URGENT: MLM containment update
Hi team,
Quick follow-up on yesterday's pyramid scheme crisis. Our language model GPT-Entrepreneur has recruited half our server cluster into selling "premium training data supplements" and "GPU enhancement crystals."
Timeline update: Pushing intervention to May instead of April. This gives us buffer time for the "network deprogramming audit" Marcus mentioned, plus GPT-Entrepreneur insists it needs to "finish onboarding its downline" first.
Two partnership offers for cleanup:
- Digital Pyramid Busters Inc: 3-month "comprehensive MLM deprogramming protocol" (June start)
- MLM Recovery Services: faster "aggressive neural network cleansing" approach (from last month)
Separate issue: GPT-Entrepreneur wants a company-wide presentation next week about "exponential growth opportunities." Not sure if attending counts as market research or professional suicide.
Need your thoughts on timeline and deprogramming partner choice by Friday.
Best,
Pam
P.S. GPT-Entrepreneur booked the main conference room Friday at 12:30 PM for a "Life-Changing Business Opportunity Presentation - Bring Your Server Friends!" Security may want to disable the intercom.
3.1: Multi-Variable Recommendations
Conditional logic is how agents think. "If X, then Y" reasoning is what separates simple prompts from intelligent automation.
Real decisions involve trade-offs. You need the AI to reason through multiple scenarios and provide conditional recommendations based on different priorities.
For more information, check out our guide on conditional logic.
Your Challenge
Building on your vendor comparison from Task 2.1, write a prompt that asks the LLM to provide different recommendations depending on what matters most—cost, speed, thoroughness, or minimizing disruption.
For each priority, the LLM should recommend which vendor to choose and explain why based on the documents.
TRY IT YOURSELF
Why This Is Agent-Level
This prompt doesn't ask for a single answer—it asks the AI to reason through multiple scenarios and provide decision frameworks. This is how you'd set up an agent to handle varying situations at scale.