AI Investigations

Overview

When Argus detects an anomaly, it can automatically launch an AI investigation. The AI agent analyzes correlated metrics, reviews recent changes, and identifies the most likely root cause.

How It Works

1. Anomaly Detection — Argus detects unusual metric patterns (e.g., CPU spike, memory leak)
2. Context Gathering — The AI collects related metrics, recent events, and historical patterns
3. Analysis — The AI correlates data across multiple signals to identify causation
4. Root Cause Report — A natural language summary with the identified root cause and recommended actions
5. Remediation — If webhooks are configured, the AI can execute remediation steps

Investigation Features

• Cross-metric correlation — Analyzes CPU, memory, disk, network together
• Timeline reconstruction — Builds a timeline of events leading to the incident
• Service dependency awareness — Understands relationships between services
• Actionable recommendations — Suggests specific fixes, not just descriptions

Assigning Investigations

Investigations can be assigned to team members:

1. Navigate to the Investigations page
2. Click on an investigation
3. Use the Assign dropdown to assign to a team member
4. The assigned member receives a notification

Investigation History

All investigations are stored and searchable. Use them to:

• Build a knowledge base of past incidents
• Identify recurring patterns
• Track mean time to resolution (MTTR)