AI Insights

AI Insights help teams spot anomalies, summarize trends, and recommend actions based on DSAL metrics and workflow events.

How It Works

1. Metrics and events feed into the Insights service once per sync cycle.
2. Vertex AI (or your configured provider) analyses trends, compares against thresholds, and generates summaries.
3. Insights appear in Analytics → Insights and can be pushed to Slack or email.

Insight Types

• Trend Summaries – Natural language write-ups highlighting increases/decreases in throughput, cycle time, or backlog size.
• Anomaly Alerts – Detect sudden spikes or drops and provide probable causes (e.g., "Review state exceeds WIP limit for 3 days").
• Recommendations – Suggest actions such as rebalancing workloads, splitting issues, or updating workflows.

Configuration

• Enable AI Insights in Organization → Analytics Settings.
• Choose model provider (Vertex AI by default). Configure API keys/secrets via environment variables or Secret Manager.
• Set data retention policies to control how much history the AI can access.

Custom Prompts

• Override default prompts by adding .kiket/analytics/prompts/*.yaml with context and formatting instructions.
• Use variables ({{ metric.value }}) to embed numbers into the generated text.

Privacy & Compliance

• Sensitive fields can be excluded via redaction rules in analytics/settings.yaml.
• Audit logs record every generated insight and who viewed it.

AI Insights turns raw metrics into actionable narratives so teams can react quickly to changes.