Looker Introduces Agentic Workflows for Automated Data Monitoring and Root-Cause Analysis

Automated Monitoring and Analysis

Looker has introduced Agentic Workflows, a new capability in preview that automates the monitoring of business metrics and performs root-cause analysis. This feature aims to reduce the manual effort typically required by data analysts to identify reasons behind metric changes, moving beyond traditional alerts that only indicate a change has occurred.

Natural Language Configuration

Building on Looker's Conversational Analytics, Agentic Workflows allows users to configure automated monitoring routines using natural language prompts. Users can ask the system to monitor specific metrics weekly or notify them if certain values exceed defined thresholds. The agent then interprets the intent, confirms conditions, and generates a workflow plan for review before activation.

Diagnostic Summaries and Integration

When a monitored metric crosses a defined threshold, the background agent automatically performs a Key Driver Analysis (KDA) on the underlying data model. This analysis identifies specific factors contributing to the change, such as product categories or customer cohorts. The diagnostic summary is then delivered directly to team workspaces via Slack or email, providing immediate insights without manual intervention.

Deeper Investigation and Oversight

Each notification includes a direct link to Conversational Analytics in Looker, pre-loaded with the agent's diagnostic findings. This allows users to conduct follow-up investigations and test hypotheses interactively. Looker also provides a centralized workflow management interface for administrators to oversee, adjust, or disable workflows across the entire instance, balancing user flexibility with enterprise governance.

Availability

Looker Agentic Workflows is currently available in preview. It requires Looker version 26.08 or later. Administrators can enable the feature through the Gemini in Looker settings page.