How Atlan builds lineage for Snowflake AI models
This FAQ covers how Atlan constructs lineage for Snowflake AI models, what upstream dependencies are resolved, and the scope of lineage extraction.
How does Atlan build lineage for Snowflake AI models?
Atlan builds upstream lineage for AI model versions automatically during the crawler run, using Snowflake's SNOWFLAKE.CORE.GET_LINEAGE() function. No separate lineage workflow is needed.
How deep does Atlan trace upstream lineage for AI models?
| Property | Value |
|---|---|
| Upstream depth | 5 levels |
| Lineage granularity | Module level |
What upstream asset types does Atlan resolve for Snowflake AI models?
| Upstream asset type | Description |
|---|---|
| Tables and views | Standard Snowflake tables and views used by the model |
| Dataset versions | Snowflake ML dataset versions that are upstream inputs |
| Feature views | Feature views from Snowflake Feature Store that feed into the model |
| Other models | Other registered models that are dependencies of this model |
Why doesn't my model show lineage after crawling?
The most common cause is a missing USAGE privilege on the model. SNOWFLAKE.CORE.GET_LINEAGE() requires USAGE on the model object to resolve its upstream dependencies—without it, lineage extraction is silently skipped for that model version. See Permissions for Snowflake AI models for the full grant required.
See also
- Crawl Snowflake AI models: Configure the crawler to discover and catalog AI models and extract lineage.
- Permissions for Snowflake AI models: Full breakdown of each privilege required and what it enables.