Skip to content

demo-script.md

Latest commit

 

History

History
161 lines (136 loc) · 5.64 KB

demo-script.md

File metadata and controls

161 lines (136 loc) · 5.64 KB

Predictive Maintenance Software - Demo Script

Overview

This script provides a step-by-step walkthrough for demonstrating the Predictive Maintenance Software. Follow these 10 steps to showcase the key features and workflows.

Demo Walkthrough (10 Steps)

1. Login as Manager

  • Navigate to the login page
  • Enter credentials:
    • Email: manager@example.com
    • Password: Password123!
  • Click "Sign in"
  • Point to demonstrate: Role-based authentication, JWT cookie storage

2. Show KPI Cards on Dashboard

  • After login, you'll see the Manager Dashboard
  • Point out:
    • Overall Efficiency: 92%
    • Downtime Hours: 24h
    • Maintenance Cost: $45K
    • Equipment Status: 28/30 operational
  • Point to demonstrate: High-level metrics, department overview, recent reports

3. Navigate to Machines Page

  • Click "Machines" in the navigation bar
  • Show the list of 6 machines with their statuses
  • Point out:
    • Different machine types (CNC Lathe, Milling Machine, etc.)
    • Status badges (Operational, Warning, Maintenance)
    • Efficiency percentages
  • Point to demonstrate: Machine inventory management, status visualization

4. Open a Machine and Run Prediction

  • Click the "Predict" button on "Machine B - Production Line 2" (has warning status)
  • In the modal:
    • Show date range selector (default: last 30 days)
    • Click "Run Prediction"
    • Wait for prediction to load (~1 second)
  • Point out:
    • Gauge visualization showing score
    • Risk level badge
    • Explanation text
    • Top 3 contributing factors with impact values
  • Point to demonstrate: Predictive analytics, explainable AI features

5. Create Alert (if risk is HIGH)

  • If the prediction shows risk: HIGH:
    • Click "Create Alert" button (red button at bottom)
    • Alert is created successfully
    • Modal closes automatically
  • Point to demonstrate: Alert creation workflow, risk-based actions

6. View Alerts in AlertsPanel

  • Return to Dashboard (click "Dashboard" in nav)
  • Scroll to see the AlertsPanel at the top
  • Show the newly created alert
  • Point out:
    • Alert title and machine name
    • Risk level badge
    • Score and status
    • Creation timestamp
  • Point to demonstrate: Real-time alert monitoring, alert management

7. Switch to Maintenance Role

  • Logout (click "Logout" button)
  • Login as Maintenance:
    • Email: maintenance@example.com
    • Password: Password123!
  • Show the Maintenance Dashboard
  • Point out:
    • Pending Tasks: 7
    • In Progress: 2
    • Completed: 15
    • Upcoming maintenance tasks with priorities
    • Predictive maintenance alerts
  • Point to demonstrate: Role-specific dashboard, maintenance workflow view

8. Maintenance Acknowledges Alert

  • Scroll to AlertsPanel on Maintenance Dashboard
  • Find an active alert
  • Click "Acknowledge" button
  • Point out:
    • Alert status changes to "Acknowledged"
    • Shows who acknowledged it (maintenance@example.com)
    • Shows timestamp of acknowledgment
  • Point to demonstrate: Alert acknowledgment workflow, user logging

9. Manager Views History

  • Logout and login back as Manager
  • Click "History" in navigation bar
  • Show the History page with two tabs:
    • Predictions tab: Shows all prediction records
    • Alerts tab: Shows all alerts (including acknowledged ones)
  • Point out:
    • Date filtering options
    • Table view with machine names, scores, risk levels
    • Timestamps for all actions
    • User attribution (who ran predictions, who acknowledged alerts)
  • Point to demonstrate: Audit trail, historical data tracking

10. Export CSV

  • On History page, ensure there's visible data
  • Click "Export CSV" button at the top right
  • Point out:
    • CSV file downloads automatically
    • Contains all visible data (predictions or alerts based on selected tab)
    • Includes all relevant fields (machine, score, risk, dates, users)
  • Point to demonstrate: Data export functionality, reporting capabilities

Quick Tips for Presenter

  1. Before starting: Run /api/seed endpoint to populate demo data (predictions and alerts)
  2. Timing: Each step takes 1-2 minutes; full demo ~15-20 minutes
  3. Highlights:
    • Focus on role-based features (Manager vs Maintenance)
    • Emphasize explainability (top features, SHAP-like values)
    • Show real-time updates (alerts appearing after creation)
  4. If something breaks: All endpoints are mock APIs, so data resets on refresh
  5. Key selling points:
    • Predictive analytics prevents downtime
    • Explainable AI builds trust
    • Role-based workflows streamline operations
    • Historical tracking enables continuous improvement

Demo Account Credentials

Role Email Password
Operator operator@example.com Password123!
Maintenance maintenance@example.com Password123!
Manager manager@example.com Password123!

Additional Features to Mention (if time permits)

  • Machine Details: Click on the machine name to see detailed information
  • Date Filtering: Filter predictions/alerts by date range
  • Risk Levels: Color-coded risk indicators throughout the app
  • Model Versioning: Track which ML model version was used
  • Future Enhancements:
    • Email/Slack notifications
    • Full SHAP explanations
    • Interactive charts and graphs
    • Mobile app for on-site maintenance

Troubleshooting

  • Can't see alerts: Check if prediction was created with risk=HIGH
  • CSV not downloading: Ensure there's visible data in the current tab
  • Login fails: Verify password is Password123! (case-sensitive)
  • No demo data: POST to /api/seed endpoint to populate sample data