How the Valsight AI Assistant Works

The Valsight AI Assistant is accessed through a chat interface where questions are answered conversationally. Each question is evaluated against the currently available model and project context and answered in real time.
Responses are streamed progressively and presented in a structured, readable format. Conversations persist per browser tab and remain available after refresh.

What the AI uses to answer questions

The AI Assistant builds its answers exclusively from non-numeric, descriptive information, including:

Model structure

Node names, types, hierarchies, and relationships
Dimensions and levels
Formula logic and syntax

Data sources

Table and column names
Column metadata and units

Project information

Project configuration
Simulations and submodels

Annotations

Node descriptions
Comments and labels

Model changes are reflected immediately in the AI’s context.

What the AI cannot access

To preserve data privacy and correctness, the AI Assistant does not have access to:

Numeric values or calculated results
Historical data values
External systems outside Valsight
Personal user information or credentials
Projects or data beyond your permissions

Response behavior

AI responses may include:

Structured explanations using headings and lists
Scrollable code blocks
Formula explanations

While a response is being generated, a thinking indicator signals ongoing processing.

Responses are informational in nature and intended to support understanding and orientation.

Interaction limits and safeguards

The AI Assistant operates strictly in read-only mode
It cannot modify models, data, or scenarios
It does not perform numeric analysis or comparisons
Answers reflect documented structure and metadata only

AI-generated responses should always be validated against business knowledge and modeling intent.

Access requirements

Available in Valsight version 7.0 or later.

The AI Assistant respects your access rights:

You see what you can access - AI context includes only projects and models you have permission to view
No privilege escalation - The AI cannot access restricted projects or data on your behalf
Read-only mode - The AI can view context but cannot modify it