Snowflake Semantic View Autopilot: AI-Powered Semantic Modeling in Minutes

Governed, trusted semantics are table stakes for AI-ready data. Today, we're announcing the general availability of Semantic View Autopilot (SVA), a system that autogenerates semantic views from your existing queries and BI assets.

The problem is a lack of definition, not the LLM

In 2025, teams building AI agents learned that even the smartest models struggled with inconsistent business logic. The barrier wasn't AI capability — it was data definition.

Prashanth Sanagavarapu, SVP of Engineering at VTS, notes that "building and maintaining a consistent semantic layer required significant manual effort" to prevent conflicting numbers. That's why we built Semantic View Autopilot: to automate creation of that governed, trusted layer.

"Semantic View Autopilot provides our AI systems with a consistent, governed understanding of business metrics ... allowing us to deliver reliable personalization and AI-driven engagement that our customers can trust,” says Matt Walker, CTO at Simon AI.

Snowflake automates semantic view creation

Semantic views provide context on your data's meaning and intent, not just its structure. They instruct LLMs how to translate data into business concepts — but creating them is often time-consuming and heavily manual.

For data teams, the goal is consistent logic. But manual creation is a burden. The product team might define "Monthly Recurring Revenue" one way, unaware that the finance team excludes one-time setup fees. These hidden rules surface only after deployment, when numbers don't reconcile.

SVA closes this gap by automating the creation and governance of semantic views. Instead of requiring engineers to code definitions from scratch, SVA proposes candidate metrics and filters learned from query history and trusted BI assets, so teams can review, certify and deploy in minutes instead of weeks.

How it works: Learning from consensus patterns

SVA's core principle is that your semantics are already defined in your query history, data usage and dashboards. This transforms semantic modeling from coding into curation — teams can now just focus on reviewing logic that SVA surfaces. These governed definitions power Snowflake Cortex Analyst, Cortex Agents and Snowflake Intelligence for more accurate, trusted results.

SVA analyzes three key signals, as described below.

Pattern recognition and consensus-based extraction

SVA uses clustering algorithms that analyze query patterns and natural language questions to identify consensus business logic. When conflicting definitions exist — such as different "active user" filters — SVA surfaces the most common pattern as a proposal.

For example, if 200+ queries consistently calculate "active user" as user_engagement_score > 50 AND last_login_days < 30, SVA proposes this filter even if a user recently ran a different query.

Multisignal learning from high-confidence sources

The highest-confidence source is often existing BI dashboards, where years of business logic already live. Tableau is the first BI tool SVA supports, with more coming through our 20+ OSI partners. SVA turns static dashboards into conversational AI in minutes (see the hands-on lab).

Teams can also upload trusted SQL queries directly. SVA extracts relationships and metrics and stores them as verified queries for future use. And because it all runs inside Snowflake, SVA can analyze your actual data. Column cardinality reveals relationship types and informs suggestions such as adding a Cortex Search service for better accuracy.

Continuous iteration based on evolving usage

SVA monitors usage patterns to keep semantic views current. If an organization launches a "Pro" subscription tier, SVA notices that new queries include subscription_tier = 'pro' and proposes incorporating it — so answers stay consistent as business rules evolve.

The shift from BI to AI agents requires a semantic foundation grounded in data usage, not LLM assumptions. Semantic View Autopilot is the fastest path to governed, context-aware AI — available now in all Snowflake regions where Cortex Analyst is available.

Get started today. Try SVA in your account and learn best practices for creating semantic views for Cortex Analyst.