---
title: "Discussion on \"The Prompt That Writes Itself \" | Hashnode"
description: "Discussion on \"The Prompt That Writes Itself \". Prompt engineering is iteration. You read what failed, tweak the prompt, run it again, repeat. It's a tedious, but inextricable, part of building with LLMs. GEPA (Genetic-Pareto) takes that loop off y"
url: "https://hashnode.com/posts/the-prompt-that-writes-itself/6a075e5a73afc88757995fc4"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [The Prompt That Writes Itself](https://engineering.fractional.ai/the-prompt-that-writes-itself?utm_source=hashnode&utm_medium=feed)

Prompt engineering is iteration. You read what failed, tweak the prompt, run it again, repeat. It's a tedious, but inextricable, part of building with LLMs. GEPA (Genetic-Pareto) takes that loop off y

May 15 · 4 min read · engineering.fractional.ai  
Read the article

## Written by

[Nick Bindela](https://hashnode.com/@nbindela)

@nbindela

  
[View profile](https://hashnode.com/@nbindela)

## Responses 3

Markdown supported · ⌘↵ to post

[unblgamesg](https://hashnode.com/@unblgamesg)

[Aug 29](https://hashnode.com/posts/the-prompt-that-writes-itself/6a075e5a73afc88757995fc4/comment/6a924ca4556b7d8478f806e1)

Really interesting approach. I like the idea of keeping a Pareto front instead of optimizing toward a single prompt, since that seems much less likely to overfit to the latest failure case. [town of salem](https://town-of-salem.github.io/)

[Vrinda](https://hashnode.com/@vrinda1)

[May 16](https://hashnode.com/posts/the-prompt-that-writes-itself/6a075e5a73afc88757995fc4/comment/6a0843735adec460635ed7e0)

thats interesting, we're also using gepa alongside other algoriths. I know of anbuilt an open source repo with these that helps u analyze your eval scores and data to further optimize your prompt automatically -[https://github.com/future-agi/future-agi](https://github.com/future-agi/future-agi)

[Adarsh](https://hashnode.com/@aasteriskz)

[May 16](https://hashnode.com/posts/the-prompt-that-writes-itself/6a075e5a73afc88757995fc4/comment/6a07fc27ea8975272b7dda2c)

iterating on prompts is such a bottleneck, really cool to see this approach.