
Snatched
AI photo editor for flawless skin & body in one tap
Hey IH 👋
I'm a solo founder and just got my first app, Snatched, into TestFlight — figured I'd share the journey and ask for feedback.
The idea: every photo editor I tried was either too complicated (endless brushes/sliders) or made people look fake — that plastic, over-smoothed AI skin everyone spots instantly. I wanted one-tap edits that keep you looking like you on a good day.
What it does: one-tap AI blemish heal (keeps natural texture — this was the hard part), realistic body sculpting, one-tap background swaps.
Stack: SwiftUI, Supabase Edge Functions proxying Replicate (Seedream for skin, Flux for backgrounds), AWS Rekognition for moderation, RevenueCat for subscriptions. Body slimming is a local Metal/Core Image warp — NOT AI-repainted, because letting AI repaint bodies produced cursed results.
What I learned:
- Making AI not over-smooth is harder than making it smooth.
- Solo-launching = you're the QA, designer, lawyer, and marketer. Account deletion, privacy manifests, App Store review… it's a lot of non-code work.
- Content moderation for a face/body editor is non-negotiable and genuinely tricky.
It's in beta now, heading to the App Store soon. Looking for honest feedback — what makes you trust an app with your photos, and what makes results feel "real" vs "fake"?
Site: https://www.snatchedllc.com — happy to share a TestFlight invite if anyone wants to kick the tires 🙏
About
Building my first iPhone then Android app. 1 year ago I had no experience coding (or vibe coding). This is what I always wanted to do growing up, but life took me into the culinary field. New to this but determined!!

2 Comments
Very valid point, thank you for your feedback!
The interesting challenge isn't making photos look better—it's making people trust that they'll still look like themselves. I'd keep validating which edits users consistently accept without questioning authenticity. That trust boundary may become a stronger differentiator than adding more editing features.