AI Didn’t Kill ETL — It Made Data Modeling More Important

Is ETL dead in the AI era? Not even close. In this Guy in a Cube video, Patrick LeBlanc responds to a great community discussion around the AI-native data stack, ETL, data modeling, semantic models, medallion architecture, and what AI agents really need from your data. This conversation was sparked by a Reddit thread in the Microsoft Fabric community:   / ejv0hqetd2   Patrick is going to show you why AI does not remove the need for ETL, data warehouses, or modeling. In fact, AI makes modeling even more important. The real shift is not that ETL is dead. The shift is that ETL should not always be the automatic answer. Patrick breaks down why messy transactional systems still need clean, trusted, business-ready data, why semantic models and context matter more than ever, and why pointing agents directly at raw source data is not architecture. It is gambling. In this video, we want to separate the AI hype from the real architecture conversation: Why ETL still matters Why data modeling is not going away Why semantic models are critical for AI and agents Why copying data should be an architectural decision Why agents need context, meaning, boundaries, and trusted data If you are working with Microsoft Fabric, Power BI, data warehouses, lakehouses, medallion architecture, semantic models, or AI agents, this is a conversation you do not want to miss. See you in the Cube. 📢 Become a member: https://guyinacu.be/membership ******************* LET'S CONNECT! ******************* --   / guyinacube   -- https://bsky.app/profile/guyinacube.b... --   / guyinacube   --   / guyinacube   #PowerBI #MicrosoftFabric #GuyInACube