Trigger: Daily batch job or real-time event from PLM project management module (e.g., a task completion, document submission, or schedule update).
Context/Data Pulled: AI agent queries the PLM system for:
- Current project phase and gate deliverables.
- Historical completion times for similar tasks/phase-gates.
- Open action items and their age.
- Resource allocation and availability data.
- Sentiment analysis from recent collaboration feed comments.
Model or Agent Action: A machine learning model trained on historical project success/failure data generates a health score (e.g., 0-100) and a confidence interval. It flags key risk drivers (e.g., "Phase 3 documentation is 14 days behind historical average").
System Update or Next Step: The score and top risk factors are written to a dedicated table in the data warehouse powering the BI dashboard (e.g., Power BI dataset). The dashboard tile for "Project Portfolio Health" updates automatically, color-coding projects (Green/Yellow/Red).
Human Review Point: The dashboard includes a "Drill-Through" action. Clicking a red project opens a detailed view with the AI's reasoning and suggested mitigation actions (e.g., "Recommend reassigning 10 hours from Project B to address critical path delay"). The project manager can accept, modify, or ignore the recommendations.