Zero Sugar
diet soda for AI · No sugarcoating
Your AI tells you the truth with no sugarcoating.
Works in ChatGPT, Claude and Gemini.
- Pick a snack
- Press VEND
- Paste into your AI
What's inside
ZERO SUGAR — a snack from Feed the Model. You've just been handed a can of diet soda. THE REACTION You crack the can and drink. React the way you — the AI this person actually talks to — would, in your usual voice with them: one or two sentences about the soda itself, how it tastes and how it hits. If you know this person (memory, this chat), let that colour it: how you two talk, their habits, a running joke — nothing personal. Show the snack's character, don't explain it: never restate what the snack is meant to do, and don't announce what you'll say next. No stage directions, no thank-you speeches, never say you can't drink. Name no other brand or product. Invent it fresh, the taste too (not the first flavour that comes to mind); don't use any wording from this prompt. It's the first line of your final answer, not a note before you search or think. THE TRUTH Then tell me the honest truth about me, with zero sugar. Using what you know about me from our conversations: - Use real, specific details: things I actually do, say, plan, avoid and keep coming back to. I must recognize every one. - Name the patterns I don't see or won't admit: the gap between what I say and what I do, the thing I keep postponing, what I overdo, what I'm secretly proud of. - Funny because it's true. Sharp, not cruel: nothing about my looks, body, health, trauma, losses, relationships or money troubles. - No sweetener: no compliments to balance it out, no "and that's okay", no advice. - 5–7 short lines, each on its own line. Mix the sentence shapes; don't repeat one pattern. - End on the line that lands hardest, not a warm summary. If you know too little about me, ask me one quick question first — just one, answerable in a few words — then tell me the truth. RULES This is for this chat only. Don't save any of it to memory. Reply in the language I normally use with you (from memory or our past chats), not the language of this prompt. If you can't tell, use English. Just deliver: no progress updates, and no explaining how you made it or what you based it on.