The pace of development of generative artificial intelligence (AI) is accelerating by causing tectonic fluctuations faster than we can imagine. Just over two years after entering the market, AI has become a key pillar of corporate management. Some executives are still skeptical about whether AI can make any real difference to their business, but market experts predict that the era will come when AI becomes a basic infrastructure like electricity or the Internet.
In terms of corporate management, it is also diagnosed that the speed of decision-making on introducing AI will soon lead to a competitive gap. AI is evolving at a tremendous pace even in an uncertain market environment, and now it is time for companies to specify strategies for “how to use AI” beyond “whether to introduce AI.” Mail Business listened to opinions on how companies should apply AI to Deloitte’s AI Integrated Service Group (One AI) in Korea.
Deloitte, the world’s largest accounting and consulting firm, suggested the infrastructure of △ multi-modal AI △ edge AI △ AI as a key AI technology trend that companies will pay attention to in 2025.
First of all, the rise of multimodal AI. AI is evolving beyond traditional textual foundations to simultaneously process images, speech, images, and even tactile data. Through this, companies can provide more sophisticated customized services by analyzing facial expressions and voice nuances beyond the level of customer response.
The introduction of edge AI is also rapidly spreading. Edge AI is a technology that enables AI to quickly and safely implement artificial intelligence functions without cloud connections by processing data immediately on the device itself, such as smartphones and autonomous vehicles. Edge AI provides answers to the need for systems that process data in real time and make instant judgments in the field. This demonstrates excellent competitiveness in manufacturing, where quality control is important, or in the field of autonomous driving, which requires instantaneous judgment, and accelerates the introduction of field-oriented AI.
Finally, the infrastructure of AI means that AI is no longer an additional technology in corporate operations, but an essential resource such as electricity and the Internet.
In particular, ‘AI as a Service (AIaaS)’ which is a cloud-based service-type model that provides artificial intelligence functions is spreading. It is a structure that makes it easy to use AI functions necessary through the Internet without building a separate AI system or having large-scale infrastructure. This means that an environment is being prepared in which small and medium-sized companies can easily secure AI capabilities at the level of large companies without building complex systems.
CEO Checklist for Successful AI Transformation
According to the Deloitte report, 67% of companies around the world are expanding their investments in AI, but only 10% have a mature operating system. Agnostic AI contributes to productivity and competitiveness improvement by enabling autonomous decision-making and execution beyond simple command execution, but there are also great concerns about the loss of responsibility and control.
Based on the “Basic Act on Artificial Intelligence,” the Korean government is also working on sub-laws such as AI risk classification, self-inspection system, and algorithmic transparency, and Sovereign AI (Sovereign AI) strategies such as the development of super-large Korean models and the establishment of AI data centers are also in full swing.
Private technical expertise and policy linkage structures are also strengthening. Competitiveness in the age of agentic AI depends on ‘responsible design and control’ rather than simple introduction.
Deloitte advised that corporate management should strategically create and manage a CEO checklist such as △AI governance building △ cybersecurity △ talent strategy for a successful AI transition. Above all, he stressed that the establishment of an AI governance system that considers both innovation and risk management should be recognized as an essential infrastructure for a successful AI transition. First of all, corporate CEOs must establish their own AI governance system that does not impede innovation while strategically utilizing government norms and infrastructure. “It is important to secure transparency and accountability throughout the life cycle from AI planning to operation, and to have a governance structure based on cooperation between the board of directors, technology and the field,” said Kim Jin-sook, partner of the Korea Deloitte AI Integrated Service Group. “We also need an approach to increase the level of awareness of the organization by internalizing AI ethics in corporate culture.”
The second checklist is the establishment of a security framework across the AI lifecycle. In 2023, ChatGPT operated by OpenAI exposed sensitive information from some users, including payment history. The service was temporarily suspended, causing inconvenience to users, and awareness of AI security also increased.
As AI utilization has increased, data collection, learning, and distribution have become more frequent and open source utilization has also increased, increasing risk factors in terms of security. Therefore, it is difficult to fully manage AI-specific risks with existing security strategies alone, making the establishment of a systematic security framework throughout the AI life cycle more important. “The AI life cycle is divided into stages such as planning, design and development, verification and inspection, distribution, operation, re-evaluation and renewal, and disposal, and model security, data integrity and confidentiality, access control, monitoring, supply chain and infrastructure security, and re-evaluation should be systematically applied according to the characteristics of each stage,” said Lee Chang-sung, partner of Korea’s Cyber & Resilience Integrated Service Group. “In particular, AI stability can be secured only when these security activities are organically linked.”
Finally, it is essential to create a cooperative environment between organizational members and AI. The spread of AI calls for a shift away from the existing “Human in the Loop (HITL)” method, which was directly verified and modified by humans, to the “Human on the Loop (HOTL)” method, where human intervention is minimized and AI makes decisions independently.
Accordingly, management should have a system to educate members to use advanced AI tools freely but operate them responsibly. The content of education should include clearly recognizing the role and limitations of AI. Some studies show that if the Generative AI is used within an appropriate range, the performance of high-skilled talent increases by up to 40%, while if it is used outside of this, the performance decreases by 19%. Furthermore, there is a need for an organizational culture that encourages members to try AI in a new way and share their experiences with each other. In fact, many global companies are systematically introducing AI throughout their organizations, with financial service company USAA including AI-considered talent development as the core of its employee value proposal, and Amazon systematically strengthening individual employees’ capabilities through an AI-based coaching system.
Kim Sung-jin, a partner at Deloitte Consulting Korea’s Human Capital Headquarters, suggested, “As a space where members can freely make decisions and experiments, they can also consider building platforms such as Talent Marketplace and Digital Playground where members can develop specific capabilities and exchange real-time feedback.”
All you have to do is press IT… CEO Transitions AI First
In the era of artificial intelligence (AI), Korean companies are also rushing to secure global competitiveness. According to Deloitte’s analysis, companies in key industries such as domestic manufacturing, electronics, and finance have already entered the Early Majority stage, which promotes AI Transformation (AX) strategies beyond the introduction of AI technology.
In fact, Deloitte’s domestic AI project in collaboration with global big tech partner Nvidia has reduced operating costs by 15-30% and has achieved remarkable results, with the customer recommendation index rising by 35 points. This is an example that shows that the agile execution power unique to Korean companies is also acting as a competitive advantage in the introduction of AI.
However, there are many challenges to be solved. Above all, compared to advanced AI companies such as North America and Europe, there is a gap in the level of large-scale use of AI and integrated approaches throughout the organization. Many companies often fall into the so-called “trap of PoC” where they fail to spread it across the company even though they produce results in the Proof of Concept (PoC) stage.
Chung Chang-mo, managing director of the Korea Deloitte AI Integrated Service Group (One AI), said, “Spreading AI across companies is like making a car. Just as a good car cannot be completed with a good engine alone, AI cannot succeed with algorithms alone. “Companies should create a company-wide integration strategy so that all elements, including data processing systems, computing power, and operating processes, can operate organically, including hardware infrastructure,” he said.
Amid these changes, Deloitte has proposed three action strategies for sustainable AI innovation for companies.
First, it is necessary to data the experience and know-how of experts and informal knowledge accumulated from their interactions with customers so that AI can learn.
Second, it is necessary to establish a system that can clearly measure the performance of AI introduction. Continuous investment and expansion within the organization are possible only when specific indicators such as sales growth and customer churn decrease are available.
Finally, it is essential to improve management’s AI literacy. AI is no longer just a task for information technology (IT) departments, but an area that CEOs and executives must directly understand and approach as part of their business strategies. The management must first become familiar with AI and know its possibilities and limitations to make correct instructions and decisions.
“Because technology is already mature enough, the success of AI innovation depends on organizational flexibility and efforts to data intangible assets beyond technology,” said Jeong. “2025 is expected to be the first year to create new competitiveness through core capabilities combined with AI, not whether AI is introduced or not. Now that AI changes its management paradigm, it will be the best opportunity for Korean companies to take a leap forward in the global market, he stressed.
For companies that find it difficult to build such a complex AI ecosystem independently, integrated platform strategies such as “AI Factory” are emerging. The AI factory can be a smart choice to maximize the effectiveness against investment and reduce the risk of failure by integrating and providing the entire process necessary for AI transformation, such as infrastructure construction, model learning, and operational optimization, into one system. Deloitte is developing AI solutions for businesses based on its partnership with global big tech. Deloitte is partnering with global cloud powerhouses such as AWS, Salesforce, and SAP as well as Nvidia, which makes AI chips (GPU).
Deloitte announced “Zora AI,” an agent AI platform developed in collaboration with Nvidia, at the Nvidia annual developer conference in March this year.
Meaning “goddess of dawn,” Zora AI can automate core tasks in a variety of areas of the enterprise, including finance, procurement, marketing, and sales, and is quickly deployed in both cloud and local environments and flexibly integrated with existing systems.
The reliability of Zora AI is demonstrated through practical field application cases. Deloitte’s own financial sector introduced it first, resulting in substantial cost savings. The first major external customer, Hewlett-Packard Enterprise (HPE), officially announced the introduction of Zora AI during its recent fourth-quarter earnings call.
Zora AI specializes in high value-added tasks such as reporting, data analysis, scenario modeling, and workflow management. Data security and real-time decision-making are a major advantage for companies that are important, especially in that they analyze and provide results immediately at the enterprise site without transmitting data to the outside.
“Deloit is establishing itself as the ‘AI engineering consulting firm’ that companies look for first,” said Kristen Ann Deloitte, global Nvidia Alliance partner. This is because they have not only expertise in Agentic Al, but also a deep understanding of Nvidia’s NIM software that enables corporate-tailored AI driving,” he explained.
Deloitte is working with Nvidia to design a customized AI factory for each country or company, and is working on various national projects together, including major Middle Eastern countries and Canada, he said. He added, “Korean major companies are also increasingly interested in Sovereign AI.”
Deloitte is also introducing AI solutions one after another in the audit and accounting sectors.
First of all, Deloitte Global is systematically introducing AI technology into its audit work, focusing on “Omnia,” an audit platform that has invested about $750 million over the past five years. The accounting audit division of Deloitte Group Korea is also actively applying the core technologies of its global headquarters to domestic audit sites. In addition, “Smart Review,” which automatically reviews errors or omissions in audit reports, and “Doc AI” and “Smart AI,” which extract important information from various documents, are contributing to enhancing actual audit quality. Digital audit infrastructure is being advanced based on the three-stage evolution roadmap leading to ‘tool→ solution→ integrated platform’. In addition to automating the audit sector and introducing and utilizing AI, the accounting audit sector is also actively exploring the market for corporate solutions based on financial accounting data through the “AI Asset & Analytics” group.
Deloitte announced earlier this year its plan to invest $3 billion by 2030 using the industry’s leading corporate Generative AI as its next-generation growth engine. Deloitte explained, “It is a strategic move to secure a differentiated competitive advantage in the corporate digital transformation market by leading the establishment of a comprehensive AI-based ecosystem beyond the introduction of new technologies.”
[Reporter Kwon Oh Gyun]
