The Scenario
A South African retail bank receives thousands of app reviews a day on the Google Play Store. The product team wants to use an LLM (like GPT-4 or Gemini) to classify these reviews into exactly three buckets: "Bug Report", "Feature Request", or "Praise". If a review is just complaining without actionable info, it should be marked "Spam".
The Brief
Write a single prompt that will reliably classify any given review. You must use "few-shot prompting" (providing examples within the prompt) to show the model exactly how to handle edge cases.
Deliverables
- The System Instructions (defining the AI's role and the categories)
- 3 to 5 Few-Shot Examples (Input: [Review], Output: [Category]) including at least one tricky edge case
- The exact prompt template where the new review will be injected (e.g., "Review to classify: {{review}}")
Submission Guidance
Zero-shot prompting (just giving instructions) often fails on edge cases. Few-shot prompting (giving examples) anchors the model. Make sure your examples cover South African slang (e.g., "The app is eish, it keeps crashing" -> Bug Report).
Submit Your Work
Your submission is graded against the rubric on the right. If you pass, you get a public Badge URL you can share on LinkedIn. There is no draft save, so work offline first and paste your finished response here.