←back to thread

We're Joining OpenAI

(www.alexcodes.app)
192 points liurenju | 1 comments | 03 Sep 25 18:38 UTC | HN request time: 0.2s | source
Show context
asdev ◴[03 Sep 25 19:36 UTC] No.45119639[source]▶
isn't this useless with Claude Code now?
replies(1): >>45119706 #
liurenju ◴[03 Sep 25 19:45 UTC] No.45119706[source]▶
absolutely no. Alex is heavily optimized for xcode. If you work on an ios project that has more than 700 files, you'll understand how accurate it captures the context.
replies(2): >>45119928 #>>45120841 #
sumedh ◴[03 Sep 25 22:00 UTC] No.45120841[source]▶
> Alex is heavily optimized for xcode.

Is there a reason why the Claude cannot do the same thing?

replies(1): >>45122529 #
_mu ◴[04 Sep 25 01:43 UTC] No.45122529[source]▶
It's not that Claude can't but iOS / Apple development is such a bear and beast. You could really build a whole model just to solve for that ecosystem. There are huge issues in the Apple ecosystem with documentation and so much tribal knowledge. In many cases it's hard to know what is the right thing to do, and there is keeping up with all of Apple's required changes.

So Claude could do it, it just seems like they're focusing on a different set of developers for the moment.

replies(1): >>45123886 #
1. hobofan ◴[04 Sep 25 05:29 UTC] No.45123886[source]▶
I don't think that makes it as special as you think it does.

> There are huge issues in the Apple ecosystem with documentation and so much tribal knowledge.

I struggle to come up with an ecosystem where that doesn't apply. React, Angular, .NET, .... Though some of them probably even suffer from overdocumentation, e.g. React with the same beginner level tutorials / open source code regurgitating bad patterns, and you then have the challenge of separating the wheat from the chaff.

The question is really whether maintaining an ecosystem-specific model would be able outperform a better generalized coding model, and even further whether the marginal improvements would justify the additional maintenance process/cost.