Agentic Dimensional Modeling

What happens to dimensional modeling when the consumer is no longer only a human analyst—but an AI agent that can interpret data, make decisions and act? Dimensional modeling gave enterprises a durable way to turn operational data into understandable facts, dimensions and business measures. That foundation still matters. But agentic systems expose a new problem: much of the meaning humans once supplied through experience, context and judgment can no longer remain implicit. An AI agent needs more than a technically correct star schema. It needs to know: what a business term actually means which source is authoritative what was known at the time a decision was made which policy applied whether evidence was sufficient which identity or entity was involved who had authority to act whether an attempted action became a committed business outcome Agentic Dimensional Modeling introduces a practical architecture for meeting those requirements. At the center of the book is Agent-Ready Dimensional Modeling (ARDM)—an approach that preserves the strengths of dimensional modeling while extending the analytical contract for machine consumers. Rather than replacing facts, dimensions, grain, conformance and history, ARDM asks a more important question: What must become explicit when analytical data is being consumed by software that can reason, recommend, decide and act? The book develops concrete patterns for: modeling business events, state, agent decisions and outcomes at the correct grain turning dimensions into governed semantic contracts separating current truth from what was knowable at decision time representing identity, MDM, source authority, provenance, quality and evidence distinguishing recommendations, decisions, approvals, attempted actions, committed actions and outcomes integrating dimensional models with semantic layers, knowledge graphs, vector retrieval, documents and operational APIs serving governed context to AI-agent runtimes designing for freshness, correction, replay and historical reconstruction testing and observing agent-ready analytical systems evolving an existing warehouse or lakehouse into an agent-ready analytical platform One of the book's most important design principles is that not every AI requirement belongs inside the dimensional model. It shows where dimensional modeling should remain authoritative—and where responsibility should instead belong to MDM, semantic layers, knowledge graphs, vector stores, document retrieval, policy services or operational systems. The book uses the continuous Atlas Industrial Group case throughout its 22 chapters. Inventory, supplier, revenue and service scenarios recur as the architecture develops, allowing the reader to see how grain, semantics, identity, time, authority, evidence, decisions and outcomes interact across one coherent enterprise environment. The book is organized around four practical stages: Define the Business Problem ? Define the Solution ? Design the Model ? Deliver the Platform A downloadable Reader Edition companion pack extends the book with ten ARDM workbooks containing supplied inputs, practitioner exercises