Making the Academy Homework Reminder

The first app I made in this lab wasn't grand. It came from one small irritation that repeated every week. Each academy my kid attends sends a homework text. The subject, page numbers, and due date are all mashed into one blob, and every academy formats it differently. Reading those texts, figuring out which kid each belonged to, and copying them one by one into the Reminders app took more effort than it sounds. With two kids, it was easy to mix up whose was whose.

Shrinking it to a weekend-sized problem

I didn't picture a grand study-management app. I decided to kill just that one chore — copying the texts over. Narrowing the scope that much turned it into something I could build over a weekend. The first version was a web app. Paste the text, and it figures out which child the homework is for, tidies up the subject and due date, and registers it into iPhone Reminders.

Every academy formats its text differently. So instead of hand-crafting a rule for each, I let an LLM read the text itself and pull out the subject, content, and due date. And since it deals with academy homework, a weekly timetable view came along naturally too. It wasn't a grand plan; it's a tool that grew as each needed piece got added.

From a web app to an app in hand

Using it on the web, the 'copy text, open site, paste' dance got a little tedious each time. So next I refined it into a native iOS app. Tap the home-screen icon, paste the text, and it goes from tidying to registering reminders in about thirty seconds. It isn't flashy, but it genuinely helps me once a week, every week.

Finished, and the question that remains

The features are all built. But I haven't shipped it yet. The main reason is that the core — reading the text — runs on an LLM inside the iPhone itself. I preferred keeping the texts on-device rather than sending them to the cloud, but today's on-device models don't yet produce tidy results consistently enough to ship.

So I'm waiting for the next Siri (Apple Intelligence) update. On-device performance is said to improve a lot, and at that point the same design might finally hit shippable quality. I built it small and carried it all the way to done — now I'm waiting for the technology to catch up.