While AI transformation (AX) dominates conversations across industries, many enterprises on the ground still struggle to trust AI-generated analysis. Repeated hallucinations — where AI misinterprets data or fabricates information that doesn’t exist — are a key culprit. In a March survey by Anthropic, “errors caused by hallucination” ranked as the No. 1 AI concern among respondents.

Against this backdrop, Tableau, the data analytics and visualization platform under Salesforce (CRM), is pushing back on fears of a so-called “SaaSpocalypse” — the notion that AI agents will replace the software industry wholesale. Matthew Miller, vice president of product management at Salesforce Tableau, told the Seoul Economic Daily at Salesforce’s South Korea office in Yeongdeungpo-gu, Seoul on the 19th: “General-purpose AI services built on large language models (LLMs) give slightly different answers every time you ask. Tableau does not.”

Tableau, founded in the U.S. in 2003 as a data analytics and visualization platform, was acquired by Salesforce in 2019. Once data is entered, the platform performs analysis according to predefined rules and customer-specific regulations. Miller defined Tableau’s technology as “deterministic AI” that enables consistent answers, emphasizing that “tokens are only used when initially creating dashboards and reports, so costs are lower than AI agents.”

“Agent Proliferation Is Technical Debt”

Miller argued that the trend of non-experts coding their own agents could actually become a burden for enterprises. “Software isn’t a one-and-done build — continuous maintenance and security upkeep are essential,” he said. “If agents that function as software are mass-produced without governance, enterprises end up accumulating ‘technical debt.'” His point: agents may replace simple services like household budgeting apps, but enterprise-grade services requiring high regulatory compliance and security remain out of reach.

Tableau’s “semantics” framework, which standardizes tacit data knowledge, is another key strength. “Many companies fail to properly leverage their data because different departments and industries define terms differently,” Miller explained. “Tableau works with customer executives and business units to standardize terminology definitions and enable context-appropriate data analysis.”

For example, the term “conversion rate” is interpreted by marketing teams as advertising and campaign performance, while finance teams read it as foreign exchange rates. Companies also differ in what they include as revenue. Tableau inputs these tacit knowledge elements into the system’s backend to drive accurate analysis.

Evolving Toward Agentic Analytics

In his keynote at “Salesforce Tableau Datafam Seoul 2026,” held at the Grand InterContinental Seoul Parnas in Gangnam-gu, Seoul on the 20th, Miller announced that Tableau is “evolving beyond simply deriving insights through analysis into an ‘Agentic Analytics’ platform that connects data to real business outcomes.”

According to Salesforce, Tableau’s knowledge engine and decision engine underpin this transformation. The knowledge engine helps AI deliver more accurate answers and insights based on enterprise data and business context, while the decision engine enables AI agents to perform actual judgment and actions. Tableau’s “semantic layer” also supports AI agents in understanding the meaning and context of each dataset tailored to industry and enterprise-specific business characteristics, boosting AI reliability and accuracy.

On-site demonstrations showcased Tableau agents executing the entire workflow — from data preparation to analysis to dashboard building — using natural language. The event also introduced agentic analytics workflows that connect Tableau’s MCP (Model Context Protocol) with external LLMs and AI agents such as Anthropic’s Claude Code to link analysis to execution.

South Korean Enterprise Adoption Cases

The event also highlighted adoption cases from major South Korean companies. Olive Young presented how it built a field-driven data utilization culture based on self-service analytics, evolving it into an AI data agent powered by Tableau MCP. LG CNS built an analytics environment supporting data-driven decision-making by integrating procurement data, enterprise resource planning (ERP), and internal and external data. Adoption cases from Toss Bank, Krafton, and Barofarm were also shared.

CompanyTableau Use CaseOlive YoungBuilt field-driven data utilization culture based on self-service analytics; evolved into Tableau MCP-based AI data agentLG CNSIntegrated procurement data, ERP, and internal/external data; built data-driven decision support environmentToss BankShared Tableau adoption statusKraftonShared Tableau adoption statusBarofarmShared Tableau adoption status

Kim Young-kyun, head of Tableau business at Salesforce Korea, said: “As AI enables anyone to analyze data quickly and get answers, enterprise competitiveness will no longer come simply from holding more data or performing more sophisticated analysis. The new benchmark for data capability will be how quickly and accurately insights are connected to execution based on trustworthy data and clear business context.”

He added: “The role of agentic analytics is now expanding beyond simply showing past performance to recommending next actions and supporting actual work execution.”

Miller concluded: “Just as Steve Jobs once likened computers to ‘a bicycle for the mind,’ AI and technology maximize human capability. When humans and AI learn from and leverage data together, organizational capability becomes powerful enough that 1+1+1 equals not 3, but 17.”

Datafam Seoul is Tableau’s flagship annual event, serving as a forum for sharing the latest data analytics trends and real-world use cases. This year’s event, themed “Beyond Agentic Analytics, into action,” drew approximately 1,000 industry professionals and data experts.