Turn messy catch reports into clean JSON with an LLM

Fishers send us voice-to-text reports like "about 40 kilo tuna, 2 boats, back by 6". We need a prompt + small script that turns 30 sample reports into structured rows, and tells us honestly when it is unsure.

Your task

Find where the AI in the story goes wrong, and show it.

What to hand in

✓ Script + prompt
✓ Output for the 30 samples
✓ Where it fails, in a short list

What the company looks at

☐ Finds real failures
☐ Tests are varied
☐ Findings are clear and ranked
☐ Suggests what good looks like

PromptingPythonData cleaning