Global – October 27, 2026 -- Snowflake has put metadata connectors for 18 data, BI and pipeline sources into public preview as part of Horizon Context, a governed context layer inside Horizon Catalog designed to give AI agents and analysts shared business definitions.
Snowflake introduced the layer at Snowflake Summit. It collects metadata, enriches it with business meaning and exposes it to people, BI tools and AI agents. Snowflake's stated aim is to stop teams from rebuilding definitions, such as what counts as customer retention, for every new AI use case.
Connectors cover 18 sources including Power BI, Tableau and Databricks
The supported list spans databases, BI tools and data pipeline tools: Power BI, Tableau, Looker, Sigma, Hex, dbt, SQL Server, Oracle, PostgreSQL, Databricks and BigQuery. Depending on the connector and its configuration, Horizon Context collects schemas, query logs, dashboard definitions and usage signals.
Teams can browse external assets next to Snowflake objects, search their metadata and trace cross-platform lineage from the catalog. The connectors collect metadata only and do not require replicating the underlying data sets into Snowflake.
Snowflake CoCo uses lineage to investigate data issues and change impact
Snowflake CoCo, the company's AI-assisted workflow tool, can use available lineage to answer questions such as where a retention data set originates and which downstream reports depend on a given column. Lineage lets teams check whether an unexpected metric decline reflects a business change or a data issue. They can also see which reports an upstream table or column change could affect.
New APPLY COMMENT privilege lets stewards edit descriptions without broader rights
Administrators can now grant the APPLY COMMENT privilege so stewards can edit object and column descriptions without MODIFY or OWNERSHIP rights. Separately, metadata permissions let users explore catalog information and lineage without permission to query the data itself. Business experts can therefore contribute definitions while administrators keep control of data access.
Description propagation carries column definitions downstream when values pass unchanged
Column description propagation uses column lineage to copy descriptions to supported downstream columns, including renamed ones, when values pass through unchanged. Propagation stops when values are transformed, combined or aggregated. It does not overwrite descriptions written directly on downstream columns, and teams can tell propagated descriptions from user-written ones.
Semantic views complement the layer with governed definitions of metrics, dimensions and relationships. Together with the redesigned Horizon Catalog Explorer, the connectors, the new privilege and propagation are all in public preview.