Challenge 03 · Chain of Thought Prompting
Chain of Thought Solver 🟠
✓
Zero-Shot
✓
Few-Shot
3
CoT
4
Self-Con.
5
ToT
What is Chain of Thought Prompting?
Chain of Thought (CoT) prompting instructs the model to show its reasoning step by step before giving a final answer. Instead of jumping to a conclusion, the model walks through intermediate steps — making complex reasoning tasks far more accurate and transparent. The key phrase: "Let's think step by step."
Scenario
A logistics company needs an AI that doesn't just give answers — it shows its working.
When planning delivery routes, cost calculations, or resource allocation,
the reasoning must be transparent and auditable.
Your Task
Build a PartyRock app called "The Logistics Thinker" that:
- Takes a logistics problem as input
- Works through the problem step by step, with each reasoning step clearly numbered
- Arrives at a final recommendation only after all steps are shown
Rules
- Your system prompt must explicitly instruct the model to think step by step
- Each reasoning step must be numbered in the output
- The final answer must appear only after all reasoning steps
- Must work correctly on both test problems below
Test Problems (Must Test Both)
Problem A — Delivery Route
A driver has 4 deliveries:
Stop A: 5 km, 10 kg
Stop B: 12 km, 3 kg
Stop C: 8 km, 7 kg
Stop D: 3 km, 15 kg
The van holds max 20 kg per trip.
What is the most efficient 2-trip plan to minimize total distance?
Problem B — Staff Scheduling
A warehouse has 3 staff.
Task X: requires 2 people, takes 3 hours
Task Y: requires 1 person, takes 5 hours
Task Z: requires 2 people, takes 2 hours
All tasks must be done today in 8 working hours.
Can they do it? If yes, what is the schedule?
⭐
Bonus +10 points: Add a second AI widget that solves the same problem
without step-by-step reasoning, so the quality difference is visible side by side.
How to Build This on PartyRock — Step by Step
STEP 1 — Create the app
1. Go to partyrock.aws → "Build your own app"
2. Describe it as: "A logistics problem solver that shows step-by-step reasoning"
3. Clear auto-generated widgets and build manually
STEP 2 — Add widgets
1. "+ Add widget" → "User Input" → label: "Enter your logistics problem"
2. "+ Add widget" → "AI Generation" (your CoT solver)
STEP 3 — Write your CoT system prompt
In the AI Generation widget → System Prompt field:
• Explicitly instruct the model to think step by step
• Require numbered reasoning steps before the final answer
• Specify that the final answer appears ONLY after all steps
Key phrases to include: "think step by step", "show your reasoning",
"number each step", "final answer at the end"
Connect User Prompt to your input widget (@input)
STEP 4 — Set Model, Temperature & Top P
Reasoning tasks need a capable model and low temperature for consistency.
• Model: Claude 3.5 Sonnet handles multi-step logic well
• Temperature: lower values (0.0–0.3) keep reasoning consistent
• Top P: consider how it affects logical coherence
⚠️ Auto-read from JSON. Make deliberate choices.
STEP 5 — Test & Export
Test with both logistics problems. Verify numbered steps appear before the answer.
Mac: Cmd+K | Windows: Ctrl+K → exports JSON
Copy share URL and submit both below.
💡 Your model, temperature, and Top P are extracted from the JSON automatically.
Scoring Breakdown (150 pts + 10 bonus)
| Criterion | Points |
|---|---|
| Correct CoT technique (explicit step-by-step in prompt) | 25 |
| Steps are numbered and logically sequential in output | 25 |
| Correct answers on both test problems | 20 |
| Real-world relevance of use case | 15 |
| Model choice + written justification | 25 |
| Temperature value + written justification | 20 |
| Top P value + written justification | 20 |
| Bonus: side-by-side comparison widget | +10 |
| Total | 150 (+10) |
Submit Your Entry