Few-Shotadvanced

Few-Shot Examples

A three-part pattern that shows the LLM 2–4 input-output examples and asks it to match the pattern on a new input. Best for stylistic tasks where the pattern is hard to describe.

The components

  1. 1
    Task framing

    A one-sentence description of the pattern the LLM should learn.

  2. 2
    Examples

    2–4 input → output examples that demonstrate the pattern.

  3. 3
    Live input

    The new input the LLM should transform using the demonstrated pattern.

Worked example

Rough prompt

Help me reply to investor 'no' emails with grace.

Super-prompt produced by Few-Shot

[Task framing] Convert short investor "we'll pass" emails into warm, brief founder replies that keep the door open without sounding desperate.

[Examples]

Example 1
Investor email: "Hi Maya — thanks for the time on Tuesday. We've decided this isn't a fit for the fund right now. We'd love to stay in touch as you progress."
Founder reply: "Hi Sam — really appreciate the clarity. The conversation on Tuesday made me think harder about the wedge motion, so it was time well spent on my end. I'll keep you posted when we hit the next milestone."

Example 2
Investor email: "Maya, thanks again for sharing the deck. After the partner meeting we've concluded the stage is a touch early for us. Happy to revisit at Series A."
Founder reply: "Thanks Helena — totally understand. I'll come back to you when we have meaningful revenue traction. In the meantime, any chance you could intro me to one or two friends who do play at our stage?"

Example 3
Investor email: "Maya — we discussed and unfortunately won't be moving forward. Thanks for letting us look."
Founder reply: "Thanks Jon — appreciate you taking the look. If there were one thing about the deal or the team that gave you pause, I'd really value the honest read. Either way, hope to keep you in the loop."

[Live input]
Investor email: {{user provides this when they paste the rough prompt below — embed it verbatim here}}

Try Few-Shot on your own prompt

Open the tool with Few-Shot pre-selected
Few-Shot Prompt Framework — Teach the LLM by Example · PromptKit