During the development of a Power BI project, testing the semantic model often takes a back seat, either because of time constraints or the effort required to validate it manually. We often assume everything should work as expected, but that does not guarantee data quality, referential integrity, or that metrics return the expected results.

In this session, we will explore how to introduce automated and reusable tests into the development of our semantic models using PQL.Assert, an assertion library that allows us to define and standardize different types of validations across the model.

We will use Semantic Link in Microsoft Fabric to automate the execution of these tests, and we will see how GitHub Copilot CLI can help us go one step further: comparing metrics against their business context and validating their results against the expected values from the source.

A practical session focused on bringing testing into semantic model development, reducing manual validation and increasing confidence in our models before they reach production.