Purpose

Define reusable, governed business logic directly in Snowflake for use by Snowflake Cortex

Modular construction and management of SQL from meta-data repositories implemented in Snowflake tables, organized as View Definitions for automated generation of semantic, business, and technical VIEWs

Business Logic Encoding

Encodes dimensions, measures, filters, relationships

Encodes full SQL expressions, aliases, CTEs, UDF calls, joins, FROM clauses and transformation logic in meta-data repositories

Governance & Versioning

Managed via Snowflake object lifecycle and meta-data

User defined governance and lifecycle management using standard Snowflake roles and security

Modularity

Moderate – definitions are centralized but tightly bound to schema

High – SQL is constructed using reusable, components (LEGO-style modularity) with managed qualification; new View Definitions created and transformed from existing View Definition

Tooling Interoperability

REST API integration required

Outputs standard CREATE VIEW and SELECTs compatible with all BI tools, dbt, or direct Snowflake consumption via AltaSQL function call

Customization

Built-in structure limits flexibility

Fully customizable SELECT clauses, CTEs, filters, WHERE and Post Where logic, UDF calls, expressions, post-processing logic and SQL code

Usage Model

Call Cortex Analyst via API from application to “prompt”

SQL developers or analysts generate governed SELECT/CREATE VIEW/CTAS via ASDF functions; end-users consume may generate or use ready-to-run SQL statements

Meta-data Repository

Stored in Snowflake catalog

Maintained in AltaSQL-managed repositories; includes lineage, transformation types, AltaSQL Tags, and semantic annotations

Schema Independence

Decoupled from physical schema to a degree

Fully decoupled – “Plug-n-Play” FROM_CLAUSE argument provides dynamic data source specification

Auto-Documentation

Limited – depends on external tools or manual annotations

Built-in meta-data catalog (AltaSQL Tags), Snowflake tagging, “sticky” column comments, lineage, column alias syntax

Multi-layered Architecture

Possible, but not automatic

Enables layered VIEW architecture: raw → transformed → business → semantic → consumption

Ease of SQL Generation

SQL statements LLM generated; not directly modifiable

Spreadsheets provide pre-configured, Ready-to-execute, AltaSQL function calls for generating SQL statements either manually or from procedural code

Tag-Based Governance

Not available

Uses AltaSQL Tags with TAG_CLASS, TAG_NAME, TAG_VALUE to modularize and govern semantic intent

Great for centralizing metrics and natural language queries

Ideal for modular SQL generation, governance, and full semantic layering

Good for business-facing consumers

Optimized for analysts, engineers, architects, and meta-data managers

Limited extensibility outside Semantic View definition

AltaSQL Tags provide a comprehensive, user definable meta-data catalog to store any type of information