Skip to main content

Predictions & Intelligence

The Predictions & Intelligence page is the platform's proactive-signals hub — it answers what's predicted to fail, where you're exposed, what's drifting, and how a change would ripple, all from one screen. Signals are derived from the knowledge graph, the alert pipeline, and the platform's scoring models. The header carries two global actions — Edit dashboard and Export PDF — that apply to whichever view is open.

Views​

The page is organised into five tabbed views:

  • Predicted failures — Entities the predictor has scored above the surface threshold in the selected window
  • Security posture — Reachability, RBAC and compliance findings; a summary of the dedicated Security Posture page
  • Anomalies & forecasts — Metrics drifting from baseline and their short-term forecasts
  • Correlations & simulations — How a change to one entity would ripple across dependent services
  • Business view — The same signals rolled up to a business-impact lens

Predicted failures​

The default view. Three counters summarise the current window:

  • At-risk entities — Count of entities currently scored above the threshold
  • Highest score — The single highest risk score across the fleet
  • With breach ETA — How many at-risk entities have a projected time-to-failure

Two controls shape the table:

  • Window — The look-back window for scoring (e.g. 1 hour)
  • Score threshold — Only entities scoring at or above this value are listed (e.g. Score ≥ 30)

Each row scores one entity:

ColumnDescription
EntityThe scored resource (pod, host, monitor, …)
ScoreComposite risk score
Top reasonThe dominant contributing factor (e.g. 5054 restarts)
ETAProjected time to failure, when one can be computed
Contributing signalsThe weighted signals that fed the score — e.g. Restart Signal, K8s Event Signal, Alert Signal, each with its own contribution

Note: Scores are composite — a single high-weight signal, such as a pod restart storm, can push an entity above the surface threshold on its own.

Best Practices​

  • Open this page at the start of a shift — it surfaces what's likely to break before an alert fires.
  • Lower the score threshold to widen the net during a fragile change window; raise it to focus only on the highest-risk entities.
  • Use the Security posture tab for a quick read, then open the dedicated Security Posture page when you need to trace an exposure.
  • Export PDF produces a shareable snapshot of the active view for standups and reviews.