Momentum One Academy · lab
Prepare a Semantic Model for AI
This lab focuses on preparing a retail-sales semantic model for natural-language analysis. Rather than treating AI readiness as a single switch, it examines the model metadata, business context and validation steps that help Copilot interpret questions consistently.
What the lab covers
- Choosing which business fields should be visible to AI-assisted experiences.
- Adding instructions that explain business terms, measures and data scope.
- Mapping common questions to verified report answers.
- Adding synonyms so familiar business language resolves to the intended model fields.
- Reviewing Copilot responses and diagnostics before marking the model approved.
Learning outcome
Participants can describe how schema simplification, model instructions, verified answers and synonyms influence natural-language analysis. They also learn to review generated responses and diagnostics instead of assuming that a fluent answer is automatically correct.
Workshop connection
This overview supports Module 08 — Direct Lake, Semantic Models, Copilot & Data Agents. It adds a model-governance checkpoint between creating a semantic model and exposing it to AI-assisted business users.
Access note
This exercise requires a paid Fabric F2-or-higher capacity and Copilot access; Fabric trial capacity does not support the capabilities used. It also requires Power BI Desktop November 2025 or newer. Features and interface labels may vary because parts of the workflow are identified as preview. Full instructions and the starter model are provided during the workshop; this page does not publish them.
Further learning
Plan your workshop