The cert and API key are account-level — set up once, reused for every app under that developer account.
After the first time, a release is three commands: bump the build, upload, submit.
Phone stays in your pocket.
I shipped an iOS app to the App Store entirely from the terminal.
No phone plugged in. No Developer Mode. No clicking through Xcode.
Here's how it works 🧵
What you still can't automate:
• Resolution Center replies — web UI only (but optional; a fixed build re-enters review on its own)
• Screenshots and first-time listing — one-time browser task
Everything else? Terminal.
TLDR:
If you use AI tools every day, your input matters a lot.
And if typing is slowing down the quality of your input, it might be worth changing the way you prompt and switching from typing to talking.
Hot take:
Most people are not getting bad AI outputs because the model is bad.
They are getting bad outputs because their input is too thin.
The easiest fix I’ve found:
Stop typing your prompts. Start saying them out loud.
My usage data made this obvious:
- 9,388 words dictated in a month (up nearly 1000% since last month)
- 225 fixes made by Flow
- 82% of my usage was AI prompts
I’m using it to give AI more context, get better first drafts, and waste less time.
Forget looksmaxxing, start postmaxxing.
People spend all day vibe coding products, but story is just as important.
I built a product that turns your codex, claude code, or open code covos into x posts or articles.
Check it out out: https://t.co/PRPSyZYwFi
Texts like this fire me up.
My friend lives on a kibbutz and the system they use to buy drinks at their local pub is literally built on paper and pen.
I spent an evening and built them an app that lets them move to the modern era.
Now it's actually being used.