When I yell at correct my AI agent, it’s not because it was wrong. It’s just… not quite right.
That’s because it’s not pulling from years of business context, definitions, and “well, that number depends…” logic sitting in my head.
That’s the part businesses are underestimating right now.
AI can only work from the meaning your organization has made explicit and consistent across systems. And in a lot of companies, that meaning still lives in dashboards, spreadsheets, YAML files, meeting history, and various brains.
This month, we get into semantic modeling and the semantic layer – which is exactly what you need to get right for AI to work reliably at scale.
We sat down with Christina Salmi, VP of Data Strategy at Analytics8, to talk through what a good semantic foundation looks like and how to get it right.
Here are a few of the takeaways:
1. Get the sequence right: context, then model, then layer.
“Semantic layer” has become one of those phrases everyone uses slightly differently.
The distinction matters more now because AI depends on semantic consistency.
Context is the business meaning behind the data: definitions, rules, calculations, relationships, and organizational nuance.
The semantic model is how that context gets structured.
The semantic layer is what makes it accessible across dashboards, tools, applications, and AI systems.
In other words, the layer is not where meaning starts. It is where meaning gets distributed.
“You can have the right 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.”
2. Even if your data platform offers native semantics, you still need to build a sound semantic model.
Platform-native semantics are a starting point, not a solution. When definitions are locked to a single platform, consistency breaks down the moment your data estate spans more than one tool.
Universal semantic tools like Coginiti are built specifically to sit across your existing infrastructure, whether that’s a single platform or a combination of several — Databricks, Snowflake, Google BigQuery, on-prem systems, and more. You define your business logic and context once, and it flows consistently to every platform, analytics tool, AI agent, and consumer downstream.
3. Defining shared definitions means alignment between people who rarely agree.
And that’s okay. Sometimes the same metric goes by two different names, and the fix is simply adding context that maps them together. Other times, definitions are genuinely different, and the right answer is to document both clearly and stop treating the difference as a problem.
An effective solution for conflict resolution is to bring everyone together and walk through the actual logic behind the metric. When people see all the layers of calculation that go into producing a number, a lot of the tension resolves on its own.
Closing the AI Readiness Gap, Part 2: The Semantic Layer Imperative
Catch the recording
Matthew Mullins, CTO and co-founder of Coginiti, a Semantic Intelligence platform, joined our latest live stream about what it takes to make a semantic layer act as a governance layer for AI… instead of just another serving layer for dashboards.
