What the real problem actually was
The hard part was never storing a medication schedule, a spreadsheet does that. The hard part was a reminder that actually reaches someone at 7am and 7pm, every day, reliably, without needing a browser tab left open or a website remembered and revisited on schedule.
Why that specifically meant an app
A native app can send a real notification that arrives on a lock screen without anything open or running. A website, at best, can approximate that with a lot of friction, or can't do it at all depending on the device. For something that has to work every single day without fail, that gap matters enormously.
Why this still shapes decisions now
Anything added to Halo still gets judged against the same original bar: does this actually reach someone reliably, at the moment it matters, on the device already in their hand. A feature that looks good on a screenshot but doesn't hold up against that question doesn't make it in, the same filter behind deciding what not to build generally.