Snowflake Horizon Context: The Governed Context Layer for AI, BI and Apps

Snowflake Horizon Context: The Governed Context Layer for AI, BI and Apps

Skip to content

Blog / Product and Technology / Snowflake Horizon Context: The Governed Context Layer for AI, BI and Apps

JUN 02, 2026 / 8 min read Product and Technology Snowflake Horizon Context: The Governed Context Layer for AI, BI and Apps

Nick El-Rayess +1

Your head of sales sees $14.2 million in Q3 revenue. Your CFO sees $12.8 million. Both asked an AI agent this morning to provide numbers. Same data. Why the discrepancy?

This is what can happen when business logic is scattered across separate tools: a metric defined inside a BI model only one team owns, a calculation buried in a dashboard, a set of instructions manually hardcoded into an LLM prompt. The result isn’t just metric drift, where sales and finance show different numbers in response to the same question. It’s a trust gap that makes it hard to move AI projects forward quickly with confidence.

Today, that changes. Snowflake introduces Horizon Context , a new capability within Horizon Catalog that offers a connected, governed semantic foundation with active context for AI and BI.

“As AI becomes increasingly embedded across our enterprise, it’s essential that applications, analytics and agents operate from the same trusted understanding of the business. Snowflake Horizon Context helps extend consistent business definitions across our broader data ecosystem, supporting more trusted and governed AI and analytics experiences at scale.” Jeff Miller

Managing Director, Global Head of Data Factory & Enterprise Data Platform, BlackRock Aladdin

Snowflake Horizon Context builds on Horizon Catalog’s metadata foundation by turning that metadata into governed business meaning. It collects context from across your data estate, enriching it with business definitions and relationships, and activating it so AI agents, BI tools and applications can automatically discover and apply trusted logic.

From a system of record to a system of intelligence

AI agents can write SQL, debug code and analyze data. But if agents have to guess when you ask what "revenue" means in your business, that is a context problem.

The disconnect between enterprise AI and the context it needs comes down to three problems:

Context is scattered: When important context is scattered across disparate databases, BI and data pipeline systems, no single system has the complete picture it needs to deliver trusted responses.

Context is raw: Making raw resources useful for AI requires enriching them with higher levels of meaning: How are these data assets related? Which ones are authoritative? What does this column mean? What is the correct way to calculate this metric?

Context is inactive: Context only works if it gets used. If that depends on users knowing where to find a specific agent or how to prompt it in a specific way, the much larger remainder of AI sessions do not benefit.

To address each of these problems, Snowflake Horizon Context collects metadata from disparate systems, enriches it and activates it so that it’s useful for your business. It turns Snowflake from a system of record into a system of understanding.

Collect: Build the complete picture

Your AI needs context from your entire data estate, inside and outside Snowflake. Horizon Context extracts context from external systems and collects it in Horizon Catalog.

Metadata Connectors (private preview): We are expanding Horizon Catalog from a data catalog for Snowflake toward a data catalog for all your data. Connect to external database, BI and data pipeline systems including PostgreSQL, Microsoft SQL Server, Tableau, Power BI and dbt and collect database schemas, query logs, dashboard definitions and more.

OpenLineage API (public preview): Configure OpenLineage producers like Apache Airflow to send lineage information directly to Horizon Catalog.

Open Semantic Interchange (OSI) : Snowflake is leading an open standard for how disparate systems exchange semantic metadata. The working group now includes 54 participating vendors and has published a specification .

Enrich: Turn raw metadata into business meaning

Raw context is a start, but it needs enrichment to create higher levels of meaning. Snowflake Horizon Context automates much of this, dramatically reducing the manual effort to build and maintain a context layer while still supporting human collaboration.

End-to-end column-level lineage: Horizon Context mines lineage from Snowflake and external query logs, BI systems and OpenLineage feeds, then stitches it together into a complete lineage graph.

Popularity: When dozens of similar-looking data assets exist, determining which to trust is key. Horizon Context uses query and access logs to calculate popularity as a signal for authoritativeness.

AI-generated data documentation: Horizon Context uses AI to generate table and column descriptions using both metadata and, optionally, sample data.

Semantic Views: Snowflake is launching several enhancements at Summit 2026. Advanced Semantics (private preview) bring level-of-detail (LOD) calculations, composable definitions and user-defined materializations with automatic query rewrite. Semantic Studio (private preview) is a full-fledged, AI-assisted IDE in Workspaces with CoCo integration and Git integration. Semantic View Autopilot takes your existing SQL, Tableau and Power BI files and creates semantic views for you.

Activate: Make context work automatically

The last mile of context is making sure it gets used. With Horizon Context, your context is discoverable, accessible and automatically activated as you interact with agents.

Context search: CoCo automatically retrieves relevant context using Universal Search, a hybrid keyword plus semantic search that uses signals like popularity for ranking and access control policies for filtering. New enhancements include search across your entire data estate (private preview), upgraded AI models for more relevant results, and search for workspace files, semantic views and agents.

Automatic semantic view discovery: When asked a data question, CoCo now automatically searches for and queries relevant semantic views, falling back to tables if none exist.

Semantic View interoperability: Expose semantic views via MCP, governed by Horizon Catalog, and connect from Claude, Cursor, Antigravity CLI or your agent of choice. We are expanding our ecosystem of supported BI platforms beyond Omni, Sigma, Hex and…