Agentic Dimensional Modeling

About this book

What happens to dimensional modelling when the consumer is no longer only a human analyst, but an AI agent that can recommend, decide and act?

Dimensional modelling has long turned operational data into understandable facts, dimensions and measures. That foundation remains essential. But much of the meaning humans once supplied implicitly can no longer stay implicit. An agent needs to know what a business term means, which source is authoritative, what was known at the time of a decision, which policy applied, whether evidence was sufficient, who had authority to act and whether an attempted action became a real outcome.

Rather than replacing dimensional modelling, the book introduces Agent-Ready Dimensional Modeling (ARDM), which preserves grain, facts, dimensions, conformance and history while extending the analytical contract for machine consumers.

The book shows how to model and govern

• business events, state and machine decisions at the correct grain
• semantic contracts that make enterprise meaning explicit
• decision-time knowledge separately from current truth
• identity, MDM, source authority, provenance and evidence
• agent recommendations, decisions, approvals, attempts, commitments and outcomes as distinct analytical events
• semantic layers, knowledge graphs, vector retrieval and operational APIs without forcing every concern into the warehouse
• context gateways that serve governed business context to agent runtimes
• freshness, correction, replay, testing, observability and reconciliation
• migration from an existing BI or lakehouse estate to an agent-ready analytical platform

A continuous Atlas Industrial Group case runs across 22 chapters in four stages: Define the business problem → Define the solution → Design the model → Deliver the platform. A downloadable companion pack provides ten ARDM workbooks with inputs, practitioner exercises, worked solutions and design rationales.

This is not a book about adding an AI layer on top of a warehouse. It is about redesigning the analytical contract so enterprise data stays understandable, reconstructable and governable when software becomes part of the decision process.