Habituu

A checkbox is easy to lie to. A partner who's waiting for your proof isn't.

Product design, end to endMobile app concept for iOS2025Core loop run live for a month

scroll to read · skip to the designs

ProductHabituu

A habit app that hands you a person, not another checkbox: one compatible partner who sees your daily photo proof and holds you to it. Designed end to end, but only after the core loop had already run for a month in six WhatsApp groups, with real pairs, that I ran by hand.

Details
iOS. A design concept, not a built appplatform
Research, IA, UI and the design systemrole
Solo and self-directed, no client behind itteam
Concept & design 2026, core loop tested over a monthtimeline
ProblemThe Lonely Checkbox

Most habit apps give you a checkbox, and a checkbox is easy to lie to. What actually keeps people going, a human who is invested in them, is the one thing these apps strip out. Twelve interviews, twelve people who'd quit a tracker within weeks.

SolutionA Partner From Day One

One compatible partner who sees your daily photo proof and holds you to it. And because a matching app is dead with nobody on the other side, everyone gets a clearly-labelled AI partner from the first session, until a human match takes over.

ResultThe Loop Has Legs

Before a single screen existed, the loop ran by hand: six real pairs, one month, me as the algorithm. Five pairs were still posting proof at week four, past the point every one of them had quit a tracker before.

Impact
5/6
Pairs Still Posting
at Week Four
A month in six WhatsApp groups (a signal, not a study)
11/12
Would Keep Going
on the Real App
The same twelve Reddit recruits the research started with
30 days
The Loop Ran
Before Any App
I was the matching algorithm and the notifications
The Problem

Every tracker they'd tried ended the same way: quietly, in week two or three. Not because tracking failed, but because nobody was watching.

I found twelve people on Reddit trying to stick to a habit. Every one had tried a tracker and drifted off it within weeks, and the same twelve later became the people I tested on.

  • Solo trackers leave you alone with your willpower, and the streak becomes a number you owe an app
  • Group apps spread accountability so thin that nobody feels personally missed
  • Coaching works, and costs more per month than most people will ever pay for a habit
you tick the box no one asks for proof the streak climbs anyway …and quietly, you stop

The lonely checkbox, drawn: nothing checks the tick, so the streak goes hollow, and then the app goes quiet.

"Streaks? I lied to it for about a week, ticking things I hadn't done. Then I just stopped opening it. What would have changed that? Someone actually waiting to see it. I don't skip the gym when a friend is already there."

The Live Test

Screens were the wrong first test. So I ran the app before it existed. Six real pairs, one month, no interface.

The same twelve, paired into six by habit and schedule, then one WhatsApp group per pair. That was the entire product: six group chats, no app, no screens. I was the matching algorithm and the notifications, both by hand.

It tested what screens cannot: whether a stranger you were paired with last week changes what you do on a Tuesday night. It did, and the nudges happened unprompted.

7:31, you post your proof it lands on your partner's phone the nudge comes back both streaks climb
Design 1 / 4

Show the Product, Then Ask

four questions, and a real match preview before any paywall

People hand this app photos of their real life, so the system stays plain: ink for structure, emerald for the other person. Four questions, each one a matching input, then a free preview of people whose mornings already line up with yours.

Design 2 / 4

A Habit Is a Frame, Not a Tick

and your own progress never appears without your partner's

There is no tick box in this app. A habit is a frame: a photograph, or an empty one asking for something. Every frame sits beside your partner's, so you cannot check your own progress without seeing theirs.

✓A month is a grid of frames too

The same unit at a different zoom. Missed days stay visible as outlines instead of being quietly dropped.

✓The freeze pass is honest about itself

One a month, and Sarah is told the streak was frozen, so the number never lies to her either.

✓Missing is survivable, hiding is not

A bad Tuesday should not erase three weeks, but it stays on the record. A record you can edit is worth nothing to the person watching.

Design 3 / 4

Matching Without the Black Box

six dimensions, visible weights, and a paywall only at peak intent

Swipe mechanics people already know, pointed at compatibility instead of attraction. The detail sheet opens all six weights, so a match is never a black box. Browsing stays free, and the wall waits until you try to Like someone.

When does the paywall appear?

✕
At launch, or during onboarding

Asking before anything has worked just trains resentment.

✕
A hard wall the moment the free preview ends

The preview proves value, but a blanket wall still lands cold.

✓
At the first Like, the moment of highest intent

You've seen the person you want. That's when the ask feels fair.

Design 4 / 4

The Partner: AI First, Human as Soon as Possible

one thread, honestly labelled, and built to hand itself off

A matching app with nobody on it is dead, and the wait for a human is where these apps lose people. So everyone gets an AI partner in session one: labelled AI, capped at a visible daily limit, and built to hand over the moment a human lands. One partner means one thread, so there is no inbox to manage.

How do you fill the empty day-one gap?

✕
Make people wait until a human matches

A dead app on day one, which is exactly when people leave.

✕
Seed the queue with fake or inactive profiles

Dishonest, and it poisons the trust the app runs on.

✓
A real, clearly-labelled AI partner from minute one

Accountability before any human shows up, without pretending.

Why it worked: it never pretends to be a person, it keeps the loop alive from session one, and it earns the right to charge only once you have felt that loop work.

The Results

Twelve people is a tiny room, and I read it that way. But the signal held for a month.

Five of the six pairs were still posting at week four, past the point they had all quit a tracker before, and eleven of the twelve said they would use the real app. Everything past that is guesswork: twelve strangers over one month prove nothing about retention at scale, conversion, or safety.

It proved the only thing I needed this early: the loop has legs. Only then did I start drawing screens.

"I didn't want to be the one who broke the streak for both of us. That guilt did more than any app ever has."

Reflections

01

Run the loop before you build it

A month of hand-run pairs answered the question screens never could: does a matched stranger change what you do on a Tuesday night? Everything designed afterwards was built on a yes.

02

Accountability needs a face

Guilt toward a person who's waiting beats a streak owed to a database. Every design decision that worked kept the partner close to the habit; every one that didn't, hid them.

03

The honest synthetic beats the fake human

A clearly-labelled AI partner with a visible message limit builds trust while it fills the empty room. A queue seeded with fake profiles would have spent that trust before day two.

What I'd Build Next

The loop is proven. The partnership is the product.

The trust layer ships first: reporting, blocking, and exact control over what a partner sees. Strangers hand each other a camera here, so trust is the feature everything else stands on.

The first build: these screens become a real app with live matching, its first cohort recruited the way the research was, from people already asking strangers for accountability.

The synthetic-partner experiment: the question the hand-run month could not touch. Does accountability survive knowing your partner is an AI?

Partner milestones: anniversaries, a hundred-days-together card. The relationship is what compounds, not the checkbox, so that is where the roadmap goes.

Somewhere out there is a stranger who got matched with you three weeks ago, and tonight they'll check whether you did the thing. Better go do the thing.