AI Features Overview

Sifflet uses AI to help you detect data issues faster, understand incidents more easily, and keep your data documentation up to date. This page explains the AI-powered features available in the platform, what value they provide, and how they handle your data.

We distinguish between two types of information:

Metadata: information about your assets, such as table and column names, descriptions, tags, and SQL queries.
Data: the actual rows and values stored in your data warehouse or data lake. This can contain sensitive or personal information.

Our AI features are designed to primarily work with metadata, and only use limited samples or aggregates of data where strictly necessary.

Features using metadata only:

AI Assistant in the product
- Text to SQL
- Text to Monitor
- Text to Regex
- Monitor Title & Description Suggestion
- SQL Query Correction
Monitor Recommendation (Sentinel)
Incident Grouping
Incident Root Cause Analysis (Sage)

Features using metadata and data:

Metadata Suggestion

Configuration and Controls

We recognize that every organization has its own constraints and policies around AI and data protection. Sifflet provides centralized controls so you can align AI usage with your governance requirements.

Typical controls include:

Global AI enable / disable:

Turn AI-based features on or off for your environment, depending on your security posture. This setting is not available through the UI and needs to be requested directly to Sifflet. Note that disabling all AI features may result in a degraded experience with Sifflet platform.
Feature-level configuration:

Disable only the AI features that use your company's data. This setting is available in the administration settings under AI Settings.