elpakal

Hmm. It says my city has 0% chance of visibility but I'm looking at it outside right now.

Working on adding Apple Intelligence to my macOS app built to analyze iOS app size metrics. I'm hoping to have a locally running assistant that can act like an iOS build engineer to provide optimization opportunities and more: https://apps.apple.com/us/app/dotipa/id6742254881.

Right now my app allows users to export build metadata as JSON which can be interpreted by LLMs for analysis, but I'd like to have this work on-device.

Sir the vibe coding didn’t work, break the glass and call in dev!

App Store Connect did not come out yesterday

which is definitely not buttery smooth, I use it every day

I mean isn't this just considered data poisoning?

Hello, I'm a former Staff Engineer branching out to help law firms and attorneys with AI product development and due diligence. Looking for partners in the legal space who want to build the next big thing in legaltech or need help understanding how to best adopt AI legaltech.

  Location: Golden, CO
  Remote: Onsite, remote, or hybrid are all okay.
  Willing to relocate: Negotiable
  Technologies: AI / PyTorch / Tensorflow / Typescript / Next.js / AWS / Node.js / Vercel / Swift / iOS / macOS
  Résumé/CV: https://fractional-engineer.com
  Email: kenny@ndukt.com

This is a great application of this quote. Insurance providers have 0 incentive to make their AI "good" at processing claims, in fact it's easy to see how "bad" AI can lead to a justification to deny more claims.

But then who’s gonna spend all day Googling stuff when the weather changes their plans again?

> PRs are a bad way to review and iterate on code

idk, we’ve (humans) gotten this far with them. I don’t think they are the right tool for AI generated code and coding agents though, and that these circles are being forced to fit into those squares. imho it’s time for an AI-native git or something.

Not so much project as task, but I have been really happy with "vibe coding" to fix my production crashes. In some cases these crashes are impossible for me to reproduce as a solo, and I trusted the fixes provided by codex which seemed to have worked based on the absence of these crashes now.

Maybe more suitable for a vibe debugging addendum.

I build and maintain an iOS app size analysis tool that runs locally on your machine. I'm researching how well Apple Intelligence is for on-device insights into builds—my app currently has the ability to export build metadata as a json file, which you can then paste into your favorite LLM for insights and optimization tips. But I'd like to have a locally running alternative (your own local build agent). It's been super interesting seeing Foundation Models in action, and the context window restriction has been a new kind of challenge.

https://apps.apple.com/us/app/dotipa/id6742254881

Fitting that the first image example they showed spit out "NSURL ass".

Nobody uses NSURL anymore...

Looks like it supports custom backends

The benefit I see is that it meets users where they presumable already are (GPT). As other comments allude to here, it's clear they see themselves as a staple of the user's online experience.

Are they still expecting us to get paid based on “revenue sharing”?

Hello, I'm an experienced Staff-level engineer branching out to help law firms and attorneys with AI product due diligence/development. Looking for partners in the legal space who want help with product adoption, risk analysis and/or development.

  Location: Golden, CO
  Remote: Yes
  Willing to relocate: Negotiable
  Technologies: AI / iOS / macOS / Swift / PyTorch / Tensorflow / Typescript / Next.js / AWS / Node.js / Vercel
  Résumé/CV: https://fractional-engineer.com
  Email: kenny@ndukt.com

I’ve been playing around with Apple’s Foundation Models, their on device llm has a 4k context window. That’s really been an interesting exercise in context engineering coming from others like Claude and GPT. I think those larger context windows have made me take context engineering for granted.

Wish I was cool enough to have an invite code. Oh well, as an iOS build nerd next best thing I can do is inspect their ipa I guess. Interesting that they have some pretty big duplicate mp4s nobody caught in NoFaceDesignSystemBundle: cameo_onboarding_0.mp4 & create_ifu_1.mp4 | 7.3MB and cameo_onboarding_2.mp4 & create_ifu_0.mp4 | 5.2MB.

Also I find it neat that they still include an iOSMath bundle (in chatGPT too), makes me wonder how good their models really are at math.

Adding a chat feature to my iOS app size analysis tool that runs locally on your Mac. My goal is to make everyone a build engineer, where you can chat with your builds and get insights and improvement areas. Testing out on-device Apple Intelligence models but need to find the time to do more validation testing.

https://apps.apple.com/us/app/dotipa/id6742254881

A lot of both defeatist and overly optimistic pro AI comments in here. Having built legaltech and interfaced heavily with attorneys over the last 5 years of my career, I will say that there is a wide spectrum of experience, ethics and intelligence in the field. Blindly copying output from anything and submitting it to the court seems like a mind boggling move, it doesn’t really make a difference if it was AI or Google or Bing or Thompson Reuters. This attorney is not representative of the greater population and probably had it coming imho.

There is definitely benefit to using language models correctly in law, but they are different than most users in that their professional reputation is at stake wrt the output being created and the risk of adoption is always going to be greater for them.

Neat, but not really addressing my point. My point is that you still need to roll out changes and LLM ApIs just work.

Depends on if he’s hanging upside down or not

Fellow inverter and this theory makes total sense

Came here to ask about how they view Apple Foundation Models as a threat.

> guarantees privacy by default, works offline, and doesn't rack up a massive API bill at scale.

I’ve been really interested in on-device ML for most of my career, and now I wonder how valuable these benefits really are. LLM vendor APIs are pretty performant these days, security is security, and with an on-device model you have to provide updates every time a new model comes out.

I’ll take that bet. You will be able to disable it but that won’t mean they won’t still run it.

I agree with the premise of the article, and have felt that we're probably seeing AI code gen tools being limited by the constraints being put on them by traditional source code management tools like git and GitHub. Those tools were designed for incremental changes (patches), and have worked well for humans to organize changes so they could be more easily reviewed, maintained and reasoned about. Units of work in the form of features, patches etc rely on "pull requests" which are a function of the above.

or move the disclaimer to the top. or better yet, have aggregators like HN add a badge if it's likely AI generated

The on-device foundation models framework is interesting to me. So far the responses have not been good but the potential is appealing.

Curious, what happens to those car accident attorneys if/when these become ready for the wild.