Snowflake Horizon Catalog¶

Horizon Catalog is the agentic catalog for all your data, whether it is inside or outside of Snowflake. It is open, interoperable, and compatible with any engine, any data, and any cloud. You can find the right data fast, understand it through rich semantic context, and trust every answer with enterprise-grade governance, from sensitive data protection to AI guardrails.

Problems that Horizon Catalog solves¶

Organizations with data and AI agents spread across teams and systems face common challenges around discoverability, context, and trust:

Disconnected data slows discovery. When data is spread across disconnected silos and incompatible formats, neither AI agents nor human users can find the authoritative information they need.
AI requires governed, contextualized data. Without business context, AI agents cannot distinguish authoritative assets from noise. This results in technically accurate but operationally irrelevant answers.
Fragmented security leaves gaps. When access controls and lineage are siloed across systems, sensitive data becomes difficult to track and protect as it moves across formats and platforms.

Connect your entire data estate¶

Horizon Catalog provides full interoperability across data inside and outside of Snowflake, with any engine, any data, and any cloud.

External metadata (Preview): Connect your source data and BI tools using out-of-the-box connectors.
Apache Iceberg tables: Read and write Iceberg tables from any engine. Supports both Snowflake-managed and external Iceberg tables.

Discover & understand data¶

Find the right data across your entire estate and understand it through rich context, whether you’re a person exploring the catalog or an AI agent retrieving a trusted answer.

Search Snowflake objects and resources: Search across databases, schemas, tables, and other objects to quickly locate the data you need.
End-to-end data lineage: Trace column-level lineage across Snowflake, external databases, BI tools, and OpenLineage feeds to understand where data comes from and how it’s used.
Internal Marketplace: Discover and share governed data products across teams without copying data.
Metadata access control (Preview): Grant metadata-only access so users can explore the catalog without access to the underlying data.

Build your AI context layer¶

Collect, enrich, and activate context across BI tools and data so that both humans and AI agents operate on the same trusted semantics. Together, these capabilities make up Horizon Context, the governed context layer in Horizon Catalog.

Business Ontology (Preview): Define canonical business terms, metrics, and relationships, then bind them to Semantic Views and other assets so people and AI use the same approved meaning.
Auto-generated descriptions: Snowflake Cortex generates table and column documentation from metadata and sample data, consistent across every engine.
Semantic views: Create governed, business-aligned definitions that AI agents use to understand your data. You can create them from scratch by using Autopilot, or ingest them from Tableau and Power BI.
Object tagging: Consistent metadata ensures that table definitions, query logs, joins, and lineage are presented the same way to every engine.

Deploy governed, trustworthy AI¶

Governance policies execute at the query engine layer, not the application layer. They apply automatically to every caller: human analyst, BI tool, or AI agent. There is no separate governance configuration for AI workloads.

Governance with Snowflake CoCo: Configure governance policies in natural language, assess your governance and observability maturity with scoring skills, and get prioritized recommendations. No SQL expertise is required.
Access control: Role-based access controls, Time Travel, and access history let you review past activity and data states. Trust Center continuously monitors for misconfigured roles and unprotected columns.
Data protection policies: Masking and row-access policies are enforced across any Iceberg REST Catalog-compatible engine.
Sensitive data classification: Automatically discover and classify columns. Snowflake applies tags and policies without manual intervention as schemas evolve.
Data quality monitoring: Continuous monitoring with root cause analysis ensures that every query, whether from a human or an AI agent, returns fresh and accurate data.
AI Guardrails: Detect, redact, and block personally identifiable information (PII) and protected health information (PHI) from agent outputs. Agents operate under the same role-based access control (RBAC) policies as human users.

Business continuity¶

You can manage primary and secondary environments, keeping them consistent through database and account replication across regions and accounts.