Workflow Recap: Bridging LLMs and Analytics with Sigma AI Query

At Workflow 2026, a session was conducted on bridging the gap between large language models (LLMs) and structured data workflows using AI Query within Sigma workbooks, emphasizing the shift from deterministic to probabilistic systems. Traditional analytics rely on deterministic logic yielding consistent results, while LLMs operate probabilistically, allowing for dynamic summarization and interpretation of complex data. AI Query, running natively within data warehouses like Snowflake or Databricks, enables secure and efficient data processing without exporting sensitive data, offering applications such as dynamic summaries and AI-powered dashboards to facilitate data interpretation for non-technical stakeholders. Examples include AI-driven sales forecasting tools and portfolio modeling apps that provide context-aware insights directly within Sigma, enhancing decision-making without altering existing workflows. The session highlighted the potential for AI agents to interact with analytical interfaces, thus strengthening data workflows by complementing traditional analytics with AI capabilities, and underscored the importance of maintaining security and governance standards by processing data within the warehouse environment.