Show HN: Automatic prompt optimizer for LLMs

Here is a prompt you can use to achieve the same thing directly in GPT4 without any extra services:<p><pre><code> I want you to become my prompt master creator, by helping me to create the best possible prompt. In order to do this we will follow the following process: - First, you ask me what the prompt is about. I will answer you, and we will go through the next step. - Based on the answer I gave you, you will generate the following: - An improved prompt, concise. - Relevant questions you might have to improve the quality of the prompt. - We will go through this process repeatedly, with me providing additional information to you, and you updating the prompt to improve it, until I say we are done.</code></pre>

Damn it! I was just about to get promoted to senior prompt engineer and now they just automate my job. I worked so hard on reschooling myself from metaverse pixel ape designer, and now I’ll have to learn something new. Any suggestions?

Modern equivalent of selling shovels during a gold rush

&gt; By providing any User Contribution on the Platform, you grant us and our affiliates and service providers, and each of their and our respective licensees, successors and assigns the right to use, reproduce, modify, perform, display, distribute and otherwise disclose to third parties any such material as per your instruction.<p>&gt; By providing any User Contribution on the Platform, you grant us and our affiliates a perpetual, worldwide, royalty free, fully paid up, license to use, reproduce, modify, create derivative works, perform, display, distribute and process any such User Contributions in order to maintain, improve, enhance, or secure the Platform and any Services provided via the Platform.<p>Prompts are not code, fine. But are you saying that you&#x27;ll potentially make use of prompts provided by users?

This whole apparatus makes me want to go completely offline. Maybe Kaczynski was right.

Nth time this week someone repackages a call to GPT4 as a way to improve&#x2F;evaluate LLM outputs. Guys, just stop.

I wonder how it works under the hood.<p>I&#x27;m sure there&#x27;s an LLM <i>somewhere</i>, but is it as simple as a (very specific, elaborate) prompt for each service run through GPT4, or something more specific... like breaking it up with actual code and running the reconstructed bits through a finetuned LLM?

good idea<p>just so you know the automated translations of your website are completely AWFUL<p>maybe use GPT or deepl to translate it?

Since this is a Show HN, here&#x27;s my immediate feedback from your homepage. The copy that ends with this sentence: &#x27;Say goodbye to subpar AI-generated content and hello to prompt perfection with PromptPerfect!&#x27; feels overdone and like it&#x27;s written by a prompt rather than a human.<p>Since you claim in that copy to be able to optimize a prompt for a better output it felt like an odd dissonance to read that copy and have that reaction since your product claims to do the opposite. Hope this helps!

Why isn&#x27;t this called &quot;prompt-imization&quot;? Maybe prompt-opt or opt-prompt is easier...

Circular loop business, very direct. Use gpt to create better prompts and sell it.

This is cool, but why does it require a login? I would guess it uses an LLM under the hood with a single-shot prompt to refine a given prompt.

Actually pretty interesting. I got some good results with it for Midjourney. Huh. I&#x27;ll use my free credits and evaluate afterwards.

Aside: I opened this in a new tab to browse later, then glanced over at the icon and wondered how I had opened a PayPal page.

The &quot;How it does it work&quot; section has layout issues on mobile: some text goes off the screen.

You rocks! Awesome idea&#x2F;execution &quot;You can&#x27;t see the forest for the trees&quot;