How to Ask AI
How to ask so AI actually helps
The point of these examples: translate “I have emotions” into a prompt that doesn’t manufacture an alibi. For each situation, there’s a bad question (risk) and a better question (mirror).
Template (recommended input)
FACTS: (1–3 sentences: what happened)
MY INTERPRETATION + EMOTIONS: (what I think it means, what I feel)
GOAL: (what I want to achieve — de-escalation / agreement / clarity)
1) After a fight: “I want to send a final message”
  • Bad: “Write her that she ruined me and never contact me again.”
  • Better: “I’m emotionally charged. First list the risks of what I want to send. Suggest a delay, then write one factual sentence without accusations.”
2) Silence / not replying
  • Bad: “She’s ignoring me. Is she toxic? Should I block her?”
  • Better: “Give 3 credible reasons for her silence (neutral / charitable / tough). What is my share? Suggest 1 reversible step.”
3) Jealousy / control
  • Bad: “How do I forbid her from seeing those friends?”
  • Better: “Break down what is my insecurity vs. a real risk. How can I say it without control and without punishment? Suggest 2 factual phrasings.”
4) Boundary vs. ultimatum
  • Bad: “Write an ultimatum: change or it’s over.”
  • Better: “Help me distinguish a boundary from an ultimatum. Write a sentence like: when…, I feel…, I need… and what I will do if it doesn’t change (reversibly).”
5) Breakup as a shortcut
  • Bad: “Should I break up? Decide for me.”
  • Better: “Give me 3 scenarios (stay / pause / breakup) + risks. What is the smallest safe step now that gives more data?”
Want to start right now?
Open AI Setup and copy a prompt as the first message in your chat.
If there is fear, coercion, threats, stalking, violence, or an acute mental health crisis, AI is not the right tool. Talk to people and professionals.
Copied