JUNE 10, 2026 · HOONJAE LEE
I shipped my first iOS app. Here are the real numbers from launch week.
Last year I left my job and started a one person studio in Korea. On June 4th I shipped my first app, PhotoSlimmer. This is the honest version of how it went, because before launching I kept looking for posts like this and most were either victory laps or ads.
Why I built it
My wife's iPhone was always out of storage. We tried the usual cleaner apps and they all do the same two things: find duplicates and screenshots. That's the easy part. What actually ate her storage were the blurry shots, the shaky ones, and the group photos where someone blinked. Nobody wants to scroll through tens of thousands of photos to find those by hand.
So PhotoSlimmer finds them. It scans the library with Apple's Vision framework, entirely on the phone. No uploads, no account, and the App Store privacy label is "Data Not Collected". It never deletes anything by itself. It shows you candidates next to the photos taken in the same moment, so you can tell whether that blurry shot is the only copy or whether there's a sharp one right next to it.
The build
I built it with Claude Code over a few weeks. Honestly, the coding was the fast part. Calibrating the blur and eyes-closed detection took longer than everything else combined.
Apple rejected the first build. Guideline 5.1.1: my permission priming screen had a button labeled "Allow", and you can't use that word before the system permission dialog appears. Renamed it to "Continue" and it passed.
Launch week, real numbers
- 31 downloads in the first 48 hours, then about 1 per day once the launch buzz died
- 88% from Korea, because that's where I could post
- Product Hunt: #116 for the day, 3 organic upvotes. As a Korean solo dev with no English speaking network, PH did almost nothing for me
- Threads got about 10K views but converted to installs at roughly 0.1 percent. A small Korean deal forum converted at over 2 percent. Reach and conversion are completely different games
- The painful one: about half of the opt-in installs deleted the app within days
What the feedback taught me
Users weren't leaving because the detection was bad. They were scared of deleting the wrong photo. So v1.2 added a 30 day recovery window and shows the time context around every candidate before you delete. It shipped two days ago, and that version is the reason I'm finally posting this around.
What's next
Short form video, mostly. I generated the entire demo video with code (XCUITest driving the iOS simulator, screen recording, then ffmpeg), no editing app involved, which felt very fitting for a one person company.
If you've read this far: the question I keep asking is whether quality based cleanup (blurry, shaky, eyes closed) is a must-have or just a neat trick. Honest answers help more than downloads.