■ DevOn AIND coding tool drives 26% software development productivity gain, outperforming GitHub Copilot and Codium by 2X

■ The more difficult and complex the development project, the greater role AI can play, with stronger impact observed in finance and manufacturing verticals

■ The findings derive from a joint study with Sogang University that rigorously analyzes AI’s productivity impact, with results published in the international journal JGITM

Seoul, South Korea February 12, 2026 – LG CNS has demonstrated the effectiveness of AI‑based software development, finding that its proprietary AI development solution delivers roughly twice the productivity gains of general‑purpose AI coding tools when used in enterprise environments.

The findings are based on a year-long joint study with Sogang University’s Professor Hyunkyu Park to verify and evaluate the productivity of AI-enabled software development. Professor Park’s team worked closely with LG CNS developers on everything from concept definition to research design and analysis, as well as the development of productivity measurement frameworks.

The study examined 26 real-world LG CNS development projects to quantify changes in productivity before and after AI adoption. Findings showed that LG CNS’ proprietary coding tool, DevOn AI Native Development (DevOn AIND), increased development productivity by 26.1% on average, compared with 14.1% achieved by general-purpose AI coding tools such as GitHub Copilot and Codium – roughly doubling the productivity improvement.

DevOn AIND is an AI-based solution trained on LG CNS’ development standards, quality criteria, and deliverables. It automates the execution of the development lifecycle from project design and code generation to testing and quality assurance. It also provides customized features, such as generating source code for large-scale enterprise development and operations management projects.

The study found significant productivity improvements across major industries, including finance (23.1%), manufacturing (15.5%), chemicals (11%), electronics (6.3%), and batteries (4.2). The impact was more pronounced as project complexity increases. For example, tasks that previously required developers to manually cleanse irregular raw data collected from process control systems can now be handled automatically by AI using AIND, leaving developers to focus on relatively higher‑value activities such as review and optimization.

The performance gains are driven by DevOn AIND’s development-specialized capabilities, powered by AI trained on diverse datasets, including the company’s extensive experience and know-how across industries. DevOn AIND produces source code tailored to System Integration (SI) and System Management (SM) projects and supports on-premises enterprise deployment.

Based on this joint research, LG CNS and Sogang University published a paper examining the impact of AI on development productivity in the prestigious Journal of Global Information Technology Management (JGITM). This international journal covers key topics including IT strategy, information systems management, and digital transformation, with all published submissions receiving rigorous evaluation of research methodologies and findings to ensure a high level of credibility and academic authority.

“Before fully adopting AI, global companies evaluate technology management metrics such as productivity, quality impact, and economic feasibility,” said Professor Hyunkyu Park of the Graduate School of Management of Technology at Sogang University. “This joint study is a pioneering case of evaluating AI productivity in real‑world operations.”

“We have established a clear vision that goes beyond AI-assisted development to AI-native development,” added Sunjung Kim, Senior Vice President and Head of the Delivery Innovation Center at LG CNS. “We are rapidly approaching a whole new era, where tasks that once required the work of 20 people over several weeks can be completed in just 30 minutes with the help of AI.”