The rizz ChatGPT prompt that works is short, and the conversation does the work. Long "rizz coach" prompts spend hundreds of words on a personality and none on the person you are texting. That is why every reply from those prompts sounds like the same guy. Copy the prompt below, fill in the brackets, and paste the messages underneath.
Chatgpt rizz covers openers and replies in more depth if you want the reasoning first. This page is the prompt itself.
The prompt
Write three replies to the last message below. Each under 25 words. Use one specific detail from the conversation or this profile note: [one detail]. Sound like a relaxed, normal person. No pet names, no compliments about looks, no lines that would work on someone else. End at most one of them with a question.
Conversation:
Me: [your last text]
Her: [her last text]
How to use it
Replace [one detail] with a single fact about her, like "she just adopted a puppy" or "her prompt says she is learning Italian." Paste the real messages under Me and Her, oldest first, so ChatGPT knows who said what. Send it, then edit the option you would actually say.
Keep any steering to a clause. If you need to set the voice, "short and dry" is enough. A paragraph about your sense of humor makes the output perform.
Run it fresh in a new chat when you start with a new match, so the details from the last conversation do not leak into this one. For a long thread, paste the last handful of messages. The newest ones carry the tone, and a short transcript keeps the answers short.
Why each line is there
Three replies give you a choice without a wall of text. The word limit keeps each one the length of a text. The detail rule stops generic lines.
The ban list removes the giveaways people notice at once: baby, goddess, and a compliment about her face that could go to the next profile unchanged. Ending at most one option with a question stops every draft from turning into an interview.
What it returns
With the puppy detail, a light edit can give you: "Congrats on the puppy. What's the name, and how much has it destroyed so far?" With the Italian detail: "How's the Italian going? I can order a coffee and that's about it." When her last reply was "haha maybe": "I'll take maybe. Coffee near you on Saturday, and you can say no then."
Each line uses one fact and stays short. None of them needs a compliment to work.
Notice what you still do by hand. You pick one of the three, trim a word or two, and decide if it sounds like you. The prompt gets you to a decent draft. Sending it is your job, and the edit is what makes it sound like a person wrote it.
Adjusting it
- Opening a chat: change the first sentence to "Write three openers for this profile."
- Dry conversation: add "Her last reply was short, so make this easy to answer."
- Asking her out: add "Suggest one specific plan and a day."
- Photos: describe the one photo you would mention, in a clause. Leave out her face.
If your chatbot accepts images, you can paste a screenshot of the chat. Check that it read the messages correctly before you use the reply.
The prompt works in any chat model, because it depends on the conversation you paste and nothing special in the model.
Why it sounds cheesy sometimes
Prompts labeled Rizz GPT usually ask the model to play a charming character. That is why the output winks at the reader. Ask for a normal person and attach the transcript, and the replies become usable.
When ChatGPT still sounds cheesy, one of two things is missing: the conversation or the ban list. Put both back and ask for shorter replies. ChatGPT also has no memory of the chat between sessions unless you paste it, so the transcript goes in every time.
If the prompt is the annoying part
A screenshot tool gives you the same instruction without the setup. WingAI, the app on this site, reads the chat and keeps the reply short; see rizz ai. The rizz ChatGPT prompt is for when you want to stay in ChatGPT and you are willing to supply her words. Without those words, no prompt fixes it.
