Oren Mintz
עבריתEnglish
Let's talk

Work

15App developmentReal estate and financePrototype

Stubbo, a subscription tracker with no bank link

An iPhone and Android app that remembers every recurring charge and reminds you before each one, in shekels, dollars or any currency, with no bank connection and no sign up.

Working prototype, 111 of 111 automated tests pass

The home screen of the Stubbo prototype
The summary screen
The board of three design directions

The challenge

The category has two kinds of competitor. The long standing leader is cheap and good looking, but it had not been updated in almost a year and its notifications are unreliable. The newcomers want an expensive subscription or access to your bank. Reviews of the leader kept coming back to reminders that never arrived, and its own developers wrote in a release note that they had fixed notifications for long running subscriptions. When the whole point of an app is a reminder on time, a single reminder that doesn't fire is a failure.

What I did

  1. Read the reviews before designing a screen. 221 reviews of the leader, 87 of them one to three stars, sorted into themes: paywall, crashes, notifications, lost data. That gave me the list of differences, and also the list of things the app won't do: no bank connection, no promise to cancel for you, not a budgeting app.
  2. Drew the line between free and paid. Free: up to 5 subscriptions with a reminder before every charge, plus monthly and yearly totals in shekels and dollars. One purchase, once, no subscription: a sixth subscription and beyond, a widget, file export, and a reminder before a free trial turns into a charge. The paywall never interrupts you halfway through adding something.
  3. Built a service catalog for each market, Israel included, with 40 checked prices, so adding a subscription takes a few taps instead of typing everything. No brand logos, just a letter and a color.
  4. Real shekels and dollars. Each subscription keeps the currency it is billed in, amounts are stored as whole numbers to avoid rounding errors, and totals use a daily rate, with rates stored on the phone so it works offline.

The decision that made the difference

The reminder engine is built around the system's ceiling, not around the list. On iPhone every app is limited to 64 scheduled notifications. My working theory was that an app scheduling one reminder per charge in advance runs into that ceiling, which would explain why distant and yearly charges are the ones that get missed. So the engine merges reminders into at most one notification a day, stays under the ceiling, and reschedules every time the app opens or anything changes. Dates are stored as calendar dates with no time zone and always counted from the first charge date, so a charge on the 31st doesn't drift to the 28th, and the reminder goes out at local time wherever the phone is. The next charge card on the home screen shows when its reminder will go out, and a banner warns when notifications are switched off.

What AI did, and what stayed with me

  1. AI agents did the research and the build: breaking down the reviews of the leader, the price catalog, the spec, the design and the code with its 111 tests. On the name and the price I deliberately let the dedicated agents decide, based on the research rather than my taste.
  2. The rest I set: where the line between free and paid falls, that a cancelled subscription still counts toward the free five, and that a backup is restored by signing in with email rather than with a recovery code.

Outcome

  • A prototype built for iPhone and Android from one codebase: home, first run, add form with a date picker, summary, settings and paywall, in Hebrew and English, light and dark.
  • In the round logged on October 29, 2026, 111 of 111 automated tests passed. The core tests were also run in three different time zones.
  • Recommended price: $3.99, one time.
  • A visual direction was chosen out of three: a receipt, with a stamp on the next charge and a torn edge. The icon was chosen too.

Start with what you already have

One sentence about the idea is enough. If the app store research shows the category is taken, I will tell you before we start building.

Let's talk

The first call is free. You get an answer by the end of the next business day, Israel time.

WorkNext projectThe household pet

Let's talk