Made these for my first iOS launch two weeks back.

First time doing App Store screenshots and I'm planning to redo the listing in the next update. Looking for honest craft feedback before I commit.

What I tried to do:

• 8 screens that each lead with the value, not the feature
• Hand-drawn elements (washi tape, doodles, weather stamps) for warmth
• Pastel theme background to give the listing a unified look in a category that mostly defaults to dark UI screenshots

What I'm not sure about:

• Headlines feel a touch precious in places — am I overdoing the "the good days find you again" voice?
• 8 screens vs the usual 5 — too many?
• The handwritten/scrapbook aesthetic vs cleaner editorial — too niche for the general App Store audience?

Open to anything. Tear it apart.

https://apps.apple.com/us/app/reflect-ai-journal-diary/id6762427801

Wrapped a 4-month solo build and shipped Reflect on iOS last week.

It's a journal app — voice transcription in 10 languages, paper-journal OCR, and AI insights over your own entries (Yearly Narrative, "Ask AI" with citations from your writing).

Stack:

- React Native + Expo SDK 54, EAS Build

- Firebase (Firestore + Cloud Functions on Node 22)

- Gemini via Vertex AI server-side, ADC — no client-side key

- RevenueCat for subs

- Native Apple Watch companion

- ~52 screens, 10 languages (EN/FR/ES/PT/DE/IT/AR/KO/JA/HI)

App Store: https://apps.apple.com/app/id6762427801 (Disclosure: my app.)

Happy to answer anything about the architecture, Expo 54 stability, or the server-side Gemini setup.

A few things I'd love this sub's take on:

1. Vertex AI vs. AI Studio key. I went Vertex + ADC to keep the key off the client. It added boilerplate. Worth it for you, or do you stick with a key behind a proxy?
2. Apple Watch companion. Has yours actually driven discovery, or is it purely retention?
3. Cold launch with 0 followers. Beyond ASO, what actually worked for your initial distribution?
4. Localization. Did shipping in 5+ languages pay off commercially, or would English-only have been fine for early validation?