How it works
The mechanical answer to how the household intelligence layer runs your house. The approval ladder + gradual-trust mechanic, three layers of memory, a nightly synthesis loop, and eight days from application to a held week. The same answer in the same order every time, because the question every household asks is some variant of okay, but how does this actually work on a Tuesday.
TL;DR
Full auto by default — Aule closes loops end-to-end. Two beats of confirmation: before any grocery order, and any single spend at or above $200. The layer gradually loosens as it earns your trust. Three layers of memory underneath; a nightly synthesis loop on top. Eight days from "apply" to "the week is held."
How decisions get made
Auto is the default rung — Aule closes the loop end-to-end without the household in the middle. Two gates are kept on by default: any grocery order, and any single spend at or above $200. These are the only two beats of confirmation a household ever needs from the start; the layer gradually loosens both as it earns your trust.
Rung 01
Default. Closes the loop end-to-end without you in the middle. Receipts land on the third memory layer and surface in the next morning brief.
Rung 02
Any grocery order, any size, asks first. After ~30 days and at least four successful confirmations, the household can graduate to "approve the weekly meal plan, run orders against it" — the cart still drafts every week; you just stop signing each one.
Rung 03
Any single spend at or above $200 surfaces with the quote, the context, and the recommendation. The household can raise the threshold, lower it, or remove it once the layer has earned the latitude.
The gradual-trust principle
The two gates are a starting position, not a ceiling. Households can lock anything tighter at any moment, and they can accelerate the loosening at any moment. The point is full auto with two beats of confirmation today, and full auto with fewer beats as the layer earns them.
Full policy — including escalations, child-identity rules, and the human-in-the-loop pattern — lives on /products/trust-and-safety.
The shape of the work
Underneath every closed loop, the work is the same five operating motions applied across every domain. The long version of these lives on /products; this page is the mechanical view.
01
The layer notices what is coming before the household has to remember. School registrations open, car-service intervals approach, a child’s birthday lands in six weeks — Aule is on the page before the calendar event triggers.
Example: Pediatric well-check window opens; Aule surfaces three suitable Saturday slots with the right copay column.
02
When work has to be done, Aule decides who or what should do it. Vendor short-list, sitter rotation, the right partner for the school email, the AI itself if the job is fully scoped — the right routing is half the work.
Example: "Kitchen-sink leak" routes to three plumbers, one recommended in context — used twice before, mid-quote, earliest window.
03
You set the rules at onboarding. Aule decides inside them. New vendors above a spend threshold ask first; recurring workflows the household has approved once execute automatically; anything touching a child’s identity escalates, always.
Example: A $87 grocery delivery within the weekly budget executes; a $340 outside-vendor invoice surfaces for your approval.
04
Most household to-dos linger because nobody circled back. Aule follows up until the loop is closed — the plumber came, the appointment was kept, the gift arrived, the form was submitted, the refund posted.
Example: Day-after-visit message to the plumber confirms the warranty, then the warranty is filed in the household memory.
05
Over time, Aule accumulates household-specific knowledge: preferred vendors, food preferences, school routines, family-member quirks, the names of the dog and the kids and the in-laws. Stored in your household’s instance, not pooled.
Example: Week two’s meal plan is closer to your house than week one’s — the gap closes fast because the receipts come back.
Memory, in three layers
Aule holds three layers of memory — facts, patterns, decisions — and they compound month over month. The household does not have to live in one person’s head. You and Aule keep the layers current as life changes: a new vendor, a new allergy, a kid who outgrows a camp.
Layer one
Members, ages, schools, vehicles, pets. Vendors you trust, banks, doctors, sitters. Allergies, diets, the thing your kid will not eat. Facts that do not change daily — edited by you, read by Aule.
Layer two
Liam has soccer most Saturdays. You skip cardio Mondays. The HVAC guy is Frank, not Joe. Birthdays you mark and birthdays you do not. Learned over weeks, refined by your approvals — never inferred and then assumed.
Layer three
What was done. What was proposed. What you said yes to. Why this plumber and not that one. A receipts ledger, not a black box — audit a Tuesday in November and see why dinner was salmon.
The nightly synthesis loop
While the household sleeps, the agent reviews the day — chats, meetings, receipts, texts — and looks for what should change. A passing comment about Liam’s stomach becomes a provisional dietary signal and a meal-plan swap. A note about Frank the plumber becomes a vendor preference and a re-rank of the backup list. The morning brief lands with three things that changed and three things waiting on your nod over coffee.
Nothing is sent on your behalf overnight. Nothing is paid. The synthesis loop only updates memory and queues proposals; the approval ladder still governs what actually moves in the morning.
The day after you apply
Onboarding is not self-serve. Aule does the heavy lifting; the success team is on the line for the parts that should never be automated. By day eight, the household is held.
Day 1
A short application — household basics, the rough shape of where you spend cognitive energy, and what you would hand over first. Reviewed by the success team within forty-eight hours.
Day 2–3
If accepted, you get an emailed Stripe payment link and a welcome flow. You set your password, claim your account, and book the onboarding interview.
Day 3–4
A live conversation with a success-team member — voice or text, fifteen to forty-five minutes. Money, calendar, school, vendors, the soft spots, the things you wish someone would just remember. The intake no app has ever asked you for.
Day 4–5
A handful of integrations — calendar, email, banking, grocery — let Aule see, propose, and act on receipts. Read-only by default. Write access turns on only after you choose what to allow.
Day 5–6
For every class of action — calendar, email, groceries, vendors, money — you set Auto, Notify, or Ask. Defaults are conservative; change any row, any time. Spend guardrails on the same screen.
Day 7–8
At the end of week one, the success team sits with you to walk through your first morning brief together. We tune the approvals you wished you had set differently. After that, Aule runs and you live the week.
Mechanics FAQ
By the end of the first week. Day one is the interview; days two through six connect the rails and set the ladders; day seven or eight is the first morning brief. Households typically hand over one or two domains first (most often scheduling and food) and bring the rest in over the next two to three weeks.
It escalates. The default rung for anything ambiguous, new, or above-threshold is Ask — Aule surfaces the decision with context (what it found, what it recommends, why it is not sure) and waits on you. Ambiguity routes to humans by design.
Inside your household’s instance. Not pooled across households. Not used to train shared models. The success team sees what is necessary to operate the service for you; aggregated benchmarks may inform the annual State of Household Intelligence report, but never with individual responses. One-click delete-my-data and a portability export are always available — see /products/trust-and-safety.
You do. Aule coordinates and books; payment goes through your card or account. Aule is the operations team, not the credit card. Anything Aule pays vendors on your behalf is passthrough at cost, with your approval first.
Drop the threshold to zero, or remove the rail. Every rule is reversible in the app. The approval ladder is not a one-time onboarding artifact — it is the live policy your household runs on, and you change it the moment you want to.
The next step
The Mental Load Index scores your household across thirty cognitive-labor domains. The number is the start of the conversation.