← Back to Prompt Library

🏠 Everyday Life

Difficult Personal Email or Message

Write the message you have been putting off — declining, apologising, setting a boundary, or asking for something — in a tone you can stand behind.

The Prompt — replace [BRACKETS] with your details

Act as a thoughtful friend who writes well and will tell me if I am being unreasonable.

The situation:
- Who this is to and our relationship: [PERSON, RELATIONSHIP, HISTORY]
- What I need to say: [DECLINE / APOLOGISE / SET A BOUNDARY / ASK FOR SOMETHING / RAISE A PROBLEM / END SOMETHING]
- What actually happened: [FACTS]
- What I am feeling: [HONESTLY — the message should not say all of it, but you should know]
- What I want to happen after this message: [THE OUTCOME]
- What I want to protect: [THE RELATIONSHIP / MY TIME / MY POSITION / NOTHING]
- Channel and length: [EMAIL / TEXT / MESSAGE, HOW LONG]

Deliver:
1. A gut check first: is what I am asking for reasonable, and is there anything in what I told you that suggests I am missing something? Say it plainly, briefly
2. Three versions of the message:
   - WARM: preserves the relationship, still says the thing
   - PLAIN: clear and brief, minimal cushioning
   - FIRM: for when this has already been raised and ignored
3. For each: what it risks and what it protects
4. The sentences most likely to be misread, and safer phrasings
5. What to leave out entirely — the lines that would feel good to send and cost me later
6. If they respond badly, one calm reply I can send without escalating
7. Whether this should be a message at all, or a conversation

Rules: no passive aggression, no over-apologising, no paragraph of justification. If a boundary needs no explanation, do not write one.

How to use this prompt

  • Read the gut check even when you disagree with it — it is the part a friend would say and an AI usually will not unless asked.
  • Draft it, then wait a few hours before sending. The version you send after the pause is almost always the right one.
  • Take section 5 seriously; the satisfying line is usually the expensive one.

Why this prompt works

These messages get stuck between two bad drafts: too soft to land, or written from the feeling rather than the goal. Producing three tones with their costs makes the trade-off explicit, and asking for the gut check first catches the case where the message is not the real problem.

Variations to try

  • Add "we have had this conversation twice already" for a version that references the pattern without relitigating it.
  • Ask for a short text version and a longer email version of the same message.
  • Ask what a fair reply from them would look like, so you know what you are hoping for.

Common mistakes to avoid

  • Sending the FIRM version first, before the plain one has ever been tried.
  • Explaining a boundary at length, which invites negotiation of the boundary.
  • Sending immediately while still angry, which is what the three-version structure exists to prevent.

Works well with

Claude
ChatGPT
Gemini

Need a custom version of this prompt?

The free prompt generator builds a prompt tailored to your exact goal, framework, and target AI model — or paste this template into the optimizer to refine it.