Demo Runbook
Pre-Demo Checklist
Section titled “Pre-Demo Checklist”# 1. Reset databasecd packages/backendnpx convex run seed:reset
# 2. Start all appspnpm devVerify:
- 3 batches created (001=normal, 002=critical, 003=high)
- 15 LIMS tests scheduled
- No pending proposals
Demo Script
Section titled “Demo Script”Phase 1: Set the Scene (2 min)
Section titled “Phase 1: Set the Scene (2 min)”Open MES App → Batches
“We have three batches in production today for Insulin Lispro. Batch 002 is for a critical customer—a major hospital chain running low on inventory. If we don’t deliver on time, they’ll switch to a competitor.”
Show the batch priorities:
- Batch 001: Normal (routine production)
- Batch 002: Critical (hospital chain)
- Batch 003: High (standard order)
Phase 2: Trigger the Delay (1 min)
Section titled “Phase 2: Trigger the Delay (1 min)”“Unfortunately, Batch 001 just experienced a 200-minute delay in upstream manufacturing. Let’s see how this affects our critical customer’s batch.”
Execute (not visible to audience):
npx convex run scheduling/testing:triggerSchedulingAgent \ '{"batchId": "LISPRO-API-YYYYMMDD-001", "delayMinutes": 200}'(Replace YYYYMMDD with today’s date)
Wait 5-10 seconds for agent to compute.
Phase 3: View AI Recommendation (3 min)
Section titled “Phase 3: View AI Recommendation (3 min)”Navigate to: LIMS App → Approvals
Click the pending proposal.
“The AI detected the delay and computed multiple strategies. It’s recommending ‘Minimize Cumulative Delay’ which results in about 6 hours of total delay across all batches.”
Point out:
- Batch 002 (critical) shows a cascade delay of ~140 minutes
- The AI explains why: resource conflicts with the delayed batch
Phase 4: The Key Question (3 min)
Section titled “Phase 4: The Key Question (3 min)”This is the demo highlight.
“Wait—Batch 002 is for our critical customer. Is there an option that gets Batch 002 out without any delays?”
Click on “Aggressive Priority (Batch 002)” scenario card.
“Yes! The AI computed an option that completely protects Batch 002. But look at the trade-off: total delay jumps from 6 hours to 23 hours. That’s why it wasn’t recommended—but the choice is ours.”
Key message:
“The AI doesn’t just give one answer. It shows the trade-offs so you can make an informed decision based on business priorities.”
Phase 5: Approve and Show Results (2 min)
Section titled “Phase 5: Approve and Show Results (2 min)”Either:
- Approve the recommended strategy (balanced approach)
- Or select the aggressive option to protect Batch 002
Navigate to: MES App → Batches
“The schedule updates flow automatically to all systems. Full audit trail of who approved what and when.”
Voice Agent Version
Section titled “Voice Agent Version”If using the voice agent for approval:
Agent calls: “I have a schedule change proposal. Batch 001 has a 200-minute delay that cascades to Batch 002. I’m recommending the minimum cumulative delay option.”
Manager asks: “Can we protect Batch 002 completely? That’s for a critical customer.”
Agent responds: “Yes, there’s an aggressive priority option that eliminates Batch 002’s delay. However, total delay increases from 6 hours to 23 hours. Would you like me to apply that instead?”
Quick Reset
Section titled “Quick Reset”Between demos:
cd packages/backendnpx convex run seed:resetTroubleshooting
Section titled “Troubleshooting”| Issue | Solution |
|---|---|
| No pending proposals | Wait 10s, then check Convex dashboard for errors |
| Wrong batch IDs | Use format LISPRO-API-YYYYMMDD-00X with today’s date |
| Scenarios not showing | Run seed:reset and trigger again |