In this special rant episode, Sonny Rivera joins Juan and Tim during a live recording and rant about AI-generated dashboards, semantic layers, and what it takes to trust an answer. The situation: dashboards slop is easy to build which makes foundation matter more, not less. We discussed: - Building a dashboard was never the hard part. Being able to explain where a number came from is. - Speed without a governed foundation produces wrong answers faster. As Sonny put it, “fast and wrong is still wrong.” - Modeling data is really modeling the business. The job has not changed. How it gets done is changing. - A single source of truth did not hold up. Sonny favors an interoperable set of semantic layers and context, with shared definitions where people actually care. - Logic written in the moment, in a dashboard or a pipeline, often never makes it back into the semantic layer. Closing that loop is the gap to fix. - Friction between teams is a sign that a definition matters. Make the differences visible rather than forbidding them. - Trust is what data leaders can least afford to lose, which is why testing belongs in dashboards too. - Writing code is no longer the slow part. Product management and testing are becoming the bottlenecks.