I'm a product owner, not a developer. I vibe-coded an iPhone app and shipped 5 versions in 5 days

I'm a product owner and spend most of my day in meetings. Writing the minutes afterwards was the part I hated, so I built CalNote AI. It's an iPhone app that starts recording when a meeting on your calendar starts (one tap on the notification), stops 10 minutes after it ends, and saves a clean note to Files, Obsidian or Notion.

I'm not a developer. The first UI came from Rork, and the server, code reviews, tests and App Store submissions were done with Claude Code.

What happened since 1.0 went live on Oct 28:

• Oct 29: 1.1, a full redesign the day after launch

• Oct 31 to Nov 2: three more updates (a second nudge if you miss the start, keeping Pro after a reinstall, sharing the summary as plain text)

Honest numbers (Nov 2):

• About 30 devices have opened the app, including my own test phones

• 5 to 6 active devices a day

• 1 App Store rating

The biggest lesson: "it works" in the AI's answer and "it works on a real phone" were different things. Recording quietly stopped a few seconds after the screen locked, and I only caught it by measuring the length of the audio files. Since then every update gets tested end to end on a real iPhone before it ships.

A question for other makers: how do you test the things your AI is confident about but can't actually see, like background behaviour on a real device?

I'm launching CalNote AI here tomorrow (Nov 3). If you have a second, I'd love to know whether this one-liner makes sense to you: "Tap once when a meeting starts. Notes land in your folders."