymakers to rethink talent pipelines to match demand rather than traditional supply. He argued that talent programs must be redesigned to align with this demand-driven environment.
Within companies, changes are already evident on the ground. The number of developers working with a single software product manager is shrinking as teams reorganize to leverage AI-enabled workflows, and developers’ roles are beginning to evolve accordingly. The shift underscores a broader need for professionals who can define problems, design systems, and verify results in AI-augmented settings, not just write code.
Education and training were central themes. Participants emphasized the need for graduates who combine deep software fundamentals with domain expertise and AI fluency. Song Ho-cheol, head of the DX division at Duzon Bizon, argued that AI now handles coding, so developers spend more time shaping AI behavior and defining how systems should work, prompting universities to prioritize design and verification skills over routine coding.
Other policymakers urged integrative approaches in higher education. Shin Jeong-gyu, CEO of RableUp, noted that universities should favor interdisciplinary courses and project-based learning that let students tackle real-world problems, rather than adding new departments for the sake of expansion. The aim is to cultivate both advanced software specialists and “convergence” talents who can apply AI across diverse sectors.
There was also a call to strengthen infrastructure for AI education. KAIST professor Seong Min-hyeok argued that Korea lacks campus-level GPU clusters needed for hands-on AI training. He suggested a mix of public, university, and industry solutions—potentially repurposing enterprise GPUs for education and creating a multi-university cluster—to ensure training capacity keeps pace with the demand.
The government signaled its intent to deliver an overarching talent policy that fits the agentic-AI era. Deputy Minister Ryu Je-myeong of the MSIT said the administration is identifying the new profiles of talent required in practice and is reexamining current AI talent development and support mechanisms to ensure they align with on-the-ground demand and AI-native work environments. For U.S. readers, Korea’s approach illustrates how advanced economies are recalibrating their education and workforce pipelines to compete in a rapidly evolving AI-driven software ecosystem, with implications for global supply chains, multinational software vendors, and the cross-border flow of AI talent.