predictive-action-intelligence

name	predictive-action-intelligence
description	Use when designing proactive Dataverse recommendations such as churn risk, follow-up needs, API limit risks, compliance violations, pipeline risk, or next-best actions. Produces analysis design only.

Predictive Action Intelligence

Purpose

Design proactive recommendations that detect Dataverse risks and suggest actions before users ask.

Use this skill for:

• pipeline risk analysis
• churn and follow-up recommendations
• flow/API limit risk warnings
• compliance risk detection
• stale record and process bottleneck detection
• next-best-action recommendations

V1 Boundary

This skill produces prediction and recommendation design guidance only. It does not train models, query Dataverse, monitor flows, or create alerts.

Workflow

1. Define the prediction target and business outcome.
2. Identify candidate signals from Dataverse records, relationships, status changes, activity history, flow runs, and process milestones.
3. Define recommendation types: warn, explain, ask, create task, start workflow, escalate, or request approval.
4. Define confidence, evidence, thresholds, false-positive handling, and user feedback capture.
5. Define governance: who can see predictions, when action requires approval, and how explanations are logged.

Output Format

Return:

• Prediction scenario: risk or opportunity to detect.
• Signals: data and process indicators to inspect.
• Recommendation model: suggested actions and confidence thresholds.
• Explainability: evidence and rationale shown to the user.
• Governance: visibility, approvals, feedback, and audit.