Adding INSANE amounts of butter and cream to everything is the core of French cuisine. That's why it tastes good 🥲

Optimally Packed Gridfinity by Xowap | Download free STL model | Printables.com

Finally got a decent handle on Printables 🥲

All thanks to my new revolutionary model

Optimally Packed Gridfinity by Xowap | Download free STL model | Printables.com

It has been time I didn't release any 3D printable model, but I'm confident that this one will be of major use.

Meh...

OpenCode with a cheap model like Gemini 3 Flash will match Claude Code on every metric for less than $1/h instead of Claude's $10+/h

Plus, having a big AGENTS file sucks, you're much better off with well-sized (=small) skills

I mean... Launch videos from any Apollo mission are fucking incredible. Third parties like Everyday Astronaut are able to have their own amazing images on the ground. And Artemis serves us a pixel soup along with lens switch and blackouts at every key moment?

Given that the Columbia disaster is mostly due to sunken cost fallacy, I can imagine that they changed their approach to safety.

But also... The lunar orbit is more the job of Space X/Boeing, and neither seem to be able to deliver so far 😅

It's less ambitious than Apollo 8, the first flight ever out of Earth's orbit though

If you got a printing farm and no compelling reason to replace everything by 2031, I guess you're pretty pissed about this?

From the way they royally fucked-up the shots of the take off, I'd say indeed they didn't prepare as well

Apollo 8 _did_ orbit the moon, here they just orbit Earth and happen to meet the moon on the way. That's cheating honestly. The "most powerful rocket ever" has limitations it seems?

Artemis II's flight path seems extremely unambitious, or is it just me?

Like in terms of "going to the moon" that's the easiest possible thing?

Using Gemini Flash instead of Pro really made my life so much better 🥲

A dark-themed "Certificate of Completion" from the website amIaTechBro.com. The header reads "Am I a Tech Bro?" in large white text. Centered in a bold, vibrant green font is the classification: "NOT A TECH BRO." Below this, text indicates a "11% agreement with tech bro ideas," accompanied by a progress bar that is mostly empty, anchored on the far left under the label "not a tech bro." Small text at the bottom jokingly claims the certificate is "100% legitimate" and should be added to LinkedIn. The issue date is listed as April 1, 2026.

Got scared for a hot minute, but everything's fine now

That's a surprise

not an april fools joke: svelte is no longer just a web framework (or rather, won't be once we merge @paolo.ricciuti.me's magnum opus). have seen some extremely cool demos of this

Got so many hopes about that

However with AI, we have an opportunity to do things that wouldn't scale with humans. Like consider every zebra scenario. Skim through insane amounts of data. Simulate every measurable biological system.

Could this fix things?

Which makes me start to wonder about the medical method and bottlenecks. It's not how smart or knowledgeable the doctors are apparently. It's the algorithm itself. As an engineer, I couldn't deliver an algorithm that works as bad.

Trying to sovle some personal medical situation, after getting bounced by ERs I've decided to implement the "Dr House algorithm" with n8n and LLMs. It give me a bad/incomplete diagnostic but yet it was essentially the same than 3 teams of ER doctors, with suggestions that _did_ find the actual issue

Funding models are not binary. Beyond philanthopy and VC, there are other patterns including mission-locked, co-op, concessionary, community & oss.

Here's the updated "Funding Models Are Not Binary" slide from my Feature/Product/Business talk #ATmosphereConf. Turns out there are even more great examples already in the Atmosphere!

Common knowledge of eye movement tells you that people will read screen diagonally from top left to bottom right corner... However, I'm pretty sure that now you start by the thumbnail/quoted frame and then read the tweet/comment above

I'm told that the fall of the Roman Empire was more than 3 years ago

How good does it work on VisionOS? One year ago it was HORRENDOUS

Well, if you chunk down the work into a gazillion skills, foreseeing in advance all the issues that are going to arise... Yeah sure. It's gonna take longer than coding it by hand, though.

And more importantly, why the fuck it's always the same ~100 tracks in the list... The algorithm should automatically discard stuff that you've heard over and over again.

Conversations of AIs wanking off each other are wild (ly boring)

I don't care if it can spot a security vulnerability in 10 million lines of code, I'm sure it still can't architecture any of my basic projects properly

Regarding Anthropic's newest model, in French a "mytho" is a big fucking liar and that's all I've got to say on the topic

Is that me or is the number of products that suffer serious bugs in production is completely skyrocketing?

Not to point any fingers at vibe coding obviously