One number.
Every team, every agent.
A semantic layer defines the entities, dimensions, and measures behind your metrics. Atlan generates them from the BI logic and query history you already have, then gives every definition an owner, a version, and a test suite.
A semantic model: three entities with primary keys joined by named N to 1 relationships, dimensions to slice by, and two measures defined once on top of them.
Trusted by AI-forward enterprises
"Atlan captures Workday's shared language to be leveraged by AI via its MCP server. As part of Atlan's AI labs, we're co-building the semantic layer that AI needs."
Joe DosSantos
VP Enterprise Data & Analytics, Workday
"We're focused on how we can scale context development as much as possible, and where can we leverage Atlan AI to build the most robust definitions across our data estate and make sure we're propagating it upstream and downstream."
Takashi Ueki
Head of Enterprise Data & Analytics, Elastic
"With Atlan we cataloged over 18 million assets and 1,300+ glossary terms in our first year, so teams can trust and reuse context across the exchange."
Kiran Panja
Managing Director, Cloud & Data Engineering, CME Group
"AI initiatives require more context than ever. Atlan's metadata lakehouse is configurable, intuitive, and able to scale to hundreds of millions of assets."
Andrew Reiskind
Chief Data Officer, Mastercard
WTF IS THE CONTEXT LAYER
A bi-weekly live series for AI leaders and builders. One burning question per episode, an open AMA floor, and guests who've actually built context infrastructure.
A semantic layer teaches agents business language and meaning. But is meaning enough to act? David Mariani (co-founder and CTO of AtScale) joins Austin Kronz on where semantic and context layers diverge, and whether they are two names for the same solution.
One question.
Three kinds of context.
Knowledge is which number the question means. Expertise is how to aggregate it. Norms are whose definition is the authoritative one.
Question It Raises
Context Type
Answer It Needs
Which number is the metric?
avg_dt_secs, not the POS number. The week runs Monday to Sunday, store local time
How is it aggregated?
Weighted by car count. A mean of means over-weights the quiet stores
Whose definition wins?
Operations owns avg_dt_secs at v4. Finance owns the POS metric. Changes need approval
No agent runs on
a semantic layer alone.
It defines what the numbers mean. Which table to compute them on, how one record relates to another, and what was decided last time come from somewhere else.
How are business metrics defined and measured?
Entities, relationships, dimensions, and measures over tabular data. A graph walks from one record to its neighbours. A semantic layer collapses millions of rows into a rate.
Explore Semantic Layermeasure: refund_rate label: "Refund rate" expr: COUNT(refund) / COUNT(payment) filters: - status != 'reversed' # 9 in 10 queries - NOT is_test_account entities: [payment, refund, invoice] dimensions: - region # circles 1-4 + acquired - fiscal_quarter # Apr 1 - Jun 30, local tz owner: finance-ops version: 4 # changes need approval
Your metrics are already defined, badly, in six places.
Atlan reads all six and reconciles them.
Definitions mined from what the business does, not from a blank spreadsheet.
Every metric you care about is already implemented somewhere: in a BI tool's semantic logic, in a dashboard's calculated field, in the SQL your analysts copy between queries. Atlan reads those implementations and proposes the governed version.
The definition of a metric is production code. Atlan treats it that way.
A generated definition is a proposal. In Context Engineering Studio it gets tested against historical cases, reviewed by the person who actually owns the number, and versioned like any other production asset.
One definition of refund rate, wherever the question gets asked.
A metric definition locked inside one BI tool serves the people who open that tool. Governed definitions have to reach every analytics agent, copilot, and notebook that will ever ask the question.
Cortex AnalystSee the context layer, built live.
Watch Atlan engineers bootstrap, test, and deploy enterprise context on real data — from cold start to production-ready agents.
The teams winning with AI are not the ones with the best models. They are the ones building context that compounds. Watch Atlan mine knowledge from your existing stack, build and certify a context layer through a development lifecycle, and deploy it to any agent platform.
Learn more about the Semantic Layer with Atlan.
Semantic Layer for AI Agents
Why a governed semantic layer is what lets an agent compute a metric the way your business defines it, not the way the schema suggests.
How to Build an AI-Ready Semantic Layer
The practical build: which entities and measures to model first, how to certify them, and how to keep definitions from drifting once agents depend on them.
Semantic Layers Failed. Context Graphs Are Next
Semantic layers solved meaning over rows and stalled on everything else. What agents need on top, and why the graph is the next layer.
Semantic Layer for AI Agents
Why a governed semantic layer is what lets an agent compute a metric the way your business defines it, not the way the schema suggests.
How to Build an AI-Ready Semantic Layer
The practical build: which entities and measures to model first, how to certify them, and how to keep definitions from drifting once agents depend on them.
Semantic Layers Failed. Context Graphs Are Next
Semantic layers solved meaning over rows and stalled on everything else. What agents need on top, and why the graph is the next layer.
One definition of the number.
Every agent, every time.
