Anomaly Insights¶

Overview¶

Anomaly Insights uses AI to detect unusual patterns in your data. It identifies statistical outliers, time series anomalies, pattern deviations, and data integrity issues - helping you catch problems before they impact your application.

Key Features¶

• Scheduled Analysis: Automatic anomaly detection on a schedule (hourly, daily, weekly)
• Manual Triggers: Run analysis on-demand when needed
• Configurable Sensitivity: Tune detection from conservative to aggressive
• Severity Levels: Insights are categorized by severity for prioritization
• Notifications: Get alerted when anomalies are detected

Insight Types¶

Severity Levels¶

Configuration¶

Anomaly detection is configured per structure in the Console UI under the AI Insights tab.

Analysis Frequency¶

• Hourly: Best for high-volume, real-time applications
• Daily: Recommended for most use cases
• Weekly: For low-volume or less time-sensitive data

Sensitivity Level¶

• Low: Only significant anomalies (fewer false positives)
• Medium: Balanced detection (recommended)
• High: Detect subtle anomalies (more comprehensive, may have more false positives)

Analysis Types¶

Select which types of analysis to run: - Time Series: Detect temporal patterns and trends - Statistical: Find outliers using statistical methods - Pattern: Identify deviations from expected patterns

Lookback Period¶

How far back to analyze records (1-90 days). Longer periods provide more context but take longer to analyze.

Minimum Records¶

Minimum number of records required before analysis runs. More records produce more reliable results.

Using Anomaly Insights¶

Enable Detection¶

1. Navigate to your structure in the Console
2. Click the AI Insights tab
3. Toggle Enable Anomaly Detection
4. Configure frequency, sensitivity, and analysis types
5. Click Save Configuration

Run Manual Analysis¶

To analyze immediately:

1. Go to the AI Insights tab
2. Click Run Analysis Now
3. Results appear automatically when ready

Review Insights¶

Active insights appear in the Anomaly Insights table:

• Acknowledge: Mark as reviewed (you're aware of the issue)
• Dismiss: Mark as not relevant (false positive)
• View Details: See full insight information and affected records

Insight Details¶

Each insight shows: - Severity: Critical, Warning, or Info - Type: Category of anomaly detected - Title: Brief description of the issue - Field: The affected field (if applicable) - Confidence: How confident the AI is (0-100%) - Description: Full explanation of the anomaly - Affected Records: Links to records involved

Notifications¶

Configure which severity levels trigger notifications: - Critical: Recommended for all users - Warning: Recommended for data-sensitive applications - Info: Optional, for comprehensive monitoring

SDK Usage¶

The SDK provides methods to work with anomaly insights:

// Trigger anomaly analysis
const result = await centrali.anomalyInsights.triggerAnalysis('orders');
console.log('Analysis started:', result.data.batchId);

// List active insights
const insights = await centrali.anomalyInsights.list({
  status: 'active',
  severity: 'critical'
});

// List insights for a specific structure
const orderInsights = await centrali.anomalyInsights.listByStructure('orders');

// Get insight details
const insight = await centrali.anomalyInsights.get('insight-id');
console.log('Title:', insight.data.title);
console.log('Affected records:', insight.data.affectedRecordIds);

// Acknowledge an insight
await centrali.anomalyInsights.acknowledge('insight-id');

// Dismiss an insight
await centrali.anomalyInsights.dismiss('insight-id');

// Bulk acknowledge
await centrali.anomalyInsights.bulkAcknowledge(['id1', 'id2', 'id3']);

// Get insights summary
const summary = await centrali.anomalyInsights.getSummary();
console.log('Active:', summary.data.active);
console.log('Critical:', summary.data.bySeverity.critical);

Realtime Events¶

Subscribe to anomaly events via SSE:

const subscription = centrali.realtime.subscribe({
  structures: ['orders'],
  events: ['anomaly_insight_created', 'anomaly_detection_completed'],
  onEvent: (event) => {
    if (event.event === 'anomaly_insight_created') {
      console.log('New anomaly:', event.data.title);
      console.log('Severity:', event.data.severity);
    }
  }
});

Best Practices¶

Start with Medium Sensitivity¶

Begin with medium sensitivity and adjust based on false positive rates.

Set Appropriate Lookback Period¶

• High-volume data: 7 days
• Medium-volume: 14-30 days
• Low-volume: 30-90 days

Act on Critical Insights Quickly¶

Critical insights indicate potential data integrity issues. Investigate and resolve promptly.

Review Dismissed Insights Periodically¶

Occasionally review dismissed insights to ensure you're not missing recurring issues.

Use Notifications for Critical¶

Enable notifications for critical severity to catch urgent issues immediately.

Limits¶

Related Documentation¶

• AI Validation - Data quality validation
• Schema Discovery - Automatic schema evolution
• Structures & Records - Data schemas and entries