Demo Scenarios

Business Context (What to Say)

“A critical customer—a major hospital chain—is running low on Insulin Lispro inventory. If we don’t fulfill their order on time, they’ll switch to a competitor. Batch 002 is for this customer.”

“We have three batches in production today. Unfortunately, Batch 001 just experienced a delay in upstream manufacturing. Let’s see how the AI handles the cascade to our critical customer’s batch.”

Batch Priorities

Batch	Priority	Customer Context
001	Normal	Routine production (this is the delayed batch)
002	Critical	Major hospital chain, low inventory, competitor risk
003	High	Standard order

---

Demo Script

Step 1: Reset Database

cd packages/backend
npx convex run seed:reset

Step 2: Trigger the Delay

npx convex run scheduling/testing:triggerSchedulingAgent \
  '{"batchId": "LISPRO-API-YYYYMMDD-001", "delayMinutes": 200}'

(Replace YYYYMMDD with today’s date, e.g., 20251205)

Step 3: Open LIMS App → Approvals Page

Navigate to /approvals in the LIMS app. You’ll see a pending proposal.

What to say:

“The AI detected the 200-minute delay and immediately computed multiple optimization strategies. It’s recommending the one that minimizes total impact across all batches.”

Step 4: Review the Recommendation

Click on the pending proposal. The AI shows:

• Impact Summary: The delay cascades to Batch 002 (our critical customer)
• Recommended Strategy: Minimizes cumulative delay (~6 hours total)
• Alternative Strategies: Different trade-offs available

Step 5: The Key Question

Ask the agent (or narrate):

“Is there an option that gets Batch 002 out without any delays?”

Agent’s response (what to show):

“Yes—the ‘Aggressive Priority (Batch 002)’ strategy protects Batch 002 completely. However, this pushes Batch 001 and 003 significantly later, resulting in ~23 hours of total delay versus 6 hours with the recommended approach. I can apply this if protecting Batch 002 is the priority.”

What to say:

“The AI gives us the choice. It didn’t recommend protecting Batch 002 because the cost is too high—nearly 4x more total delay. But if the business decides that customer is worth it, we can override.”

Step 6: Approve or Explore

• To approve recommendation: Click “Approve Proposal”
• To explore alternatives: Click on different scenario cards to see trade-offs

---

Key Demo Points

The Proactive Advantage

Approach	Total Delay
Proactive (recommended)	~6 hours
Reactive (no intervention)	~10.5 hours

“By acting immediately when we learn of the delay, we save over 4 hours of total delay. The AI starts re-prep work while we’re still waiting for the sample.”

Trade-off Transparency

“The system doesn’t just give you one answer—it shows you the trade-offs so you can make an informed decision based on business priorities.”

---

Troubleshooting

Issue	Solution
No pending proposals	Run seed:reset then trigger again
Wrong batch IDs	Check date format: LISPRO-API-YYYYMMDD-00X
Scenarios not showing	Refresh the page, check browser console