"We have a blameless culture, so people can feel that their mistakes are learning opportunities, despite causing misery to every single customer that depends on our shit"

Ich benutze Sachen selten so, wie sie eigentlich gedacht sind

LLMs are getting there 😂

🇧🇷 o @DarioDurigan como ministro foi uma boa surpresa. E assim, oficialmente, o Largo de São Francisco produziu mais ministros da economia brasileiros do que a Escola de Chicago.

🇧🇷 Se alguém tiver ritmos do teclado yamaha ai sobrando, me envie por favor 😂 psr.modalharmony.net/#style=…

SaaS CEOs are getting squeezed from every direction. Renewals are due. Customers are gutting budgets to chase AI. The US economy is cracking. And if you don't have an AI story, you're not just losing deals, you're losing the conversation entirely. The pressure is coming, and there's nowhere to hide.

POV: Real life Hitchhiker's Guide. You're living in the German Vogon Sector. You have access to world-class Siemens medical technology, but the nurse is staring you down because you didn't bring a towel as explicitly stated in paragraph 4, page 7 of the examination request form.

Can someone explain the new "algorithm" like I'm a golden retriever? Someone I follow posts something => I don't see it on my timeline. One hour later, a random account with way less relevance (and one I don't even follow) reposts it, and that shows up on my timeline.

I have a test rig built with @turingpi -- still V1 :/ -- that acts as chaos coordinator for testing my little database; I always had to ssh into it to run tests, now I have a UI :P

Early in my career I was someone else's claude code agent Back in 2005, I worked on a clean-room design project porting an obscure proprietary AIX library that a production system depended on. The only requirements were passing an algorithms and data structures interview and signing an NDA stating I had never used that specific library in my life. I would sit in an office and receive emails. Alessandro would send me specs and plain text files with inputs and expected outputs. I would write my own test cases, fix the bugs, commit to CVS, and go home. Next morning the results would come back: performance regression on X, tests failing on Z, here are new input files. I never had access to the real system. Found out later it was running inside an internal pipeline that optimized cutting patterns. In hindsight, I was someone else's Claude Code.

I'm barely using the new M5 Pro but it seems with Tahoe, 64GB is not enough

The moment companies mandate "use AI for everything," engineers will do exactly that, which includes asking AI to evaluate the company's past strategic tech and product decisions. When it comes back saying every single one was wrong, the next question writes itself: "Great. So when do we replace the CXO with an agent?"

regrets: studio display without nano texture

I can’t even caption

If you experienced the #F1 Ferrari/Schumacher era, it’s an odd feeling to hear the hymns playing in reverse order.

Taskfile + Lua = <3