AI is only as consistent as your definitions
Define each metric once in your warehouse, or bring the one you already have, and dashboards, apps, and AI answers all use it.
One definition per metric
Revenue means the same thing in workbooks, apps, and Sigma Assistant’s answers.
Bring your semantic layer
Use Snowflake semantic views, Databricks metric views, and dbt Semantic Layer metrics instead of rebuilding them.
AI that uses your definitions
Sigma Assistant, Sigma Agents, and Claude through the Sigma MCP server answer from the model, so AI and dashboards agree.
Trusted by 2,000+ leading enterprises around the world
One definition
Three teams, three versions of revenue
When the logic lives in each report, revenue gets defined once per dashboard. AI answers inherit the drift, and a fix means editing each copy.
| Today | With Sigma |
|---|---|
| The same calculation rewritten in each dashboard | Each metric defined once in the model or in dbt |
| Copies of reports that quietly disagree | Endorsed models people know to build on |
| Chat AI on exports gives a different answer | AI that answers from the dashboard’s metrics |
| A definition change means editing dozens of reports | Change it once and the reports pick it up |
Built for the people who own the number
The same governed metric shows up where each team works: in the model, in the board pack, and in the answer an agent gives.
Data and analytics leaders
Define each metric once, endorse it, and stop being the help desk for what a number means.
Finance and business leaders
One version of revenue, margin, and pipeline, with sensitive fields limited to the people who should see them.
Engineering and AI teams
Give your agents and AI tools a governed layer to query, and promote changes with version tags.
Why teams define it in Sigma
Keep what you’ve built, run it in your warehouse, and give AI the same definitions as your dashboards.
Keep semantics
Keep the semantics you already built
Add a Snowflake semantic view or Databricks metric view like any table, and its metrics are ready to use, defined where they live. dbt Semantic Layer metrics query through custom SQL.
- Snowflake semantic views
- Databricks Unity Catalog metric views
- dbt Semantic Layer for dbt Cloud
Runs in your warehouse
Model in a spreadsheet and run it in the warehouse
Build tables, joins, and metrics with familiar functions. Queries push down to your warehouse, and people add their own calculations on top without changing the model.
- Joins and relationships without SQL
- Reusable metrics
- Logic runs in your warehouse
Ground AI
Give AI the same definitions
Sigma Assistant and Sigma agents answer from your models, and Claude, ChatGPT, or Cursor query the same metrics through the Sigma MCP server, as each person.
- Metrics AI tools can call by name
- Answers that match the dashboard
- Each person’s row-level security applies
Trusted changes
Show what to trust and change it safely
Badge models as endorsed, warning, or deprecated. Version tags promote changes from dev to prod, and version history restores an earlier model.
- Endorsed, warning, and deprecated badges
- Dev to prod with version tags
- Restore an earlier version
Governance starts in the model
Permissions and policies defined in your warehouse carry through models, dashboards, and agents.
Endorsed models
Badge the models people should build on, and flag the ones they shouldn’t.
Per-user OAuth
Connect with OAuth so queries run with each person’s own warehouse role.
Row- and column-level security
Reads and writes both inherit your table policies.
Audit trail
Audit logs record user activity, AI conversations, and input table edits, so you can see who changed what.
Your data stays put
Sigma queries your warehouse in place and never copies the data out.
Certified
SOC 1, SOC 2, SOC 3, HIPAA, GDPR, CCPA, and ISO 27001.
How data teams settled the number
See how data teams moved their logic into governed models, so dashboards, reports, and answers start from the same numbers.
1,500 reports consolidated. 33,000 objects simplified.From 1,500 Reports to One Workbook: Inside Saddle Creek’s BI Reset with Sigma
“We’re working toward ensuring there’s a data model for every pod. That way, teams can self-serve answers through a consistent structure.”How Bilt Made Sigma Central to its Data Strategy
“Since Sigma, we’re able to have a lot more trust in our metrics. Now we know when people go into Sigma, the data is up to date. People know what they're doing, they understand the data.”How Sigma and HashiCorp drove record BI adoptionStraight from the source
“They have invested in data models, semantic layer connections, and security to ensure there is a degree of trust in every dashboard.”
Senior Director of Data and AnalyticsBanking · Gartner Peer Insights (opens in a new tab)“I love the ability to interact with and create metrics with our data without making core changes to the governed and centralized data warehouse.”
Product AnalystSoftware · Gartner Peer Insights (opens in a new tab)“Sigma has empowered our organization to leverage the full potential of our semantic layer, democratizing data analysis across departments.”
Data Analytics and Engineering ManagerHealthcare and Biotech · Gartner Peer Insights (opens in a new tab)“We can either write custom SQL scripts, or leverage models that were previously created by our team using the UI.”
BI AnalystSoftware · Gartner Peer Insights (opens in a new tab)“The ability… to connect multitudes of models using data lineage provides the clarity an organization needs to rely on the reporting and numbers generated.”
EngineerBanking · Gartner Peer Insights (opens in a new tab)“The seamless integration with Snowflake without having to schedule extracts to keep data fresh ensures we are all looking at the most up to date data.”
Data and Analytics ManagerManufacturing · Gartner Peer Insights (opens in a new tab)“The version tagging feature enables us to safely and dependably deploy workbooks to production.”
Principal Software ArchitectSoftware · Gartner Peer Insights (opens in a new tab)“Sigma queries directly against the warehouse, so users are always working with current data, not stale CSVs.”
Director of FinanceFinance · Gartner Peer Insights (opens in a new tab)
Data models FAQ
What is a data model in Sigma?
A data model is a shared set of tables, relationships, and metric definitions that workbooks, apps, and agents read from, so everyone uses the same logic.
Do we have to rebuild our semantic layer?
No. Sigma uses Snowflake semantic views, Databricks metric views, and dbt Semantic Layer metrics, so the definitions stay where you maintain them.
Do we model in Sigma or in our warehouse?
Either. Build data models in Sigma on your warehouse tables, or keep definitions in dbt or your warehouse’s semantic layer and query them from Sigma. The logic runs in the warehouse either way.
Can Claude or ChatGPT use our metric definitions?
Yes. Through the Sigma MCP server, AI tools find your data models and query their metrics by name, with each person’s Sigma permissions.
Where does the logic run?
In your warehouse. Sigma pushes queries down to it instead of computing in a separate engine.
Put one definition behind every number
Bring the metric your teams argue about, and we’ll show you how to define it once.