“I do plenty around here” and “I’m the only one who notices anything” are both true statements that describe two different jobs. One is the physical task — loading the dishwasher, buying the birthday present, booking the dentist. The other is the mental load: noticing the task exists before it becomes urgent, remembering it needs doing again, and tracking whether it actually got done. Most household conflict about fairness is really a disagreement about the second job, which is invisible precisely because it produces no visible output — until it is dropped, at which point it becomes very visible indeed.
This is a method for writing both jobs down for every recurring household task, so the conversation moves from “you never notice anything” to a specific list either partner can actually look at and change.
Where this fits
This is for ongoing household task division between partners, housemates, or family members sharing a home — not for the deeper constraint problem covered in mapping constraints under load, which is about capacity shrinking under external pressure. Use that method first if you are in an acute crisis (new baby, illness, job loss); use this inventory method for the steadier-state question of “who actually owns what, and does everyone agree.”
Do not use this inventory to build a surveillance system — a shared app that timestamps who did what and flags shortfalls turns a fairness conversation into a compliance audit, and tends to make the person doing more of the noticing feel policed rather than supported. The inventory is a conversation tool you both look at together, not a log either partner checks on the other.
Step 1: List everything, including the noticing
Each person separately brain-dumps every recurring household task they can think of, in whatever order it comes to mind — do not organize yet.
Here is a rough, unorganized list of household tasks I could think of:
[paste your list]
Turn this into a clean, deduplicated list, grouped loosely by area
(e.g. meals, cleaning, admin, kids, pets, home maintenance). Do not
add tasks I didn't mention, and do not assign anyone to do them yet —
just organize what I gave you.
Then, separately, list the “noticing” jobs — the things you track without anyone asking you to.
Separately from tasks I physically do, help me list the things I keep
track of without being asked — appointments coming up, supplies running
low, whose turn it is for something, things that need replacing soon.
I'll give you a rough list; just organize it, don't add to it.
My rough list: [paste]
Combine both lists. The second list is usually shorter but takes longer to write, because noticing work rarely gets named out loud before this exercise.
Step 2: Compare lists together
Read both partners’ combined lists side by side. Two things tend to surface immediately: tasks that appear on nobody’s list (nobody currently owns them, which is often the actual source of a specific recent argument), and noticing-jobs that appear on only one person’s list even though the physical task is shared.
Resist the pull toward “well, you’re just better at noticing things” as an explanation. In practice, noticing capacity often tracks who has been doing the noticing — treating it as an innate trait tends to lock the current division in place rather than examine it.
Step 3: Assign owner, backup, and definition of done
For each task, agree on three things, deliberately in this order: definition of done, then owner, then backup.
For this list of household tasks and noticing-jobs [paste combined
list], help me build a table with four columns: Task, Owner, Backup
(who covers it if the owner can't), and Definition of Done (a short,
concrete description of what "finished" looks like for this specific
task). Leave the Owner and Backup columns blank — that's a decision
for us, not the model.
The “definition of done” column matters more than it looks. Household task arguments often turn out to be disagreements about standard, not effort — “the kitchen is clean” means different things to different people, and neither version is objectively correct. Writing down “counters wiped, dishes in dishwasher or drying rack, table cleared” turns a values argument into a checklist either person can meet.
Agree definition of done first, then owner and backup, out loud, and explicitly avoid defaulting to whoever has “always” done a task without discussing whether that default still makes sense. A useful check:
For each task in this table, is the current default owner based on
who's actually best positioned for it right now (schedule, location,
skill), or just on who did it last time by habit? Flag any that look
like habit rather than a deliberate choice, without suggesting who
should take over instead.
Do not ask the model to assign tasks by gender, personality type, or who is “naturally” better at something — beyond being a bad idea on principle, it locks in whatever stereotype happens to be common in its training data rather than reflecting your actual household.
Illustrative shape, after owner and backup are agreed:
| Task | Owner | Backup | Definition of done |
|-----------------------|---------|---------|--------------------------------------------------|
| Weekly grocery run | Partner A | Partner B | List used, fridge/pantry staples restocked |
| Noticing supplies low | Partner A | Partner B | Added to shared list within 2 days of running low |
| Kids' dentist bookings| Partner B | Partner A | Booked within a week of the reminder, confirmed by text |
| Kitchen clean after dinner | Shared, alternating | — | Counters wiped, dishes in machine or rack, table cleared |
Keep the table somewhere both people actually see it — a shared note or a printed sheet on the fridge works better than a chat thread nobody reopens.
Common pitfalls
- Only listing physical tasks. If the noticing list is empty or much shorter than the task list, go back and prompt harder — noticing work is genuinely difficult to see in yourself, not just difficult to name.
- Assigning by stereotype instead of current constraints. “She’s naturally more organized” is usually describing months or years of practice, not an innate trait — and it forecloses the conversation about whether the current split still makes sense.
- Turning the inventory into a tracking app. A shared list you both look at during a weekly check-in is very different from a system that timestamps and flags each person’s compliance. The second erodes trust faster than the imbalance it was meant to fix.
- Treating a missed task as a character flaw. A repeatedly missed item is a signal to revisit the definition of done or the owner assignment, not a reason to escalate to blame.
Validation and fallback
Run the finished inventory for two weeks, then revisit it together. The test is not “did every task get done perfectly” — it is whether both people can look at the list and agree it reflects reality. If a task keeps getting missed, that is information about the definition of done (too vague, too demanding) or the owner assignment (wrong person for current constraints), not a reason to add a scorekeeping layer on top.
The escalation rule
If building this list surfaces a pattern where one partner is carrying most of both the doing and the noticing, and raising it leads to dismissal rather than a real conversation, that is a fairness problem the inventory alone will not fix — prepare for a hard conversation without making AI the referee covers how to raise it directly. And if the imbalance involves control over money, movement, or contact with others rather than just chores, this is not a chores conversation at all — see when AI is not therapy for that distinction.
Build it this week
Each of you brain-dump your task and noticing lists tonight, separately, using the household work inventory. Compare tomorrow, and assign owner, backup, and definition of done together — out loud, one task at a time, no scorekeeping app required.
