Skip to content
Casabase CubeCasabase Cube Documentation

Intelligence Overview

Casabase Cube integrates with Snowflake Intelligence and Cortex Agents to make governed multidimensional data available through natural-language analysis.

Instead of requiring business users to construct SQL or manually translate a question into cube coordinates, the Casabase Cube agent can interpret the request, discover the cube’s multidimensional structure, construct the required Point of View (POV), and execute the request through the Casabase Cube query engine.

Conceptually:

Business Question
       │
       ▼
Snowflake Intelligence
       │
       ▼
Casabase Cube Agent
       │
       ├── Discover Cube Metadata
       ├── Resolve Dimensions and Members
       ├── Construct the POV
       └── Select the Appropriate Cube Tool
       │
       ▼
Casabase Cube Query Engine
       │
       ├── Apply Security
       ├── Resolve Hierarchies
       ├── Aggregate Data
       ├── Evaluate Formulas
       └── Apply Time Behavior
       │
       ▼
Governed Result
       │
       ▼
Natural-Language Response

The important distinction is that the agent does not replace the Casabase Cube analytical model. It provides a natural-language interface to that model.

Governed Multidimensional Analysis

Section titled “Governed Multidimensional Analysis”

Casabase Cube already defines the business structures required for multidimensional analysis, including:

  • Cubes
  • Dimensions
  • Hierarchies
  • Members
  • Member properties
  • Formulas
  • Solve order
  • Time behavior
  • Row-level security

The agent uses those existing structures rather than requiring business users to recreate multidimensional logic in a separate AI model.

For example, a user might ask:

Show actual revenue by region for the current year.

The agent can determine which cube, dimensions, members, and hierarchy selections are required, construct the appropriate POV, and execute the query through Casabase Cube.

The resulting values are produced by the same Casabase Cube query engine used by supported SQL and interactive query paths.

Why Use the Casabase Cube Agent?

Section titled “Why Use the Casabase Cube Agent?”

Traditional natural-language analytics must often infer business meaning directly from relational tables.

Casabase Cube provides an additional governed analytical layer:

Snowflake Data
      │
      ▼
Casabase Cube
      │
      ├── Business Hierarchies
      ├── Multidimensional Calculations
      ├── Time Intelligence
      ├── Aggregation Behavior
      └── Security
      │
      ▼
Casabase Cube Agent
      │
      ▼
Snowflake Intelligence

This allows the agent to work with an established multidimensional business model instead of independently reconstructing that model for each question.

What the Agent Can Help With

Section titled “What the Agent Can Help With”

Depending on the configured agent tools and the user’s privileges, the Casabase Cube agent can assist with tasks such as:

  • Discovering available cubes
  • Exploring dimensions, hierarchies, and members
  • Translating business questions into cube POVs
  • Running multidimensional queries
  • Drilling into hierarchy branches
  • Changing the analytical point of view
  • Explaining Casabase Cube concepts and usage
  • Assisting with supported diagnostic workflows

For example:

What cubes are available?
Show revenue by product category for FY26.
Break North America down by entity.
Compare actual and budget operating income.
How is Gross Margin calculated?

The exact capabilities available to a user depend on the tools configured for the agent and the privileges granted in Snowflake and Casabase Cube.

The Agent Uses Casabase Cube Query Semantics

Section titled “The Agent Uses Casabase Cube Query Semantics”

For analytical questions, the agent should use Casabase Cube’s supported query interfaces rather than independently reproducing cube calculations against the underlying fact table.

This is important because a multidimensional result can depend on more than stored fact rows.

Casabase Cube query semantics can include:

Stored Fact Data
      +
Hierarchy Relationships
      +
Dynamic Aggregation
      +
Member Formulas
      +
Solve Order
      +
Time Behavior
      +
Row-Level Security
      │
      ▼
Analytical Result

Querying the underlying fact table alone does not reproduce all of these behaviors.

Using the Casabase Cube query engine keeps AI-generated analysis aligned with the same governed business logic used elsewhere in the application.

Natural Language to POV

Section titled “Natural Language to POV”

A central responsibility of the Casabase Cube agent is translating natural language into a cube Point of View.

For example:

Show January revenue for the United States.

can be translated conceptually into:

{
  "Account": ["Revenue"],
  "Period": ["Jan"],
  "Entity": ["United States"]
}

The actual POV depends on the dimensions and members defined in the selected cube.

The agent can use cube discovery capabilities to understand that structure before constructing the query.

This is especially useful when users know the business question they want to answer but do not know the underlying SQL or Casabase Cube procedure syntax.

Conversational Analysis

Section titled “Conversational Analysis”

Because the agent operates within a conversational AI experience, users can refine an analysis through follow-up questions.

For example:

Show revenue by region for FY26.

followed by:

Break North America down to the next level.

and then:

Compare that with FY25.

The conversational layer can maintain context while the agent translates each request into the appropriate cube operations.

Snowflake controls the conversational experience, orchestration behavior, model capabilities, and other Snowflake Intelligence or Cortex Agent features. These capabilities can evolve independently of Casabase Cube.

Snowflake Intelligence and Cortex Agents

Section titled “Snowflake Intelligence and Cortex Agents”

Casabase Cube provides the cube-specific capabilities used by the agent.

Snowflake provides the agent platform and user experience.

At a high level:

Layer Responsibility
Snowflake Intelligence / agent experience Natural-language interaction and presentation
Cortex Agent Request orchestration and tool selection
Casabase Cube agent tools Cube discovery, querying, and supported application operations
Casabase Cube Multidimensional modeling, calculation, aggregation, and application security
Snowflake Data platform, compute, identity, roles, and platform governance

This separation is useful when troubleshooting or administering the integration because a user must have the required access at each applicable layer.

Security

Section titled “Security”

Using the Casabase Cube agent does not create a separate path around Casabase Cube security.

Queries executed through the supported Casabase Cube query path are evaluated using the applicable user context and Casabase Cube row-level security.

Conceptually:

User
 │
 ▼
Snowflake Access
 │
 ▼
Agent Access
 │
 ▼
Casabase Cube Application Access
 │
 ▼
Casabase Cube Row-Level Security
 │
 ▼
Permitted Result

If Casabase Cube row-level security restricts a user to a particular hierarchy branch, agent-generated cube queries are subject to that same restriction.

As elsewhere in Casabase Cube, row-level security is opt-in per user. A user with application access but no active Casabase Cube security rules for the cube is unrestricted by Casabase Cube row-level security.

See Governance and Security for the integration security model.

AI Does Not Replace the Cube Model

Section titled “AI Does Not Replace the Cube Model”

The agent should not invent new business logic merely because a requested metric is not defined in the cube.

For example, if the cube contains:

Revenue
Cost of Sales
Gross Margin

the agent can use those governed members and calculations.

If a user asks for a business metric that is not defined in the cube, the appropriate response depends on the agent’s configured capabilities and the user’s privileges. The absence of a governed calculation should not be silently replaced with an invented formula presented as though it were part of the cube.

This distinction helps keep natural-language analysis aligned with the organization’s defined analytical model.

Administration and Modification

Section titled “Administration and Modification”

Agent capabilities are governed by the tools included in the agent and by the privileges of the user invoking them.

General business users should typically be granted only the capabilities required for discovery and analysis.

Administrative operations should remain restricted to appropriately authorized users.

The presence of an AI interface does not change the Casabase Cube application-role model:

CUBE_PUBLIC
    │
    └── Query and discovery capabilities


CUBE_ADMIN
    │
    └── Administrative capabilities

See Roles and Privileges for the current application-role model.

What Snowflake Provides

Section titled “What Snowflake Provides”

Features of the Snowflake Intelligence or Cortex Agent experience are provided by Snowflake rather than Casabase Cube.

These can include capabilities such as:

  • Conversational interaction
  • Agent orchestration
  • Tool selection
  • Conversation context
  • Response presentation
  • Visualization capabilities
  • Agent management and monitoring

Because Snowflake can change these capabilities independently, use the current Snowflake documentation for platform-specific behavior and user-interface instructions.

Casabase Cube documentation focuses on how the Casabase Cube agent interacts with the cube model and which Casabase Cube capabilities are exposed through that integration.

Section titled “Recommended Reading”

Continue with:

  1. How Casabase Cube Works with Snowflake Intelligence to understand the request and query flow.
  2. Setting Up the Agent to configure the integration.
  3. What the Casabase Cube Agent Can Do for supported agent capabilities.
  4. Using Casabase Cube in Snowflake Intelligence for end-user workflows.
  5. Governance and Security for access and security behavior.
  6. Tips and Limitations for effective use and current constraints.
Section titled “Related Topics”