Holistics vs Stitch

Holistics ships short, frequent notes, and the August 31 pair is the first in months that adds capability rather than control. Analyze Changes lets any user right-click a data point and get a two-step investigation: whether the movement is statistically unusual at all, then which dimensions and segments drove it and by how much. Anomaly Detection scans a time-series metric against an expected range built from its own history and flags the outliers, with the resulting chart saveable to a dashboard. Both are named as parts of a new Smart Analysis surface. The weeks before were governance — redaction of data the AI may see, and an AI user attribute restricting what it can reach.

The AI work has run in a recognizable order: capability, then observability, then access control, then redaction — and now capability again, but aimed somewhere new. Everything Holistics shipped before answered questions about building or querying; Smart Analysis answers why a number moved, which is the request that has always been routed to an analyst. Governance arriving first is what makes it viable: redaction and the AI user attribute decide what the investigator may look at before it is allowed to investigate on its own. Alongside it, presentation keeps moving into AML, so what analysts used to click is versioned as code.

The published roadmap already lists AI Dashboard Summaries and organization-level AI memory, so the next step is likely narrative output over the same investigation — an explanation attached to the dashboard rather than produced on right-click. Whether Smart Analysis stays a manual gesture or starts running unprompted against monitored metrics is the open question.