Few-Shot Prompting: Teach by Example

Instead of describing what you want, show the AI two or three examples. Few-shot prompting is often the fastest way to get consistent output.

26 min · Reviewed 2026

Show, don't just tell

Zero-shot means you ask with no examples — you just describe what you want. Few-shot means you give the AI a few example pairs so it can pattern-match. Few-shot dramatically improves consistency, especially for formatting and tone.

Zero-shot	Few-shot
Describe the task in words only.	Give 2-5 example input/output pairs first.
Fast to write.	Takes more setup time.
Works for common tasks.	Works for weird, specific, or style-sensitive tasks.
Variable results.	Much more consistent results.

A classic few-shot prompt

Convert these product reviews into a one-word sentiment tag. Use only: positive, negative, neutral. Review: "The battery died after three hours. Very disappointed." Sentiment: negative Review: "Works fine, nothing special, does what it says." Sentiment: neutral Review: "Best headphones I've ever owned — crystal clear sound." Sentiment: positive Review: "Package arrived late but the product itself is amazing." Sentiment:Three labeled examples teach the AI the pattern. The fourth is left unfinished for it to complete.

The AI will complete the fourth one with 'positive' — it picks up on the pattern: focus on the product, not the delivery. You didn't have to explain that rule in words. The examples carried the rule.

When few-shot wins

Custom formatting that's hard to describe (e.g., a special kind of outline).
Tone of voice — examples nail the vibe faster than adjectives.
Tricky classification where the rules have exceptions.
Translating between structured formats.
Anything where you've thought 'I know it when I see it.'

Example: writing in a specific style

Rewrite historical facts as if they were sports commentary. Two examples: Fact: George Washington crossed the Delaware on December 25, 1776. Commentary: "And there he goes — Washington making the gutsiest Christmas call of his career, charging through the icy Delaware with nothing but stones in his pockets and revolution on his mind!" Fact: Neil Armstrong walked on the moon on July 20, 1969. Commentary: "One small step for Armstrong, one giant leap for America's space program — what a clutch moment on the lunar stage tonight!" Fact: The Wright Brothers' first flight lasted 12 seconds on December 17, 1903. Commentary:Style transfer via two clear examples.

End-of-lesson check

8 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-prompting-few-shot-builders

What is the main idea of "Few-Shot Prompting: Teach by Example"?
1. Instead of describing what you want, show the AI two or three examples. Few-shot prompting is often the fastest way to get consistent output.
2. Use AI as the final authority for the whole decision
3. Avoid checking the answer once it sounds polished
4. Focus only on speed instead of judgment
Which concept is most central to "Few-Shot Prompting: Teach by Example"?
1. zero-shot
2. few-shot prompting
3. pattern matching
4. in-context learning
Which use of AI fits this topic best?
1. Let the AI decide what matters without your review
2. Use the answer before checking whether it fits the situation
3. Custom formatting that's hard to describe (e.g., a special kind of outline).
4. Use the first answer without checking it
What should a careful learner remember about "Balance matters"?
1. Use AI to draft or organize ideas about few-shot prompting, then verify before acting.
2. Skip the context so the tool can guess faster
3. Treat the output as private even after sharing it online
4. Use the answer without checking the source
You want to use AI after this lesson. What is the safest next step?
1. Act immediately because the AI answer is written clearly
2. Use the AI answer as a draft, then check it against a reliable source.
3. Hide uncertainty so the final answer looks cleaner
4. Use private or sensitive details before checking permission
How should AI output about few-shot prompting be treated?
1. As proof that no other source is needed
2. As a replacement for context, consent, or expert review
3. As a draft or helper output that still needs human judgment and verification
4. As something that becomes correct when it sounds confident
Name one way to verify an AI answer about few-shot prompting.
Which action would help you apply "Few-Shot Prompting: Teach by Example" responsibly?
1. Use the tool to avoid thinking through the tradeoff
2. Keep going even if the output conflicts with a trusted source
3. Use the first answer without checking it
4. Tone of voice — examples nail the vibe faster than adjectives.

← Back to interactive lesson

Tendril · Builders · Prompting

Few-Shot Prompting: Teach by Example

Instead of describing what you want, show the AI two or three examples. Few-shot prompting is often the fastest way to get consistent output.

26 min · Reviewed 2026

Show, don't just tell

Zero-shot	Few-shot
Describe the task in words only.	Give 2-5 example input/output pairs first.
Fast to write.	Takes more setup time.
Works for common tasks.	Works for weird, specific, or style-sensitive tasks.
Variable results.	Much more consistent results.

A classic few-shot prompt

Convert these product reviews into a one-word sentiment tag. Use only: positive, negative, neutral. Review: "The battery died after three hours. Very disappointed." Sentiment: negative Review: "Works fine, nothing special, does what it says." Sentiment: neutral Review: "Best headphones I've ever owned — crystal clear sound." Sentiment: positive Review: "Package arrived late but the product itself is amazing." Sentiment:Three labeled examples teach the AI the pattern. The fourth is left unfinished for it to complete.

When few-shot wins

Custom formatting that's hard to describe (e.g., a special kind of outline).
Tone of voice — examples nail the vibe faster than adjectives.
Tricky classification where the rules have exceptions.
Translating between structured formats.
Anything where you've thought 'I know it when I see it.'

Example: writing in a specific style

Rewrite historical facts as if they were sports commentary. Two examples: Fact: George Washington crossed the Delaware on December 25, 1776. Commentary: "And there he goes — Washington making the gutsiest Christmas call of his career, charging through the icy Delaware with nothing but stones in his pockets and revolution on his mind!" Fact: Neil Armstrong walked on the moon on July 20, 1969. Commentary: "One small step for Armstrong, one giant leap for America's space program — what a clutch moment on the lunar stage tonight!" Fact: The Wright Brothers' first flight lasted 12 seconds on December 17, 1903. Commentary:Style transfer via two clear examples.

End-of-lesson check

8 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-prompting-few-shot-builders

What is the main idea of "Few-Shot Prompting: Teach by Example"?
1. Instead of describing what you want, show the AI two or three examples. Few-shot prompting is often the fastest way to get consistent output.
2. Use AI as the final authority for the whole decision
3. Avoid checking the answer once it sounds polished
4. Focus only on speed instead of judgment
Which concept is most central to "Few-Shot Prompting: Teach by Example"?
1. zero-shot
2. few-shot prompting
3. pattern matching
4. in-context learning
Which use of AI fits this topic best?
1. Let the AI decide what matters without your review
2. Use the answer before checking whether it fits the situation
3. Custom formatting that's hard to describe (e.g., a special kind of outline).
4. Use the first answer without checking it
What should a careful learner remember about "Balance matters"?
1. Use AI to draft or organize ideas about few-shot prompting, then verify before acting.
2. Skip the context so the tool can guess faster
3. Treat the output as private even after sharing it online
4. Use the answer without checking the source
You want to use AI after this lesson. What is the safest next step?
1. Act immediately because the AI answer is written clearly
2. Use the AI answer as a draft, then check it against a reliable source.
3. Hide uncertainty so the final answer looks cleaner
4. Use private or sensitive details before checking permission
How should AI output about few-shot prompting be treated?
1. As proof that no other source is needed
2. As a replacement for context, consent, or expert review
3. As a draft or helper output that still needs human judgment and verification
4. As something that becomes correct when it sounds confident
Name one way to verify an AI answer about few-shot prompting.
Which action would help you apply "Few-Shot Prompting: Teach by Example" responsibly?
1. Use the tool to avoid thinking through the tradeoff
2. Keep going even if the output conflicts with a trusted source
3. Use the first answer without checking it
4. Tone of voice — examples nail the vibe faster than adjectives.

← Back to interactive lesson