Product · 25 tokens per run
Roommate Boundary Message
You share a lease with a roommate who keeps eating your food, whose partner has effectively moved in five nights a week, who never does the dishes, or who is treating the shared unit like a solo one. Paste the category, what actually happened, how many times it has repeated, your lease dynamic, how bad it has gotten, and the medium you want to use, and get back a four-part boundary pack: a firm-but-non-hostile OPENER calibrated to text vs. in-person vs. house-meeting (NVC four-part frame with four condescension phrases stripped), ONE CONCRETE ASK phrased as a specific behavior with a bounded time or count, a DE-ESCALATION COMEBACK LINE for when the roommate gets defensive, and (only when the lease dynamic is you-are-lease-holder or you-are-subletter AND severity is on the higher half) an OPTIONAL PAPER-TRAIL EMAIL that reference-lines the conversation for the record. A coach-not-advocate safety router runs FIRST and refuses to draft anything when the inputs describe intimate-partner or roommate-partner violence (routes to National DV Hotline 1-800-799-7233), hidden cameras or recording devices in shared space (routes to non-emergency police), sexual harassment or sexual coercion (routes to RAINN 1-800-656-4673), physical intimidation or property damage (routes to non-emergency police plus Legal Aid), or landlord retaliation after a complaint (routes to Legal Aid via LawHelp.org). Every non-refuse output appends a same-day-help footer pointing at community mediation (nafcm.org / 211) so the user has a real path when the conversation goes sideways.
About this tool
What does roommate boundary message do?
You share a lease with a roommate who keeps eating your food, whose partner has effectively moved in five nights a week, who never does the dishes, or who is treating the shared unit like a solo one. Paste the category, what actually happened, how many times it has repeated, your lease dynamic, how bad it has gotten, and the medium you want to use, and get back a four-part boundary pack: a firm-but-non-hostile OPENER calibrated to text vs. in-person vs. house-meeting (NVC four-part frame with four condescension phrases stripped), ONE CONCRETE ASK phrased as a specific behavior with a bounded time or count, a DE-ESCALATION COMEBACK LINE for when the roommate gets defensive, and (only when the lease dynamic is you-are-lease-holder or you-are-subletter AND severity is on the higher half) an OPTIONAL PAPER-TRAIL EMAIL that reference-lines the conversation for the record. A coach-not-advocate safety router runs FIRST and refuses to draft anything when the inputs describe intimate-partner or roommate-partner violence (routes to National DV Hotline 1-800-799-7233), hidden cameras or recording devices in shared space (routes to non-emergency police), sexual harassment or sexual coercion (routes to RAINN 1-800-656-4673), physical intimidation or property damage (routes to non-emergency police plus Legal Aid), or landlord retaliation after a complaint (routes to Legal Aid via LawHelp.org). Every non-refuse output appends a same-day-help footer pointing at community mediation (nafcm.org / 211) so the user has a real path when the conversation goes sideways.
How do I talk to my roommate about eating my food?
Name the specific thing that got eaten from what you actually saw, not a generalization about their character, and pair it with one bounded ask: label a shelf that is yours, and everything on your shelf needs to be asked about before it gets touched. Keep it to two or three sentences if you are texting, or a spoken paragraph if you are doing it face to face. Skip the four phrases that make a roommate reply land as passive-aggressive ('just so you know,' 'I would think,' 'for future reference,' 'kindly note that'). Roommate Boundary Message routes food-theft into a labeled-shelf ask because that is the fix that survives repeat conversations rather than a one-off promise to stop.
How do I ask my roommate to stop having their partner over so much?
Frame it as headcount and rent, not relationship policing. 'I pay to live with one other person, not two' is a defensible line; 'I don't like your partner' or 'we need to talk about your relationship' is not. Pair the opener with a specific number: two nights a week, three nights a week, whatever you can live with. The tool picks the number from what you paste and phrases the ask as a specific behavior with a bounded count, so the conversation is about occupancy math and shared utilities rather than about the partner as a person.
What do I say when my roommate never does the dishes?
Use a bounded time window in the ask, not a wish. 'Sink cleared within 12 hours of use' is a rule you can point at next week if it happens again; 'please clean more' is not. Name the specific incident you actually saw ('the sink piled up for three days last week'), not a generalization about their character. If it is the second or third time you are raising it, the message shifts from a first-raise ('hey, can we set a rule together') to an escalating conversation that names the pattern and asks whether a rotation or a house-meeting would help.
Is a text or in-person better for a roommate boundary conversation?
Text works for a first raise where the ask is quick and specific, like 'hey, can we agree the sink gets cleared within 12 hours of use?' In-person works for anything with emotional weight, or where a text would read as passive. A house-meeting agenda works when you have more than one issue, or when three or four roommates need to set rules together. Pick the medium up front on the form; the tool shapes the artifact accordingly, with SMS cadence for text, a spoken paragraph and a pause cue for in-person, or a bulleted agenda plus proposed shared-space rules for the meeting.
How do I document a roommate conversation when I am the lease holder?
Send a short follow-up email the same day, three paragraphs. First paragraph references the conversation by date and category ('following up on our conversation on Tuesday about the shared kitchen'). Second paragraph quotes the concrete ask verbatim, in quotation marks. Third paragraph names the purpose plainly: you want it in writing so both of you can refer back. Never frame the email as a demand, never set a deadline, never threaten eviction or removal. If you are on the lease and the situation is already escalating, the tool generates this email for you alongside the opener, so the record exists before you need it.
When should I not send a boundary message at all?
If the situation involves your roommate or their partner hurting you, threatening you, damaging your things, or entering your room without permission, a boundary text is the wrong tool. Same if you find a hidden camera or recording device in shared space, if the roommate is reading your messages or accessing your accounts, or if the landlord has retaliated (utility shutoff, lock change, non-renewal within six months of a complaint). The tool runs a safety router first and refuses to draft anything in those situations, pointing you at the National DV Hotline (1-800-799-7233), RAINN (1-800-656-4673), non-emergency police, or Legal Aid via LawHelp.org instead.
How is this different from asking ChatGPT to write a message to my roommate?
A freeform ChatGPT prompt writes one paragraph in its default voice, drifts into moralizing about the roommate, and skips the safety check that keeps you out of a situation a text cannot solve. Roommate Boundary Message runs a single structured pass that classifies your situation into one of nine archetypes, produces a four-part pack in one shot (opener, one concrete ask, de-escalation comeback line for the pushback coming, and the paper-trail email when your lease position calls for it), strips four condescension tells that make roommate messages backfire, and refuses outright when the inputs describe violence, coercion, or landlord retaliation.
What does it cost?
25 tokens per run. Produde tokens are a prepaid platform credit; buy them once and spend on any tool.
Do I pay if a run doesn't complete?
No. You only pay for results. If a run can't finish, its full token cost stays in your wallet automatically.
Is my data stored?
Your input and the run output are saved to your account so you can browse history. They are not sold, shared with advertisers, or used to train shared models. See privacy for details.