The Defensible Decision

Direct the AI. Don't accept whatever it generates.

Power BI Copilot, AI visuals, and natural-language Q&A have collapsed the time between question and chart from hours to seconds. The bottleneck is no longer the tool. It is the analyst's ability to direct the AI rather than accept its first draft.

The Defensible Decision is the missing layer between "I know what good visualization looks like" - Kirk, Knaflic, Tufte, Few, Wexler - and "I have an AI that can build it for me." It teaches the principles of visualization first, then layers the discipline of AI direction on top.

Part I - Principles First. Six chapters covering why principles still matter when AI builds the chart, then the five Part-I principles in turn: context, story, chart selection, focus, and color/accessibility. Built by hand, the way the canon teaches. No AI yet. The principles that make a chart defensible do not change when the tool does.

Part II - AI Direction. Seven chapters on Dialog Engineering and the CSAR loop (Crystallize, Synthesize, Augment, Refine). How to direct Copilot to generate reports you can defend. How to recognize when AI misleads. How to build dashboards that survive the executive's first hard question. How to apply Microsoft's Responsible AI principles to the chart on your screen, today.

What you can do after reading this book:

• Critique an AI-generated chart against the principles that make it defensible
• Run a CSAR-loop conversation that ships a directed report, not an accepted one
• Diagnose the four common failure modes of AI-generated dashboards before your audience does
• Apply the Provenance Footer, the Four-Role Decision-Boundary, and the chart-adapted ACT pass to your own work

Who this book is for: Working analysts and mid-career professionals who already know what good visualization looks like and now have to direct an AI to produce it. MBA students whose curriculum included the canon but who graduated into a workflow that did not exist when the canon was written. Anyone whose job description still says analyst but whose actual work is shifting toward director.

What this book is not:

• Not a Power BI tool manual - those age out every quarter; this book stays at the directional level and points to Microsoft Learn for technical depth
• Not a generic prompt-engineering book - every example is a directed AI conversation in business analytics, not a writing or coding prompt
• Not a replacement for the canon - it cites Kirk, Knaflic, Tufte, Few, and Wexler and builds on top

Companion resources at thedefensibledecision.com:

• Chart Gallery - 50+ chart types across 10 analytical-intent categories, with interactive demos and color-blind-safe palette tooling
• Prompt Library - the seven canonical CSAR prompts referenced throughout Part II
• Datasets - four downloadable datasets matching the worked examples in Chapters 4, 8, 10, and 12
• Appendix F field guide - the Business Requirements Document template for AI-assisted analytics work
• Per-chapter resources, cited works, and current Microsoft documentation pointers

About the author. Dr. Fabio Correa, DBA, is an independent author, AI Transformation Leader (Microsoft Certified, 2026), and graduate MBA instructor in Business Visualization Analytics. He teaches working analysts to direct AI rather than be replaced by it.

Part of the Applied AI series from CorreaX Publishing - books on the discipline of directing intelligent systems in real work. Companion titles: The Verification Habit (the cognitive science of AI-assisted decision making) and Artificial Critical Thinking: The Ten Tenets (the framework underneath the discipline).