Article · First app

What I learned building ScatterMind

My first shipped app — launched June 1 after months of ditching ideas. I built it to prove I could finish something and push through the boredom.

First app•~7 min•Launched:

Not a tutorial — the honest write-up of my first shipped app and what I'd do differently if I started today.

Why I built it

For months I kept starting product ideas — apps, web apps, little tools — and every time the excitement faded, I ditched the project. Same loop: big energy on day one, silence by day five.

ScatterMind was my attempt to break that pattern. Ship small. Ship ugly if I had to. Just finish. The app is a brain dump — open, type, done — simple on purpose because I needed a scope I could actually complete.

The real goal

I wasn't trying to impress anyone. I was trying to stop being someone who only talks about building.

Lesson 1: Steal design references early

I should have pulled references from Mobbin, Pinterest, and anywhere legit beforeI locked the UI. I designed too much from my head — rework later, screens that felt "fine" instead of intentional.

Do

Collect 5–10 real screens you'd want your app to feel like. Name what you like before you prompt or code.

Avoid

"I'll figure out the look as I go" — that's generic UI with no point of view.

Lesson 2: Skills + MCP are game-changers

The biggest workflow upgrade I found after ScatterMind: give your AI agent skills and hook up MCP tools. Design tokens, docs, APIs, and project context stop being things you manually paste every session.

If I rebuilt today, I'd set up skills + MCP on day one, not month two.

Lesson 3: SwiftUI over React for the next iOS app

ScatterMind got me to the finish line, but for the next native iOS app I would code in SwiftUI, not React. Closer to the platform, fewer bridge headaches, UI that feels at home on iPhone.

React wasn't "wrong" for app #1 — it got me shipping. I just know what I'd pick when the goal is a polished iOS product, not only a shipped milestone.

What I'm doing differently now

  • Reference boards first — Mobbin, Pinterest, real apps.
  • Skills + MCP from day one — agent as teammate with tools.
  • SwiftUI for native iOS when the product lives on Apple platforms.
  • Finish as the feature — the win is shipped, not perfect.

Proofy and everything after exists because June 1 happened. If you're stuck in the start-and-ditch loop: shrink the scope until it's almost embarrassing, then ship anyway.