00deadbeef

It's harder than you might think. Mature software is battle-tested. All of the bugs and edge cases are handled. You can't just copy that from the UI.

But my point is it won't be true, and people like to lie and exaggerate to get clicks and sell people courses

Well now that Astra for Law is here I give it about five minutes before social media grifters claim to be 10x lawyers

It's not terrorism because it's self-defence and it's not war because Donald Trump says so

You should try and free some up. If you're using swap you're rapidly killing your SSD as it'll struggle to do wear-levelling.

There's no mention of Mission Control in this review, but lots of people are complaining it's turned into a buggy and useless mess in macOS 27.

Apple verified me by the age of my account. Claude asked to use the age verification when I opened it on my iPad. Haven’t been asked for any other kind of age verification by Anthropic.

What makes you think AI isn’t damaging for kids? Teaching them to outsource their thinking to AI can’t be a good thing.

Before writing this did you read the part where it says you don’t even need ID?

They could just sell it at an honest price with an honest profit. It’s an upsell from cheap F91Ws so they could have made more money just by selling me a thing that reads from Apple Health. It doesn’t even need to track steps itself.

I also disagree my email address and step data is worth $450 to them.

Let’s be honest. This thing is not innovative. It’s going to be using cheap commodity chips. Its selling point is nostalgia. I find it hard to believe it’s not already profitable at $55.

ffs all it needed to do was talk to Apple Health and I’d have bought it.

It's 2026, I always expect an Electron app, unfortunately

My non-technical cofounder managed to vibe code a holding page with Claude Design and it walked him through deploying it to Netlify.

However for some reason it had him deploy a single HTML file with all the assets encoded as a huge base64 blob in the code that required a massive amount of JavaScript to extract and render.

It’s cool. One issue I have is that the colours for “Castle & château” and “Ruin” are very similar.

When you tap a marker often you can’t see the castle in the photo because it’s cropped to fit a rectangle with an unusual aspect ratio. That element should be allowed to adjust its height so the image is scaled at its own aspect ratio rather than cropped. Eg you can only see the mounds of Berwick Castle and Bamburgh Castle in the North East of England.

Also this one should be included: https://en.wikipedia.org/wiki/Caisteal_Maol

It’s still easy to understand as the more capable the model the bigger the celestial body they’re named after.

One good thing to come of vibecoding is native apps are less effort, so hopefully there will be more alternatives to Electron crap apps.

I don't understand the point you're making.

It can be both the most powerful LLM on the market, and have no adoption in critical infrastructure.

I think this could kill LLM development. What's the point in pushing boundaries, when your business model is already hard to profit from, only to be blocked from selling your work to the entire world? Where's the incentive to continue?

The first number is which generation of their LLMs it belongs to.

Fable is the first model in the 5th generation.

The second number is an incremental release, not a generational leap forward.

They do

https://platform.claude.com/docs/en/build-with-claude/struct...

> Structured outputs are generally available on the Claude API for Claude Opus 4.8, Claude Mythos Preview, Claude Opus 4.7, Claude Opus 4.6, Claude Sonnet 4.6, Claude Sonnet 4.5, Claude Opus 4.5, and Claude Haiku 4.5

I doubt they'll phase out Haiku, some work needs speed more than intelligence. Haiku can answer a lot faster than Sonnet.

I haven't tried Fable yet but my experience with Claude is it does not engineer things well. Without direction from me, it will either over-engineer things to the point of absurdity, or do the total opposite and have little to no abstraction with repeated code everywhere.

But it doesn't make you the good guy, it makes you the best of a bad bunch. The least bad. Dario gets a boner every time he talks about taking your job.

Opus 4.8 already drops to Sonnet when you ask it cybersecurity or biology questions

Of course. You use their AI to ship code full of bugs and security holes and then they conveniently have the tool to fix them, for an extra fee.

5 minutes to load and it just dumps me to a documentation site with no useful information about that this is, who made it, what it can do, etc.

What if they have an actual back-end with long-running processes and scheduled tasks?

Maybe that's why all these vibe coded slop apps also use Clerk for auth alongside Supabase etc

I bought a 24TB for £365 last June. It's £905 now.

I like it so far.

Some features I'd like:

- XL bar size - even large feels a bit small on a 6K display

- I have grouped windows off, but it would be nice if there was an option to still sort the chips by app, so all the app's windows are listed adjacent to eachother

- If not using the suggest idea above, it would also be nice to be able to drag and drop chips to sort them in the order I want

- Make the Applications menu open for clicks on the entire bottom left area of the screen so I can slam my cursor in the general direction, where it ends up at the bottom left pixel, and click like I could in Windows to open the Start menu

- Ability to give a desktop a name

- Ability to map a key or sequence (e.g. opt, opt) to open the Applications menu, again like how you could open Windows' Start menu with a key

Bugs:

- Clicking a chip to minimise a window, then clicking it again to restore it sometimes causes the window to change size.

- Quitting boringBar spawned three stacks of these: https://postimg.cc/WhmwHGNz even though it already has permissions granted. Clicking "Allow" just spawns another one so they never go away. boringBar is not running in Activity Monitor. Had no choice but to reboot.