Almost every major data vendor shipped something on semantic layers this month, and that’s not a coincidence – it’s a signal.
As AI becomes part of everyday work, the quality of your decisions will depend on the quality of the data underneath them. Without trusted definitions, governed metrics, and shared business context, AI will scale confusion as quickly as it scales productivity.
📖 You're on Databricks. Do You Really Need dbt Too?
Short answer: yes, but only if you're clear on why. Our Solution Architect Keith Ludeman was invited by dbt to write a guest post unpacking the architecture decisions that determine whether these two tools work together or quietly compete for the same real estate in your stack.
Here's what you'll take away:
Where the tools overlap but what each one does better
How to give each technology a clear role in your infrastructure (Databricks for landing, storing, and scaling raw data; dbt for structure, testing, documentation, and governed transformation)
Tips on bringing sprawling transformation logic across notebooks, scripts, and workflows into single place
When to consider having both in your tech stack
“The teams that scale best aren't the ones with the most tools. They're the ones who decide early where transformation happens, where governance lives, and how business logic gets managed — and then stay disciplined about it.”
How to build a semantic foundation your AI can trust
A semantic layer isn't just a technology decision. It rests on a foundation of business definitions, metric logic, lineage, and quality checks that must be made consistent and usable across the stack.
Our Data Strategy Practice Lead, Christina Salmiput it this way:
“You can have the technology for a semantic layer, but if you haven't defined and organized your context into a sound semantic model, the layer has nothing meaningful to surface.”
The meaningful solution isn't standing up a semantic layer. It's building the semantic foundation (which includes context and semantic models) underneath it — so every tool, team, and agent consuming your data gets the same answer.
Join us on May 6 for Closing the AI Readiness Gap, Part 2: The Semantic Layer Imperative. Christina Salmi and Matthew Mullins (Coginiti) will cover where the semantic layer fits in your data lifecycle, how teams are embedding governance and lineage, and five practical steps to build a foundation that gets adopted.
ThoughtSpot released an AI-native semantic layer built for agent consumption, with a token-based approach to human-verified definitions that doesn't require SQL review. They also joined the Open Semantic Interchange (OSI) as a founding partner pushing for cross-vendor semantic standards.
The practical implication: The OSI bet is more interesting than the product itself. Interoperable standards would shift the question from "which vendor's semantic layer?" to "which definitions travel across vendors?"
Power BI Positions the Semantic Model as the Backbone for Fabric IQ
Microsoft rolled out updates positioning Power BI semantic models as the foundation of Fabric IQ — including Direct Lake on OneLake (GA), Translytical Task Flows (GA), and a TMDL View in the browser (preview) for code-first semantic modeling.
The practical implication: For Fabric customers, TMDL-in-browser brings Git-style versioning into semantic modeling. For multi-platform teams, the real question is how much business logic you want tied to one vendor's roadmap.
Databricks Publishes a Semantic Layer Architecture Guide
Databricks published a detailed guide arguing for platform-native semantics over tool-bound semantics (DAX, LookML, VizQL), anchored by Unity Catalog Business Semantics — definitions that sit alongside data and governance.
The practical implication: Read it even if you aren't on Databricks; send it to your data engineering lead. The "platform-native vs. tool-bound" framing will shape a lot of architecture decisions this year — though the question worth pressing is whether "platform-native" is broad enough for multi-platform enterprises.
Sigma Joins Atlan as a Context Layer Partner
Sigma joined Atlan as a Context Layer Partner. Atlan's Enterprise Context Layer now feeds certified definitions, lineage, ownership, and data quality signals directly into Sigma workbooks, Sigma Assistant, and Sigma Agents.
The practical implication: The "context layer across tools" pattern — Atlan sitting across warehouse, dbt, and BI — is the most architecturally flexible model in this month's batch. Worth studying even if you don't use Sigma or Atlan.
.png?width=1080&upscale=true&name=Subtitle%20(1000%20x%20800%20px).png)
