The way companies utilize data is shifting from verifying “what happened” to determining “what needs to be done” and executing it. This is due to the emergence of “Agentic Analytics” as a new method of data utilization, which goes beyond generative AI analyzing data in natural language to provide answers, to understanding the company’s context and supporting subsequent actions.
Salesforce’s data analytics platform Tableau held ‘DataFam Seoul 2026’ on the 20th at the Grand Intercontinental Seoul Parnas in Samseong-dong, Seoul, focusing on these changes in the data analytics environment. Approximately 1,000 people, including industry leaders and data experts, attended this year’s event.
This year’s theme is “Beyond Agentic Analytics, Toward Action.” The key point lies in the fact that AI is evolving from an analytical tool that reads data and finds insights into an agent of execution that supports corporate decision-making and actual business operations.
Kim Young-kyun, Head of Tableau Business at Salesforce Korea, and Matthew Miller, Vice President of Global Product Management at Tableau, delivered keynote speeches introducing the data analysis environment in the AI era and Tableau’s strategy. Vice President Miller explained that connecting insights gained from data to organizational decision-making and subsequent business operations is a crucial role for analytics platforms going forward.
For AI to ‘understand’ data… it must first understand the company’s context.
The premise of agentic analytics is that AI does not stop at simply searching corporate data, but understands the meaning of the data and its business context.
The technologies Tableau has put forward for this purpose are the Knowledge Engine and the Decision Engine. The Knowledge Engine enables AI to generate answers and insights based on the company’s data and business context, while the Decision Engine connects these to the judgments and subsequent actions of AI agents.
Here, a Semantic Layer is used to enable AI to understand the meaning and relationships of data that differ from company to company. Since even the same metric like ‘revenue’ can have different calculation standards and usage methods depending on the company or organization, the approach is that for AI to properly utilize data, it must understand not only the numbers themselves but also how those numbers are used within the company.
This change also expands the scope of data analysis users. It shifts from a model where professional analysts create dashboards and business users review them to a structure where users can directly ask questions about data using natural language, receive answers, and proceed to perform necessary tasks.
Salesforce demonstrated the process of handling everything from data preparation to analysis and dashboard construction using natural language on-site with Tableau Agents. They also introduced a method to connect with external LLM and AI agents, such as Claude Code,. The goal is to connect data with work environments, including Salesforce and Slack, to ensure that analysis results translate into actual workflows
From 5,700 dashboards to data agents within Slack… How are companies using them?
At the event, Olive Young, LG CNS, Toss Bank, Krafton, Barofarm, and others shared their real-world experiences with data utilization. The company case studies focused not only on AI applications but also on the foundations necessary to implement agentic analytics, such as data consistency, self-service analytics, and governance.
Olive Young introduced a case study demonstrating the evolution of self-service analytics, where employees directly utilize data, into an AI data agent based on Tableau MCP. By connecting verified data with AI, the company enabled employees to access information necessary for period-by-period sales comparisons or performance analysis using natural language within existing work environments, such as Slack.
Toss Bank unveiled the process of enhancing its data lineage and permission management systems, along with automation experience accumulated while operating over 5,700 dashboards and 362 projects. It highlighted that as the scale of the analytics environment grows, a system for managing who uses what data, where data is generated, and what processes it has gone through also becomes crucial.
LG CNS introduced an analytics environment that supports decision-making by integrating purchasing data, ERP, and internal and external data. Krafton focused on data governance for consistently managing business metrics and context, while Baropharm concentrated on streamlining pharmaceutical data analysis through Tableau-based data preprocessing.
As AI lowers the barrier to data access, data reliability and management systems actually become even more critical. This is because even if anyone can perform analysis using only natural language queries, rapid analysis cannot lead to accurate decision-making if the definitions of the data and metrics referenced by the AI are incorrect.
The next step after ‘viewing data’ is ‘acting data’
A recurring change at this event was reducing the gap between analysis and execution. While traditional Business Intelligence (BI) focused on visualizing data and providing grounds for human judgment, agentic analytics expands its scope by having AI intervene in the analysis process to suggest next steps and connect with business systems.
Kim Young-kyun, General Manager of Tableau at Salesforce Korea, said, “As AI enables anyone to quickly analyze data and find answers, corporate competitiveness will not come simply from possessing more data or performing more sophisticated analysis. The new standard for data capability will be how quickly and accurately insights are translated into action based on reliable data and clear business context.”
This means that simply making data analysis easy is not enough. ‘Acting data’ becomes possible only when common meanings and standards are assigned to the vast amount of internal data, enabled for AI to understand, and connected to actual business systems.
The event also featured programs that presented data analysis as a more everyday experience. In the finals of the data visualization competition ‘Viz Games,’ participants competed under the theme ‘Tableau is Food After Dining,’ analyzing real-time data from general restaurants nationwide to create dashboards and stories.
General Manager Kim stated, “The role of agentic analytics is expanding beyond simply showing past status to suggesting next steps and supporting the actual execution of tasks.”
We are moving beyond the era where AI merely answered analysts’ questions and are now transitioning to a stage where it understands data within the enterprise and connects it to the next task. Ultimately, the competitiveness of agentic analytics is not determined solely by the AI model itself. The reliability of the data a company possesses, and the accuracy with which that data captures the organization’s language and business context, form the foundation for transforming ‘analyzing AI’ into ‘acting AI.’
