AI is more than just a tool that provides answers; it is a 'work colleague' that understands the actual business context and rules of a company and takes charge of execution.

Workday, an AI platform company for corporate HR, finance, and agents, held Workday Elevate Seoul 2026 at the Westin Seoul Parnas in Gangnam-gu, Seoul on the 14th and presented a vision of a 'hybrid workforce' where humans and AI agents work together.

At a press conference held as a side event to the day, Heo Jeong-yeol, the newly appointed Country Manager of Workday Korea earlier this year, assessed that Korean companies have already moved beyond the AI experimentation stage and are entering the enterprise-wide adoption stage.


Branch Manager Heo Jeong-yeol includes manufacturing, semiconductor, and platform companies, among others.It was explained that domestic companies are re-evaluating existing growth drivers amidst geopolitical changes and the spread of AI, and that innovation based on AI, data, and next-generation technologies is gaining momentum.

However, it was warned that while the approach taken by many companies to collect source data in data lakes, connect standalone large-scale language models with proprietary tools, and implement new business operational logic is useful for initial experiments, it could lead to a risk known as 'shadow ERP' if applied to core systems.

Branch Manager Heo Jeong-yeol emphasized two conditions for enterprise AI.

One is probability-based reasoning, where AI has a strength—that is, the ability to generate ideas, understand human intentions, and predict next actions.

The other is a deterministic execution system that accurately handles failure-free tasks such as closing accounts, payroll, approvals, and compliance.

Branch Manager Heo Jeong-yeol explained, “For a secure enterprise AI to be built, these two key components must be integrated within a single system,” and pointed out that adding AI solutions externally after the fact can shift the risk to the customer.

Workday proposed a platform approach based on a single data model, a single security model, and a single architecture as a solution to this problem.

Branch Manager Heo Jeong-yeol explained that Workday has been internalizing HR, finance, payroll, and other functions into a single core system for the past 20 years, rather than attaching them as needed.

Thanks to this structure, data and logic are not separated, and the approval systems, compliance, and audit trails that customers have built with Workday become assets for AI governance.

In particular, within Workday, AI agents run only within a scope equal to or lower than the user's authority, and who, in what context, and whatHe emphasized that it remains as an immutable audit record whether [it] was ordered.

Josh Zwen, Vice President of Global Solutions Marketing at Workday, who visited Korea on this day, also shared the same concern.


Vice President Josh Zwen cited the difference between 'general-purpose AI' and 'context-aware AI' as the reason why AI is not being widely adopted in companies.

The explanation is that while general large-scale language models are strong at providing plausible answers to questions, they have limitations in tasks that require absolute accuracy, such as salary, benefits, and personnel decisions.

Vice President Josh Zwen stated that while corporate business processes are deterministic processes with clear beginnings and endings that must be executed accurately every time, general-purpose AI reason probabilistically, so '95% accuracy' is not sufficient.

The differentiator presented by Workday is the combination of 'context, process, and the right path'.

Josh Zwen, Vice President, explained that Workday already operates core business processes on an integrated data model, enabling it to understand business contexts such as organizational charts, policies, privacy, and regulatory requirements.

In addition, it was stated that all actions performed by the agent must pass through pre-established compliance and business processes, and that the agent cannot arbitrarily interpret or bypass them.

He said this structure is an uncontrollable 'lawless agent'He explained that it enables a 'legitimate agent' that complies with regulations, rather than ;

Workday unveiled 'Sana from Workday' as a core platform to realize this vision.

Sana is an integrated AI experience running on Workday, designed to enable employees to ask questions, request tasks, and even perform automation in natural language.

According to Vice President Josh Zwen, Sana consists of a new AI interface called 'Sana for Workday', 'Sana Self-Service Agent' which supports the automation of HR and finance workflows, and 'Sana Enterprise' which extends AI capabilities across the enterprise.

Sana's core functions are summarized into four categories: exploration, execution, build, and automation. Employees can obtain clear-source answers based on internal information and Workday data, such as remaining vacation days or contract amounts, and execute tasks like changing addresses or updating contract amounts within the authorization system.

Administrators can create dashboards that display hiring pipelines and interview feedback, and can also automate multi-step workflows, such as receipt review, policy checks, and sending pre-approval reports, using a no-code approach.

Workday stated that Sana does not stay within Workday but also connects with external work tools.

Sana Enterprise provides connectors with various enterprise applications such as Gmail, Outlook, Salesforce, SharePoint, Slack, Zoom, Jira, and Notion, enabling employees to work in a single interactive environment without having to search multiple systems individually.

They also explained that Workday Build and Workday Data Cloud provide a foundation for customers to create their own agents and utilize data from various systems without the need for separate data replication.

Workday suggested learning, recruitment, self-service, and field support as areas of practical application.

Workday Learning uses AI to help structure and design personalized learning content, while Candidate Experience Agents automate the hiring process, including answering questions and scheduling interviews.

Sana self-service agents provide personalized answers to complex HR questions and handle administrative tasks, while field agent agents support the task of finding replacement personnel in the event of absences.

Vice President Josh Zwen also mentioned cases of increased productivity and reduced HR cases resulting from the automation of the recruitment process and the application of self-service agents.

Branch Manager Heo Jeong-yeol stated that Korean companies can become first movers in AI agentic transformation as well.He looked ahead.

He predicted that repetitive and routine tasks would decrease with the help of AI, allowing employees to focus on more valuable work with AI teammates who understand business context.

Vice President Josh Zwen also emphasized that the new future of work Workday envisions is an environment where humans and AI agents collaborate in a safe and transparent manner to complete tasks.