Software should serve your intentions.

Explorations in maximally user-aligned iOS apps.

A tap travels toward the old default app. The shield intercepts it and redirects it up to the intended action, which lights up when the tap arrives. The vertical distance between the two endpoints is the gap. the tap the shield the reflex the intention the gap
The redirect stands in the gap.

01 · The mechanic

The space between a declared action and an ideal action.

You say you want one thing. Your thumb taps the other.

That space between what you want and what your thumb does is where we build. It is not a willpower problem. It is muscle memory. Apple's Screen Time API lets software stand exactly in that moment, and the pattern is always the same:

  1. 01

    Set up the Screen Time controls

    One-time permission. The user chooses which apps get intercepted.
  2. 02

    Snooze the legacy muscle memory app

    At the moment of habit, a shield covers the old app.
  3. 03

    Surface the new experience

    Redirect to the upgraded version of what the user actually wanted.

“Upgrade” Would Like to Access Screen Time

This lets it snooze the apps you choose and step in at the moment of habit.

Don’t Allow Allow
Inspiration PrayLock · Duolingo

We are not the first ones building in this gap. PrayLock proved the shield can carry a daily spiritual habit. Duolingo proved a well-designed loop can outcompete every default on your phone.

GIF placeholder PrayLock in action

PrayLock

GIF placeholder Duolingo's habit loop

Duolingo

02 · Our work

One mechanism, three experiments.

Tap to open

myLine

Scroll less, text your friends

Live on the App Store · Free · No account · Local · 2.1 MB

TikTokInstagramX A camera-roll picker built for sending

Our first attempt at using the redirect. The doomscrolling apps get snoozed, and in their place: a photo picker that makes it easy to send photos from your camera roll to friends and family. The lock is the nudge.

myLine app share screen: a gentle nudge to share your week, with recent photos ready to send
Nudge
Pick the photos worth sending
Pick
Sent as a text, not a post: photos go out through Messages
Send
Your people, your pace: choose your contacts and how often myLine reaches out
Tune
What we learned
Doomscrolling happens from a low energy state. Sharing photos is a higher energy activity, so the redirect creates real friction and real user frustration.
The brightest glimmer
I started sending more photos to my grandma and realized she loves anything I send.
A sunny porch selfie sent with myLine
Happy Friday GG !

Delivered

Oh my precious grandson it would be impossible to love you more!
GG is a real grandma. The reply came in minutes.
Next steps
Niche down. Rebrand around one job, send more pics to Grandma, and push. As a standalone tool the core loop is weak; it is mainly improved connectivity with family. The angle I expect to work on socials: sharing grandparents' reactions when the photos arrive.

CheckCheckRide

A rideshare aggregator you won't forget to use

UberLyft Both prices, side by side
GIF placeholder CheckCheckRide in action

Inspired by Griffin Li's Hackney, Obi, and others before us. The data is clear: neither Uber nor Lyft is consistently cheaper, yet fewer than 20 percent of riders check both.

The existential question: why has a rideshare aggregator never gotten popular? Our hypothesis: it is a muscle memory problem. When you go to call a ride you are in a rush, and you instinctively reach for Uber or Lyft.

The redirect is incredibly powerful here, because we have very high confidence the user genuinely prefers to see both options. It adds a handful of seconds of delay and returns significantly more data: price, and estimated arrival time.

The glimmer
Data from our beta testers surprised me. Highly price insensitive users were excited to choose the cheaper black car. They did not care about saving money and would never have compared manually, but when they saw a $50+ delta for the same product, they were excited to win.
The loop
An invite system: the app tabulates each user's savings, then suggests raising their referral bonus for onboarding friends. The king metric is total dollars saved, a real number derived from the price gap between ride types. Aggregate the demand side, and a new rideshare market entrant can finally break the cold start problem.

Pray For Them

You can scroll later

Coming soon to the App Store · Free forever · No account · No ads · MIT source

Every app, each morning The people on your heart

A simple, free take on PrayLock, focused on helping you pray for specific people in your life. It locks all your apps early each morning until you have started the day with gratitude, praying for the people who need it.

Screen recording of a locked home screen: tapping an app brings up the prayer shield, Time to pray for your loved ones, with a Ready to pray button
The real lock, recorded on device
Where it came from
Telling people I will pray for them, then feeling disappointed when human fallibility got in the way of the promise. I want software to serve my intentions.
On your honor, by design

Nothing checks that you prayed. You tap done, and the app takes your word for it.

Why it's free
Because it can be. 100 percent local, no logins, no servers, and it took an afternoon to make with Claude. This is the model for my vision of software: no ads, no paywalls. Just a local hyperstructure. The privacy policy says it plainly: built so that we could not read your data even if we wanted to.

03 · North star

Guiding principles.

Software needs to be user-aligned. As we give it more power to impact our lives, we need more certainty about how it will make decisions and what its north star is.

Most software is misaligned by default. Its north star is engagement, so it optimizes for the tap, the session, the streak. It points at itself. The finger, not the moon.

Engraving of a hand pointing up at the full moon above a hillside.
When the wise man points at the moon, the fool looks at the finger.Chinese proverb

The battleground has three adversaries: vendor lock-in selling commodity products, algorithms tuned for dopamine and dwell time, and plain human forgetfulness. Every experiment above is the same bet made three ways: put the intercept on the user's side.

User-aligned software points at what you actually want, and measures itself by how well it gets you there: dollars saved, photos sent, prayers kept. As software gains agency over our lives, that alignment stops being a nice-to-have. It becomes the whole game.