Don't get me wrong, just like you I personally would also prefer LLM-integrations with a privacy-focused provider and I think Apple is a good party to get that from (assuming they're using good models and keep their privacy guarantees here...)

But in the end you're still often 'sending data to an LLM provider', and the EU enforcing them to also let that be competing LLM providers still doesn't sound like a bad thing to me.

If Mistral would give the same privacy guarantees: great! If a company wants to use their enterprise OpenAI subscription: great! Etc. etc.

Let's allow for some competition here and not force a specific LLM-provider onto users just because they like the Apple hardware and software ecosystem.

The way I see it: If a user willingly (1) installs another AI app like deepseek and (2) willingly gives it access to 'full phone and app data' with a warning screen or setting of whatever that seems... like a good thing?

I may not agree with those users that it's worthwhile providing their full private data to [some AI startup X] or [Some Chinese or US AI company that will hover up as much for their own use] but if the EU forces Apple to provide this as an option, that sounds good to me.

The whole point of the regulation is that the data on the device is _the user's_ data and if Apple can have its AI services work with the user's data, competitors should be able to do the same.

From my (admittedly European) perspective it looks like Apple is just throwing a tantrum here.

https://youtu.be/QaQHdF2DeRY

I have a Shelly Pro 3EM three phase current meter device on my home battery connection just to get more accurate data into Home Assistant because the battery provider doesn't provide it. (In hindsight I should've bought a different battery install but that's not something I can change after the fact...)

This has made my planning / research phase so much better.

It's faster because it has already read most relevant files, still has the caveats / discussion from the research phase in its context window, etc.

With the context clear the plan may be good / thorough but I've had one too many times that key choices from the research phase didn't persist because halfway through implementation Opus runs into an issue and says "You know what? I know a simpler solution." and continues down a path I explicitly voted down.

Maybe that's just CLAUDE.md and memory causing the difference of course.

As a matter of preference however I like the way Claude Code works just a lot better, instructing it to work with parallel subagents in work trees etc. just matches the way I think these things should work I guess.

You click the button in the browser, choose what to create 'I want to create a password (or a note, or whatever)' and then get redirected to their web-app and be presented with a pop-up asking what you want to create (I just told you, didn't I?)

I get it, when you move to a new web-app some things can break. But after using stored passwords creating new ones is pretty much the only other thing you do in the app, it seems to be core functionality that's been broken for over a year now, it's kinda madness tbh.

Edit: To be fair they offered a 'solution' when I reported it: "Don't use the web-app, install our desktop app instead."

I generally don't like the way codex approaches coding itself so I just feed its review comments back in to Claude Code and off we go.

Some of the hidden features:

- If you add a task to a routine and end it with a question mark it becomes a conditional where you can add specific tasks or subroutines on the yes or no condition.

- Make it 'weekday?' and it becomes a switch case for the week.

- 'is it 7:30 yet?' becomes a conditional that automatically detects if it's before that time. I use it for e.g. 'before breakfast' (at 7:30) -> go play (with a timer until 7:30)

N.b. 'fun' fact, my daughter wanted an avatar with clothes that you can earn. The idea was pretty easy to implement but getting nano banana (using a great Claude nano banana skill) to generate the correct images took... some practice. I think I spent more time on the images for the avatars than all the other features combined. It took way too long to realize simple stuff like "nano banana won't generate transparency, only a fake white/grey checker pattern"

Also learned the last iPod touch (great device by the way) has a really low screen resolution which can be quite challenging at times.

> All apps i found were either shit, real shit, or didnt solve my personal need.

Wow, what an amazing coincidence. I made something that looks pretty similar from the looks of it just a few weeks ago because I found the same thing. If I knew your app existed I probably wouldn't have made it. I wasn't even thinking of selling it, just made it for my kid because it makes following various daily routines easier.

It's a mostly vibe coded (well, I made tons of visual and technical decisions but didn't look at the code much except some spot checks) PWA that can run offline on an iPod Touch.

It has some quirks and hidden features for day schedules, timers, etc.

There's plenty of yank in there (yours is almost certainly built better) but it works pretty well for most daily routines.

You can check it out here: https://girls-routine-planner.huppie.nl/

And look what happens, it turns out if you switch app-focus and go back to the app (now immediately focused on the text input) your on-screen keyboard becomes pretty much unreadable (white on light grey...)

The feedback is probably going straight to the bin but I couldn't help myself to file a second bug with a screenshot from the first feedback submission with a broken keyboard.

Screenshot of this absolute disaster: https://imgur.com/a/CyXiVy2

Maybe it could be styled a bit differently so the search bar is more prominent and in the center of the screen, but just having a search bar without any distractions is a fantastic feature.

I think you should be able to use the LMM at home to help you better understand the topic (they have endless patience and you can usually you can keep asking until you actually grok the topic) but during the test I think it's fair to expect that basic understanding to be there.

[0] https://news.ycombinator.com/item?id=46043012

A calculator can be used to do things you know how to do _faster_ imho but in most jobs it still requires you to at least somewhat understand what is happening under the hood. The same principle applies to using LLMs at work imho. You can use it to do stuff you know how to do faster but if you don't understand the material there's no way you can evaluate the LLMs answer and you will be at fault when there's AI slop in your output.

eta: Maybe it would be possible to design labs with LLM's in such a way that you teach them how to evaluate the LLM's answer? This would require them to have knowledge of the underlying topic. That's probably possible with specialized tools / LLM prompts but is not going to help against them using a generic LLM like ChatGPT or a cheating tool that feeds into a generic model.

Since the testing tool they use does notice and register 'paste'-events they've resorted to simply assigning 0 points to every answer that was pasted.

A few of us have been telling her to move to in-class testing etc. but like you also notice everything in the school organization pushes for teaching productivity so this does require convincing management / school board etc. which is a slow(er) process.

[0] https://cluely.com/

This topic has been an interesting part of the discourse in a group of friends the past few weeks because one of us is a teacher who has to deal with this on an almost daily basis and is struggling to get her students to not cheat and the options available to her are limited (yes, physical monitoring would probably work but requires concessions from the school management etc. it's not something that has an easy or quick fix available.)

[0] https://oxide-and-friends.transistor.fm/episodes/ai-in-highe...

The answers were mostly on par (though different in style which took some getting used to) but the speed was a big downer for me. I really wanted to give it an honest try but went back to Claude Code within two weeks.