Yeom Dong-hoon (염동훈), CEO of AI and cloud company MegazoneCloud, on April 2 stressed that a practical framework is essential to manage governance, security and compliance to create large-scale value through new technology. He made the remarks at the partner conference, ICON 2026, held at the Grand Ballroom of InterContinental Seoul Parnas.

In its third year, the event brought together 21 of MegazoneCloud's global technology partners and major customers to share trends in AI, cloud and security technology and real-world operational cases.

MegazoneCloud introduced its Enterprise TRUST Layer strategy, focusing on governance, security and compliance issues companies face after adopting AI and how to address them. The strategy sets five standards for managing AI operating environments across the full process of adopting and running AI: Traceability, Regulation, User Access, Standardization and Tooling.

Hwang In-cheol (황인철), MegazoneCloud's chief revenue officer, said successful AI is determined by a structure that is well operated rather than a well-made model. He said companies need to reflect complex regulations and requirements from the design stage to embed them into systems.

Gong Seong-bae (공성배), MegazoneCloud's chief AI officer, unveiled the enterprise AI OS AIR Studio V2. He said building, running and controlling agentic AI requires an enterprise AI OS that centrally integrates management. Gong said as AI use expands, companies face various challenges including data reliability, security and usage-based budget controls.

Gong said MegazoneCloud established its enterprise AI governance framework, TRUST, and applied it across AIR Studio V2 to solve these issues. He added that AIR Studio V2 integrates model and data management, orchestration and governance control functions into a single platform to support companies in operating AI in a safer, more efficient and scalable way.

A panel discussion was also held among Korean Air, global cloud security company Wiz and MegazoneCloud's HALO unit, which oversees its security business, under the theme of cloud security and changes that come from altering structures rather than tools.

Kim Hyo-jong (김효종), a team leader at Korean Air's information security office, introduced the airline's case of building a Wiz security system. Kim said Wiz enabled the security team to identify and act on the real risks it needed to focus on, and that after adoption it felt as if a fog had lifted. He said that, like distinguishing what needs immediate treatment and what requires management through a health check-up, the company can manage hundreds of vulnerabilities in a multi-cloud environment on a single screen and enable proactive controls that prevent risks early.

Matt Zwolenski (Matt Zwolenski), senior director at Wiz, said the key in cloud security is not adding more solutions but securing visibility that can identify which risks truly matter in a complex environment and manage them structurally. He said it is important for different teams to build a seamless workflow based on a common language and achieve the "democratisation of security".

Wi Soo-young (위수영), head of the HALO unit overseeing MegazoneCloud's security business, said security is not simply a defensive technology but an operational design that keeps business from being interrupted. He said securing visibility can also improve collaboration structures across organisations.

Companies participating in presentation sessions, including Intel and Articul8, also presented practical challenges in the AI adoption process and ways to address them.

Jason Tan (Jason Tan), who oversees sales development for Intel's Asia-Pacific AWS business, said many companies face a bigger challenge not in adopting AI itself but in how to stably scale and operate it after a proof of concept. He said Intel will cooperate with MegazoneCloud to help companies expand AI beyond experimentation into real business environments.

Edward Kong (Edward Kong), who is in charge of strategic accounts and global partnerships at Articul8, said many companies succeed in generative AI proofs of concept but face practical barriers in scaling to real work and service environments, including security, data and operational stability. He said it is difficult to differentiate by only providing AI and that companies need to be able to systematically apply and internalise enterprise data and domain expertise into AI.