Your dashboards show you the number. Who explains the why?
Every BI system built in the last twenty years was designed to answer "what happened." The AI era demands more: systems that investigate causes on their own, explain their answers, and act — without losing the discipline that made classical BI trustworthy. This book builds that bridge, end to end.
AI-Native Business Intelligence, Volume 1 is a complete, practice-first course in three parts:
PART I — Classical BI, rebuilt properly. Inmon vs. Kimball, dimensional modeling, slowly changing dimensions, ETL/ELT, data quality, MDM, and governance — the foundation your future agents will stand on, taught through one continuous retail case.
PART II — The AI-Native architecture. The six-layer reference model with the semantic layer at its heart: metrics as code, MCP as the controlled door, conversational and agentic BI, RAG and knowledge graphs, hallucination control, compile-time security, and FinOps for agent workloads. You'll learn why "wrap an LLM around the database and hope" fails — and what to design instead.
PART III — Transformation in practice. Greenfield, brownfield, and hybrid playbooks with roadmaps, risk registers, and rollback criteria — capped by a full multinational, multi-currency capstone (Truong Son Group) you can rebuild for your own organization, exercises and reference solutions included.
Inside this volume
- 35 chapters, 60+ original architecture figures, and chapter quizzes with full answer keys
- Ready-to-adapt templates: semantic-layer YAML, MCP server configuration, eval suites with "trap" questions, Bus Matrix, design checklists
- A 90-day / 6-month / 12-month migration playbook with quantitative stop-and-rollback criteria
- 30/60/90-day self-study paths for newcomers, BI veterans, and decision-makers
Written for data engineers, architects, analytics leaders, and CTOs who refuse to choose between AI capability and engineering discipline. No prior AI experience required; every concept is built from the ground up.
Volume 1 is the thinking book — what to build and why. Volume 2 (Design & Implementation) is the building book. They pair, but each stands alone.