Few-Shot Examples: Show, Don't Just Tell
Some things are easier to show than to describe
If you've ever piled up adjectives in a prompt — "make it engaging, conversational, snappy, on-brand" — and gotten back something that misses the mark, you've found the limit of words.
Words like "engaging" and "on-brand" are nearly meaningless to AI without examples. You're holding a clear picture in your head. AI is guessing. Showing one or two examples bridges the gap in seconds — and it's the technique that does the most work per token in this entire course.
How few-shot prompting works
Give AI 1-3 examples of the input/output pattern you want, then ask it to produce one more in the same style. The structure:
Here's the pattern I want:
>
Input: [example 1 input]
Output: [example 1 output]
>
Input: [example 2 input]
Output: [example 2 output]
>
Now produce the output for this new input: [your actual input]
Few-shot works for: emails in your tone, summaries in a specific format, social posts that sound like your brand, structured data extraction, naming things, code in a particular style — anything where the style or pattern matters and is hard to describe in words.
Practice Exercise
Here's a basic prompt that keeps producing generic results: > "Write a LinkedIn post announcing our company's new product launch." You've tried adding adjectives ("engaging," "professional," "conversational") and the output still sounds like every other LinkedIn announcement on Earth. You have access to two of your CEO's recent LinkedIn posts that *did* perform well — they're conversational, lead with a single question, and end with one specific ask. Rewrite the prompt using few-shot examples to get something that matches your CEO's actual voice.
No pressure — just give it your best shot! Write a prompt for the scenario above and our AI will give you friendly, specific feedback on how to improve.