Naming conventions are the thing every Fabric project starts with good intentions and abandons by month three. The fix is not better discipline. It is AI as a deterministic enforcement layer. Here is the standard I use on every project, in five rules.
Most of what is written about AI in data work falls into two buckets: AI will replace developers within two years, or AI is glorified autocomplete. Both are wrong. After months of daily use on Microsoft Fabric engagements, the honest answer is that AI is not a people replacement but a power multiplier, and like…
A reader asked the right question after our last post on three-layer pipeline monitoring: what does it cost in CUs? Microsoft has not published authoritative figures, and the Capacity Metrics App will not separate monitoring overhead from everything else your Eventhouse is doing. The only way to get a real number is to measure. Here…
Out of the box, Microsoft Fabric pipelines fail in ways that are easy to miss. After a couple of years shipping Fabric platforms into production, we now deploy the same three-layer monitoring pattern on every engagement: collect with Workspace Monitoring, diagnose with saved KQL queries, notify with tiered Data Activator triggers.