The organizations that have had the discipline to continue having intentional data modeling, or better the blast from the past role of data modeler (lucky enough to prove value to have data modeling as part of your SDLC), are so far ahead with data meaning and context. The data modeler/modeling tools collected the terms and definitions in the boring 'data dictionary'. As well as relationships, domains, data types, technical and business names. The data model metadata perfect for AI ingestion to understand 'context'. Term has 1:M to attributes, attributes 1:M to tables, tables have 1:M to databases... reusability and relationships. Great food for AI. And frankly, what is needed for a 'Semantic Layer'. Can 'AI' give us the ROI to have a data modeling renaissance? As using AI with Data Modeling, but it being a driver for the magical semantic layer.
Spot on, totally makes sense how this universal semantic layer thing is falling short, and it's kinda wild to think how much more training an AI needs to actually get human intent and all it's messy, subjective context.
The organizations that have had the discipline to continue having intentional data modeling, or better the blast from the past role of data modeler (lucky enough to prove value to have data modeling as part of your SDLC), are so far ahead with data meaning and context. The data modeler/modeling tools collected the terms and definitions in the boring 'data dictionary'. As well as relationships, domains, data types, technical and business names. The data model metadata perfect for AI ingestion to understand 'context'. Term has 1:M to attributes, attributes 1:M to tables, tables have 1:M to databases... reusability and relationships. Great food for AI. And frankly, what is needed for a 'Semantic Layer'. Can 'AI' give us the ROI to have a data modeling renaissance? As using AI with Data Modeling, but it being a driver for the magical semantic layer.
Can I add my favorite one. we updated the object but never updated the description to reflect the update <eyeroll!!!>
Oh my god yes
Spot on, totally makes sense how this universal semantic layer thing is falling short, and it's kinda wild to think how much more training an AI needs to actually get human intent and all it's messy, subjective context.