What is Snowflake Horizon Context? How it compares to the enterprise context layer

Emily Winks

Data Governance Expert

Updated:06/03/2026

Published:06/03/2026

14 min read

Watch Context Layer Live Get the Context Layer Ebook

Key takeaways

Horizon Context is a semantic enrichment layer inside Horizon Catalog, not a standalone product or full data catalog.
Business Glossary, the definitional layer AI agents need, is on the H2 2026 roadmap, not yet shipped.
Atlan layers on top of Horizon Context: 100+ connectors, shipped Business Glossary, Context Agents across 50+ enterprises.
Snowflake = data plane. Atlan = context plane. The two systems are complementary, not competitive.

What is Snowflake Horizon Context?

Snowflake Horizon Context is the semantic and metadata enrichment component inside Snowflake Horizon Catalog. Announced June 2, 2026, it collects metadata signals from external sources, enriches them with business definitions, and delivers context to Snowflake AI agents, Cortex, CoWork, and CoCo. Wave 1 ships 5 metadata connectors in private preview and Semantic Studio in private preview. Business Glossary is on the H2 2026 roadmap.

It includes

Metadata Connectors (Wave 1): 5 connectors in private preview (PostgreSQL, SQL Server, Tableau, Power BI, dbt)
Semantic Studio: AI-assisted IDE for business logic authoring, private preview
Semantic View Autopilot: auto-generates semantic views from existing SQL and BI files, available
AI-Generated Documentation: auto-generates table and column descriptions, available
Business Glossary: H2 2026 roadmap, not yet shipped

Is your data estate AI-agent ready?

Assess Your Readiness

Snowflake Horizon Context is the semantic and metadata enrichment layer inside Snowflake Horizon Catalog, announced June 2, 2026 at Snowflake Summit. It collects signals from external databases and BI tools, enriches them with business definitions, and delivers context to Snowflake’s AI agents, Cortex, CoWork, and CoCo. Wave 1 ships 5 metadata connectors (private preview) and Semantic Studio (private preview). Business Glossary is on the H2 2026 roadmap.^[1]

How does Snowflake Horizon Context compare to the enterprise context layer

Horizon Context is a component of Snowflake Horizon Catalog, the umbrella product Snowflake describes as “the universal AI catalog for enterprise data.” Before June 2026, Horizon Catalog covered tags, masking policies, access controls, profiling, and basic lineage. Horizon Context adds the semantic and business-meaning dimension on top of that technical metadata foundation.

Kleinerman’s framing at the Summit keynote is the clearest description of what it does: “Horizon Context helps collect signals, enrich those signals and make them available to coco, cowork or cortex agents for you to get more context and more semantic information.” The three-part logic, collect, enrich, activate, is how Horizon Context works. It is a runtime enrichment architecture, not a persistent enterprise knowledge graph.

On semantic views specifically, Kleinerman noted: “A lot of what we’ve been doing in [semantic views] continues to move forward as part of Horizon Context.” This is significant: Horizon Context is not a ground-up new product but an evolution of the semantic view work Snowflake has been building for several years, now branded and wrapped into a coherent product layer.

Horizon Context fills a real gap. When business logic is scattered across SQL queries, BI dashboards, and AI tools, agents produce inconsistent answers to identical questions. Horizon Context creates a unified context layer foundation within Snowflake — going beyond a simple semantic layer — enforcing definitions at the meaning level so a metric defined for the finance team stays consistent whether accessed from Power BI, Tableau, or a Cortex agent query.

What Horizon Context is not, as of June 2026:

Not a standalone product, it lives inside Horizon Catalog
Not a cross-system platform, it operates within the Snowflake perimeter and 5 committed external connectors
Not yet a business glossary, that layer arrives H2 2026

These are scope boundaries, not criticisms. For organizations running primarily Snowflake-native AI workloads, Horizon Context is a genuinely useful semantic foundation.

Build Your AI Context Stack

Get the blueprint for implementing context graphs across your enterprise. This guide walks through the four-layer architecture (metadata foundation to agent orchestration) with practical implementation steps for 2026.

Get the Stack Guide

What does Snowflake Horizon Context include?

As of June 2026, Horizon Context ships four capabilities, two of which are in private preview.

Metadata Connectors (Wave 1)

Wave 1 commits five external connectors: PostgreSQL, Microsoft SQL Server, Tableau, Power BI, and dbt. All five are in private preview as of June 2026. These connectors pull schemas, query logs, and dashboard definitions from those external sources into Horizon’s semantic layer, making external metadata available to Snowflake’s governance engine.

Snowflake’s architecture diagram at the Summit showed a broader list including Iceberg, Delta, AWS Glue, Unity Catalog, OneLake, and Airflow, but Wave 1 is the committed launch set. Snowflake’s own roadmap suggests broader coverage is the direction; Wave 1 is the starting point.^[3]

For comparison: Atlan’s enterprise data graph ships 100+ connectors today, all generally available, including BigQuery, Redshift, Salesforce, Looker, MongoDB, and legacy warehouses that Horizon’s Wave 1 does not reach.

Semantic Studio

Semantic Studio is an AI-assisted IDE inside Snowflake Workspaces for defining business logic without writing raw SQL. It supports Git-based versioning and integrates with CoCo, allowing teams to author metric definitions, semantic relationships, and business rules collaboratively. As of June 2026, it is in private preview.

For comparison: Atlan’s Context Engineering Studio is generally available. It bootstraps, tests, and versions context engineering for AI agents as code with CI-integrated eval suites that validate context before it reaches agents, compressing what one insurance customer described as a one-year build into one month.

Semantic View Autopilot

Semantic View Autopilot is available today. It auto-generates semantic views from existing SQL queries, Tableau workbooks, or Power BI reports, reducing the manual effort of creating semantic views for teams already using those BI tools. Semantic views have become a core technical primitive for Snowflake’s semantic layer strategy; Autopilot accelerates their creation within the Snowflake perimeter.

AI-Generated Documentation

Horizon Context auto-generates table and column descriptions using metadata and optionally sample data. This is static output, generated once, Snowflake-perimeter only, without compounding learning over time.

For comparison: Atlan’s context agents generated 690K+ descriptions across 50+ enterprise customers by April 2026, 87% rated on par or better than human writing, operating autonomously across all connected systems and compounding knowledge over time.^[8]

What is not in Horizon Context yet

Business Glossary, the layer that makes business terminology consistent across teams, tools, and agents, is on the H2 2026 roadmap and not yet shipped. Until it ships, semantic alignment across business units for production AI requires this layer to come from somewhere else. Context agents with compounding learning are also not part of Horizon Context; cross-system business glossary alignment is not in scope.

Feature status table:

Feature	What it does	Availability	Notes
Metadata Connectors (Wave 1)	Ingests schemas, query logs, dashboard definitions from 5 external sources	Private preview	PostgreSQL, SQL Server, Tableau, Power BI, dbt. Architecture diagram shows more; Wave 1 is the committed set.
Semantic Studio	AI-assisted IDE for defining business logic with Git versioning and CoCo integration	Private preview	Inside Snowflake Workspaces
Semantic View Autopilot	Auto-generates semantic views from existing SQL, Tableau, or Power BI files	Available	Core primitive for Snowflake’s semantic layer strategy
AI-Generated Documentation	Auto-generates table and column descriptions from metadata + optional sample data	Available	Static output; Snowflake-only; no compounding learning
Business Glossary	Consistent business terminology across teams and agents	H2 2026 roadmap	Not yet shipped
External agent access via MCP	Delivers semantic context to external AI agents via MCP standard	Available (via Horizon Catalog)	Enables Atlan and other tools to read Horizon context

How does Snowflake Horizon Context fit in the Horizon Catalog?

Snowflake Horizon Catalog is the governance umbrella for Snowflake’s AI trust infrastructure, covering tags, masking policies, access controls, profiling, and lineage. Horizon Context is one component within it: the semantic and metadata enrichment layer. The hierarchy is: Horizon Catalog (umbrella) covered by Horizon Context, Horizon Policies, Horizon Governance, and Horizon Security.

The data flow works in five steps:

Metadata Connectors pull signals from external sources (5 committed in Wave 1)
Semantic Studio and Semantic View Autopilot define and enrich business logic
Horizon Context layer stores the enriched context
Horizon Catalog’s governance engine enforces definitions, a finance metric restricted for the finance team stays restricted in Power BI, Salesforce, and Cortex agents
Context flows to Snowflake CoWork, Snowflake CoCo, and Cortex agents

This governance enforcement at the semantic level, not just the table level, is a genuine capability worth acknowledging. A definition restricted for the finance team stays restricted consistently, whether the query comes from a BI tool or an AI agent. That is real, useful governance for Snowflake-native workloads.

Horizon Context is also the primary data source for Cortex Sense runtime enrichment. Without Horizon Context enrichment, Cortex Sense operates on raw schema metadata only. With Horizon Context, Cortex Sense reaches 86% accuracy on structured questions compared to 24% accuracy with frontier models alone, per Snowflake’s benchmarking referenced in SiliconANGLE’s Summit coverage.^[4] The gap that context enrichment closes is real, but it is a Snowflake-perimeter stat.

Horizon Context delivers context to Cortex agents via Horizon Catalog. External agents, including Atlan’s context layer for Snowflake, can also read from Horizon Context via the MCP server for Snowflake standard, which means the enriched metadata Snowflake generates is accessible to tools operating outside the Snowflake environment. This is the technical basis for Atlan’s complementary architecture: Horizon Context produces enriched metadata; Atlan ingests it and extends it across the full enterprise stack. For a deeper look at how Atlan specifically enriches context layer for Snowflake Cortex, see the dedicated guide.

For a broader view of how Snowflake’s governance layer fits alongside third-party tools, see Snowflake Data Governance Best Practices for 2026 and Snowflake Governance vs Third Party Tools.

Snowflake Horizon Context vs the enterprise context layer

Horizon Context is Snowflake’s semantic enrichment layer within the Snowflake perimeter, while Atlan’s enterprise context layer extends coverage to 100+ systems with a shipped Business Glossary, column-level lineage, and Context Agents that have generated 690K+ descriptions across 50+ enterprises. The core architectural distinction is straightforward: Snowflake is the data plane, Atlan is the context plane, and the two systems are complementary rather than competing. Atlan ingests Horizon Context as a data source, enriches it with cross-system semantics, and serves the result back to Snowflake’s agents via MCP server for Snowflake.

For a detailed capability comparison, see How Snowflake Horizon Context and the Atlan context layer work together. The context layer for Snowflake guide provides the full native-plus-enterprise view, with the agent context layer architecture connecting these systems.

If you want to evaluate where your stack sits today, the Context Maturity Assessment surfaces the specific gaps.

Is Your Data Estate AI-Agent Ready?

Assess your context stack readiness in minutes. The AI Agent Context Readiness tool surfaces specific gaps in connector coverage, Business Glossary status, lineage depth, and agent accuracy across your enterprise.

Assess Your Readiness

Real stories from real customers: how enterprises use the context layer for AI

"We're excited to build the future of AI governance with Atlan. All of the work that we did to get to a shared language at Workday can be leveraged by AI via Atlan's MCP server...as part of Atlan's AI Labs, we're co-building the semantic layer that AI needs with new constructs, like context products."

-- Joe DosSantos, VP of Enterprise Data and Analytics, Workday

Watch Now

"Atlan is much more than a catalog of catalogs. It's more of a context operating system...Atlan enabled us to easily activate metadata for everything from discovery in the marketplace to AI governance to data quality to an MCP server delivering context to AI models."

-- Sridher Arumugham, Chief Data and Analytics Officer, DigiKey

Watch Now

Why the context plane matters for your Snowflake AI initiative

Snowflake Horizon Context is a meaningful step toward the enterprise context layer, and an honest evaluation requires acknowledging both what it delivers and where the boundary lies. For Snowflake-native AI workloads, it provides real semantic enrichment, consistent governance enforcement, and a foundation that Cortex agents can operate on with measurably higher accuracy. That is not nothing.

But the production AI deployments that reach scale, the ones in that 17% that make it past pilot, share a common characteristic: context that spans the full enterprise stack. That means a Business Glossary with stewardship workflows, cross-system lineage at the column level, context agents that compound knowledge over time, and a context store that serves any agent regardless of where the data lives. Horizon Context builds toward this; it does not yet deliver it.

The architectural relationship between Snowflake and Atlan is straightforward: Snowflake builds and enriches the data plane. Atlan builds and maintains the context plane. They work together in production at Workday, Nasdaq, Mastercard, General Motors, and 50+ other enterprises, and the combination is why those deployments reach production rather than stalling in pilot.

If you are evaluating your enterprise’s readiness to deploy AI agents at scale, the Context Maturity Assessment surfaces the specific gaps in your context stack. Or if you would like to see how the Enterprise Data Graph connects with Snowflake Horizon Context in your environment:

Book a Demo

FAQs about Snowflake Horizon Context

1. What is the difference between Snowflake Horizon Catalog and Snowflake Horizon Context?

Horizon Catalog is the umbrella product covering Snowflake’s full AI trust infrastructure, tags, masking, access policies, profiling, lineage, and security. Horizon Context is one component within it: the semantic and metadata enrichment layer that collects business definitions, enriches metadata, and delivers context to Snowflake’s AI agents (Cortex, CoWork, CoCo). Think of Horizon Catalog as the governance framework; Horizon Context as the semantic meaning layer inside it.

2. Is Snowflake Horizon Context available now?

Partially. As of June 2026, Metadata Connectors (Wave 1: 5 connectors) and Semantic Studio are in private preview. Semantic View Autopilot and AI-generated documentation are available today. Business Glossary, the layer that makes business terminology consistent across agents and tools, is on the H2 2026 roadmap and not yet shipped.

3. What is Semantic Studio in Snowflake Horizon Context?

Semantic Studio is Snowflake’s AI-assisted IDE for defining business logic without writing SQL. Available inside Snowflake Workspaces with Git-based versioning and CoCo integration, it lets business teams author metric definitions and semantic relationships. As of June 2026, it is in private preview.

4. How does Snowflake Horizon Context feed AI agents?

Horizon Context enriches Snowflake metadata with semantic definitions, then delivers it to Snowflake’s native AI agents, Cortex, CoWork, and CoCo, via Horizon Catalog’s governance enforcement layer. The architecture: collect signals from connected sources, enrich with business definitions, make enriched context available to agents. External agents, including Atlan’s context layer, can also read Horizon context via MCP.

5. Does Snowflake Horizon Context replace a data catalog?

No. Horizon Context is a semantic enrichment layer within Horizon Catalog, not a standalone data catalog. It adds business meaning on top of Snowflake’s existing technical metadata. Enterprise data catalog capabilities, cross-system discovery, business glossary, data quality, stewardship workflows, span multiple systems and use cases that Horizon Context does not address, particularly for organizations with multi-cloud or multi-warehouse environments.

6. What are Metadata Connectors in Snowflake Horizon Context?

Metadata Connectors are Horizon Context’s mechanism for ingesting context from external data sources. Wave 1 (private preview, June 2026) includes 5 connectors: PostgreSQL, Microsoft SQL Server, Tableau, Power BI, and dbt. These connectors pull schemas, query logs, and dashboard definitions into Horizon’s semantic layer. Snowflake’s architecture roadmap targets broader coverage; Wave 1 is the committed launch set.

7. How does Snowflake Horizon Context compare to Atlan?

Horizon Context and Atlan serve different architectural roles. Horizon Context is Snowflake’s semantic enrichment layer, it collects metadata, enriches it with business definitions, and delivers context to Snowflake agents. Atlan is the cross-system context plane: 100+ connectors (vs. Horizon’s 5 in Wave 1), a shipped Business Glossary (vs. H2 2026 roadmap), and Context Agents that have generated 690K+ descriptions across 50+ enterprises. Atlan ingests from Horizon Context and extends it across the full enterprise stack.

8. What is Snowflake’s Enterprise Context Layer strategy?

Snowflake Horizon Context is an early-stage semantic enrichment layer within the Snowflake perimeter, the beginning of an enterprise context layer strategy. AI agent governance frameworks ensure agents act responsibly. AI agent governance frameworks ensure agents act responsibly on this enriched context. SiliconANGLE’s June 2026 analysis identified Snowflake’s deepest gap as business process logic and cross-system semantics. The full enterprise context layer, cross-system coverage, autonomous context agents, compounding knowledge, and open context storage — available from a full enterprise context layer — is what production AI deployments require beyond what Horizon Context currently ships.

Sources

Share this article

Atlan is the context layer for AI, the governed infrastructure that delivers enterprise knowledge to every model, every agent, and every team from a single source of truth.

Book a Demo Context Studio Live