Summary

Simon Späti argues that existing semantic layers — while good at centralizing metric definitions — force teams to build extensive glue code for APIs, integrations, and agent tooling that should ship out of the box. The rise of AI agents adds yet another layer of custom wrappers needed on top of any existing semantic layer. He introduces SLayer (by Motley) as an OSS answer: an agentic-first semantic layer that bundles context primitives, an expressive query engine, and agent-agnostic MCP tooling under the MIT license. The core thesis is that a semantic layer should be a governed query engine, not a SQL wrapper, and that agents need to express query intent concisely while deterministic machinery handles joins and safeguards. Vendor lock-in from proprietary semantic layers compounds the problem, making the case for a fully open alternative stronger.

Key Insight

Existing semantic layers solve metric centralization but force teams to rebuild the same agent tooling, query engine wrappers, and context APIs every time — an agentic-first OSS semantic layer should ship all three layers out of the box so developers can focus on interfaces and harnesses instead.

Spicy Quotes (click to share)

• 7

  A semantic layer should be a governed query engine, not a SQL wrapper.

• 3

  The agent states the what — which measures, at which grain. The engine deterministically owns the how, like any query planner on a database, but on top of our defined metrics, with the intent to produce the same SQL every time.

• 4

  An agent should be able to express its query intent in a way that's as concise as possible, and just formal enough to be unambiguous, but let the deterministic machinery figure out the necessary joins, subqueries, and safeguards.

• 4

  With agents, there's one more layer that you need to build: the implementation for working with agents and for collaboration between humans and agents. So there you implement another wrapper layer.

• 6

  Pre-agentic Semantic Layers were built for dashboards that replay the same queries each time — the definition alone was never enough.

• 5

  Letting the agent ask what it needs, instead of what the semantic layer can express in its limited way.

• 4

  You want a semantic layer so agents don't end up writing raw SQL.

Tone

technical, opinionated, promotional