Few-shot prompting means showing the model two or three examples of what good looks like instead of describing it. Examples beat adjectives every time: "punchy" means a hundred different things, but a real subject line you loved means exactly one. It is the fastest way to raise output quality without writing a longer brief.
What is few-shot prompting?
The Briefing Method from Lesson 2 tells the model what to make. Few-shot prompting shows it. You paste two or three examples of work you consider excellent, label them as examples, and ask the model to match their pattern. The model stops guessing at your standards because your standards are sitting right there in the prompt.
The name comes from machine learning research, where “shots” are examples. Zero-shot is a bare instruction. Few-shot is an instruction plus samples. You do not need the jargon to use the technique, but you will see the term everywhere, so now you know what it means.
Unlock the rest of this track
Free. Join the newsletter and all 7 lessons open instantly. Unsubscribe anytime.
Frequently asked questions
- 01What is few-shot prompting?
- A technique where you include two or three examples of high-quality output in your prompt and ask the AI to match their pattern. The model imitates the structure, length, and voice of the examples instead of guessing at vague style descriptions.
- 02How many examples should I give AI?
- Two or three. One example reads as coincidence rather than pattern, and large sets dilute each other. Three strong, recent, format-matched examples outperform ten mixed ones almost every time.
- 03Do examples work better than instructions?
- For style and format, yes. Instructions like "make it punchy" are ambiguous, while an example carries length, rhythm, and tone the model can copy directly. Use instructions for the task and guardrails, examples for the voice.