The Guy
take the quiz
Claude can summarize a legal brief, debug your code, and explain quantum mechanics. ask it to text a girl back and it sounds like a LinkedIn recruiter who just discovered emojis.
ask Claude to help you reply to a girl and watch what happens. it doesn't give you a text. it gives you advice.
it gives you a paragraph about showing genuine interest, maintaining a relaxed tone, and being respectful. then — maybe — a suggestion wrapped in caveats.
you didn't ask for a therapy session. you asked what to text her. but Claude can't help itself. it's trained to be helpful, harmless, and honest. the "harmless" part is what kills you.
thanks, Claude. extremely helpful. she's definitely going to text back now that you've reminded me to respect her boundaries with a smiley face.
she said "haha nice" and you need to respond.
Claude gave you a question with an exclamation mark, a filler phrase, and the word "adventures." three red flags in one sentence.
The Guy read the subtext (she's bored, low-investment), matched her energy, and pivoted to a date. "haha nice" means "entertain me or lose me." The Guy understood. Claude didn't.
this isn't Claude's fault. it's structural.
Claude is trained with Constitutional AI — a system that rewards being safe, agreeable, and thorough. every response is filtered through "could this be harmful?" in dating, "safe" and "agreeable" are the two worst things you can be.
The Guy is built on a completely different model with a completely different philosophy — one that rewards the things that actually create attraction.
Claude writes like a polished email. The Guy writes like a text from the guy she can't stop checking her phone for.
paste these prompts into Claude yourself.
four sentences. three questions. one exclamation mark. zero personality. she sent a winky face and got a customer service response.
brackets. the AI literally left brackets in. even if you fill them in, "it was really great meeting you" is what her dentist says after a cleaning.
"makes everything feel a little brighter" — she's reading this at 11pm in bed and physically recoiling. this is a Hallmark card, not a text.
| feature | Claude AI | The Guy |
|---|---|---|
| purpose | general-purpose AI assistant | dating text specialist |
| output | x gives advice about what to say | > gives 3 ranked texts to send |
| subtext analysis | x treats messages at face value | > analyzes her energy and intent |
| tone | proper grammar, exclamation marks | lowercase, flat delivery, brevity |
| coaching | x generic advice disclaimers | > why each option works |
| stage awareness | x none | > opener / mid-convo / close |
| edge | x safety filters block anything edgy | > built for genuine edge |
| safety disclaimers | "Remember, be genuine and respectful..." | none. ever. |
| pricing | free / $20 Pro | $59/month or $149/quarter |
| philosophy | "helpful, harmless, honest" | "text like the version of you that exists when you're not overthinking" |
Claude can't flirt.
stop asking it to.
take the instinct quiz