Snowflake Horizon Catalog pairs discovery with semantic context for agents

Horizon Catalog is presented as a catalog for data in and out of Snowflake, interoperable with any engine, data, and cloud. It claims users can find data, understand it through semantic context, and rely on governance controls including sensitive-data protection, data quality, end-to-end lineage, AI guardrails, and AI governance. Stated problems are disconnected silos that slow discovery, AI answers that lack business context, and fragmented access and lineage that leave gaps as data moves.

Interoperability coverage includes Apache Iceberg read/write, the Iceberg REST Catalog API for Spark, Flink, and Trino with automatic governance, catalog-linked databases such as AWS Glue and Azure OneLake, and an Internal Marketplace for sharing governed data products without copying. The AI context layer centers on semantic views—governed, business-aligned definitions agents use—creatable via Autopilot or ingested from Tableau and Power BI, plus column-level lineage across Snowflake, external databases, BI tools, and OpenLineage feeds.

For teams building agentic access to enterprise data, the extract ties catalog entries to shared semantics and provenance rather than raw object lists. Semantic views and lineage that can trace an AI answer to origin are the bridge between platform catalogs and governed meaning across tools. That is the same seam semantic layers and interoperability programs try to keep consistent.