Hi everyone,
I’m working on an early project called Prompt Teacher, and I’ve reached a point where testing it with prompts I invent myself is no longer very useful.
The idea is simple:
People usually know what they want from AI, but their actual request may be short, messy, incomplete, or missing context. Prompt Teacher tries to understand that intention and prepare a clearer request for the AI platform being used.
I’d like to test whether it actually preserves people’s intentions rather than accidentally changing them.
If you’d like to help, please leave a reply with:
1. Any real prompt or request you would normally send to AI.
2. The platform you would use it with — ChatGPT, Claude, Gemini, Perplexity, Copilot, or another one.
Please send the prompt exactly as you would normally write it. Don’t improve it for the experiment.
I’ll run it through the current version of Prompt Teacher and reply here with the result.
Then I’d like you to judge one thing:
Did the resulting request still represent what you actually meant?
A failure is especially useful to me. If it removes something important, adds an assumption you never made, misunderstands the goal, or simply makes the request worse, please tell me.
There is no signup and you don’t need to visit another website. I want to run the experiment directly in this discussion.
I’m particularly interested in difficult, vague, or incomplete requests because those are the cases I can’t evaluate properly by myself.
Thanks to anyone willing to give it a try.