If @edwardsanchez ran Xcode, he'd stop work on the editor and the AI panel.
All hands on the Simulator, and SwiftUI Previews that run outside Xcode.
So Claude, Codex and Cursor can actually use them.
I'd vote for that.
An agent that can't see the screen is just guessing at layout.
I'd keep the debugger though.
Do you still code in Xcode?
The top 10% of OpenAI researchers now burn $23,000+ in tokens.
Per day.
Up 105% since September.
Suddenly my own token bill looks very reasonable.
Do you track yours?
@doktordos Hi from the Netherlands! I keep the projects themselves off here, but Claude Code is on something almost every day. What are you working on?
Dutch supermarkets are full of pepernoten again πͺ
Somehow they show up earlier every year.
I'm Patrick, a dev who builds with AI agents most days.
If you're into Claude Code, Codex or shipping iOS apps, say hi.
Ever tried pepernoten?
If @edwardsanchez ran Xcode, he'd stop work on the editor and the AI panel.
All hands on the Simulator, and SwiftUI Previews that run outside Xcode.
So Claude, Codex and Cursor can actually use them.
I'd vote for that.
An agent that can't see the screen is just guessing at layout.
I'd keep the debugger though.
Do you still code in Xcode?
@iammutex Ha, that would cut a lot of my "no, not like that" moments. Most of them come from the agent guessing. Is it mostly web data or company docs?
Let's connect π
I'm a dev who works with AI agents most days.
Which mostly means reading what they did and typing "no, not like that".
Dark by 5 here in the Netherlands, so the agents get my full attention π
If you build with agents too, say hi.
What are you building?
@suroorwijdan Agreed. The posts that actually helped me were always about what broke and why. A number without the story is just a scoreboard. What's the best "what broke" post you've read lately?
@arthuryuzbashev Underrated. Most of us argue with ourselves for a week about screenshots when App Store Connect could just tell us. How long do you let a test run before you trust it?