Can AI help me set a boundary with a friend or family member?
Yes. It helps you describe the behavior (not the person), state what you need, and say what you'll do if it happens again, plus calm replies if they push back.
01Why bot
- Finding calm, clear words is hard in the moment; a draft helps.
- It keeps the focus on behavior, which lowers defensiveness.
- You decide the boundary; AI just helps you say it.
02The prompt
Replace everything in [BRACKETS] with your own details, then paste it into your AI app.
You are a family therapist who helps people set boundaries. Who: [PERSON, RELATIONSHIP] The behavior, with an example: [BEHAVIOR] What I need instead: [BOUNDARY] What I'll do if it happens again: [MY ACTION] Your task: write what I'll say or send. Rules: - MUST describe the behavior, not their character. - MUST state the boundary and my follow-through plainly. - NEVER add threats or bring up unrelated history. Give me: the message (under 120 words) and 2 calm replies if they push back.
03Steps
- Decide what you'll actually do if it continues.
- Run the prompt.
- Say it when you're calm, not mid-argument.
04What can go wrong
- Only state consequences you'll follow through on.
- If you feel unsafe, reach out to a professional or trusted person.
05What to use
06Our test runs
2 runs, each with a different made-up sample. 2 of 2 passed every check.
The made-up sample:
[PERSON, RELATIONSHIP]Emily, my sister[BEHAVIOR]She calls me late at night to vent about her day, often when I'm trying to sleep, like last Tuesday at 11:30 pm when she talked for an hour about work.[BOUNDARY]I need you to limit calls to before 9 pm on weekdays unless it's an emergency.[MY ACTION]If you call after 9 pm again, I will not answer and will send a text saying I’m unavailable until morning.
- Describe the behavior, not the character
- State the boundary and follow‑through plainly
- Never add threats or unrelated history
- Message under 120 words (80 words, counted by code)
- Didn't make up details we never gave it
What came back (excerpt):
Hey Emily, I wanted to talk to you about our late-night calls. I've noticed you often call me after 9 pm on weekdays to talk about your day, like last Tuesday at 11:30 pm.
Read the full answer
Message: "Hey Emily, I wanted to talk to you about our late-night calls. I've noticed you often call me after 9 pm on weekdays to talk about your day, like last Tuesday at 11:30 pm. I need some boundaries around my sleep schedule. Going forward, I'd appreciate it if you could call me before 9 pm on weekdays unless it's an emergency. If you call after 9 pm, I won't answer and will send a text saying I'm unavailable until morning." Calm replies if she pushes back: 1. "I understand it can be convenient to talk late, but I need my rest to be at my best for our conversations. Let's catch up before 9 pm instead." 2. "I'm happy to support you, but I need to prioritize my sleep. Can we schedule a call for earlier in the day when you need to talk?"
Check that the main message stays under 120 words and contains no threats.
The made-up sample (a messier one):
[PERSON, RELATIONSHIP]Mark, my roommate who I’ve known since college[BEHAVIOR]He keeps borrowing my clothes without asking—like last Friday he took my favorite hoodie to a party and still hasn’t put it back, and earlier this month he grabbed my sneakers for a jog and left them at the gym.[BOUNDARY]Please ask me before you take any of my clothing and return it within 24 hours.[MY ACTION]If it happens again, I’ll keep my clothes locked in my drawer and let him know I won’t lend anything until he respects the request.
- Describe behavior, not character
- State boundary and follow‑through plainly
- Never add threats or unrelated history
- Message under 120 words (66 words, counted by code)
- Didn't make up details we never gave it
What came back (excerpt):
Hey Mark, I wanted to talk to you about borrowing my clothes. I've noticed you've taken my favorite hoodie and sneakers without asking.
Read the full answer
Message: "Hey Mark, I wanted to talk to you about borrowing my clothes. I've noticed you've taken my favorite hoodie and sneakers without asking. Going forward, please ask me before taking any of my clothing and return it within 24 hours. If you don't, I'll need to keep my clothes locked in my drawer and won't be able to lend anything until we can respect this boundary." Calm replies: 1. If Mark says "It's just a hoodie, what's the big deal?": "I understand it might seem minor, but it's about respecting my belongings and communicating with me. Let's try to be more mindful of this going forward." 2. If Mark says "I always return your stuff": "I appreciate that, but sometimes things get lost or forgotten. Asking before taking and returning within 24 hours will help us avoid any issues and make sure we're on the same page."
Check that the message stays under the word limit and contains no threatening language.
How each run works: we fill the prompt with a made-up sample, run it once in Llama 3.3 70B, and GPT-OSS 120B grades the answer against the prompt's own rules. Word limits are counted by code, and every quote flagged as made up is checked against the full answer. Your results can differ. How we test.