SnapGPT settings

The SnapGPT settings screen in Admin Manager lets you control SnapGPT behavior for your environment. It provides the following options:

Settings: Configure feature toggles and RAG behavior for your environment.
Standards: Create and manage Markdown-based standards that SnapGPT applies when generating or refining pipelines.

Settings

The following SnapGPT settings are available for your environment:


Setting		Purpose
Enable SnapGPT		Turns SnapGPT on for users in the environment.
Schema retrieval		When enabled, SnapGPT uses the schema to create more accurate SQL queries for SQL Snaps. If disabled, SnapGPT creates queries that will likely require editing before use.
Use preview data schema		When enabled, SnapGPT has access to preview data to generate expressions. If disabled, generated expressions might not be as accurate.
Describe pipeline		When enabled, users can ask SnapGPT to create pipeline documentation. No sensitive information is sent to the LLM. If disabled, the describe pipeline feature isn't available.
Use preview data for contextual metadata		When enabled, users can right-click any Snap of a successfully validated pipeline in Designer and select Get preview data insights in SnapGPT to receive insights into key fields, business and validation rules, and Personally Identifiable Information (PII). Learn more about getting preview data insights. Important: Enabling this setting allows SnapGPT to view and analyze sensitive information such as PII data.
Use my existing pipelines for enhanced pipeline generation	Auto	SnapGPT automatically selects pipelines from the environment to use as examples for Retrieval Augmented Generation (RAG). Users can choose Auto, Manual, or Disable in their personal SnapGPT settings.
	Manual	Users select pipelines manually for SnapGPT to use as examples for RAG. Users can choose Manual or Disable in their personal settings.
	Disable	Disables RAG for the environment and in personal settings.

To restrict specific users from accessing SnapGPT, refer to Block users from SnapGPT.

Pipeline standard

Add naming conventions, error-handling rules, or other organizational standards in Markdown (.md) format for SnapGPT to use when generating or refining pipelines. You can enter standards directly in the Markdown editor or upload a Markdown file.

Standards set to Always are enforced on every generate or refine pipeline request. Standards set to Optional appear in SnapGPT for users to select themselves.

Note: Up to 100 standards can be stored for an environment.

Pipeline standard tab in SnapGPT settings

How it applies


Value	Behavior
Always	Applied automatically for all users in your environment on every pipeline generation and refinement request. Users cannot opt out.
Optional	Stays published but is not applied automatically for your environment. Users can enable it from the Standards panel in SnapGPT.

Note: Changing How it applies requires confirmation.

Add a standard

Use either of the following methods:

Click New standard: Enter a name (.md is appended automatically), enter the information for the standard in the Markdown editor, use Preview to check rendering, then click Create standard.
Click Upload: Drag and drop or browse for an existing .md file.

When standards are published, users can apply them from the + menu in the chat input box. Refer to SnapGPT pipeline standards.