I've been quite dogmatic about using TDD when vibe coding. It has produced good results for me. But for a roster app I am making for my wife with CP-SAT it's just test-bloat running my velocity into the ground.

So funny when you start digging around open source repo's of the greatest companies currently hockey-sticking and you find prompts like this "no word salad"

I did the insta-thing for my 5 day hike in Peru! But I dont have Insta 😭

Ok hear me out. I am occasionally running Openclaw on a phone. I also have chromecast. Its serving me 5 ads in a row taking 75 seconds. That sucks. So I was thinking... home NAS/server meets multimedia always on-pc. Thoughts? What should I be looking for? Enough gpu for HD + some ram/cpu? Power?

If you cant explain in one sentence why your ai product solves the customers pain better than a Claude Code skill with bash or a custom Openclaw setup, why are you even building?

Leuke pet 😝

Institutional money made a bet that the productivity gains we're seeing in software engineering can be replicated across the board with white collar work. Which I believe to be wrong.

JP Ahonen 🇪🇺@jpahonen.eurosky.social · last mo.

I have repeatedly said even in the best use cases the enterprise ROI isn't there for current American models, except for few elite engineers world wide. Chinese models are putting that to the test: an order of magnitude, perhaps two, cheaper tokens. But no moat. Another domino falls...

A half time should last 15 minutes. This shit should be enough to ban America for 20 years. Go back to your “football” with egg shaped balls that you’re supposed to throw to each other like some fancy outdoor version of korfbal

I now put some guardrails and extensions in Pi and I think I'm never going back. Made the whole thing look pretty with Zellij too. Oh and if you're a cli-guy and dont use Zoxide, you're missing out.

Only posts feed down. Rip I guess.

redsolver@redsolver.dev · 6mo ago

Hi! SkyFeed is struggling and I need to find a solution. If you don't know SkyFeed, it's a service I built 2 years ago, enabling anyone to build and publish feeds using a visual block-based editor. SkyFeed is still hosting the most feeds on Bluesky, but that comes at a cost [1/X]

github.com/JustVugg/col... This is wild shit. TLDR: Running GLM5.2 on a 25gb ram consumer machine by putting large parts of the model on hard disk and only loading parts of it to RAM.

Running some evals for chatalytics.eu Mistral Small is consistently outperforming GPT 5 nano when it comes to latency while scoring better on the task output. Nice.

Would you use a website that allows you to browse Reddit as a grid like this?

lol

Currently setting up #OpenClaw on a spare Android. Ngl, its obscure and complicated to get things fully up and running. If you have a spare pc with normal linux distro, things are much easier.

Either Balogun or the US coach could have said "this is not how we want to do it" and not start Balogun. But statistically speaking one of them voted Trump I guess...

My god this is a wonderful feed to wake up to. Go #RedDevils

😂

Marcel Janssen@marceljanssen.bsky.social · 2mo ago

Ik hoop dat België vannacht met 50-0 wint van de usa. #FuckTrump #FuckInfantino

Yo wtf this guy is running GLM 5.2 on a single RTX 4060 Ti 16GB with 32GB of RAM. Some quality loss to make it possible, but wtf bro. huggingface.co/giannisan/GL... github.com/giannisanni/... x.com/Giannisanii/...

giannisan/GLM-5.2-ds4-gguf · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

huggingface.co

So is organising your econd brain / Zettelkast / @obsidian.md with vector embeddings a thing yet?

GTA6 is coming. Its gonna be 80 bucks. In 2026.

The Weeniest YO HO HOI 🏳️‍🌈@ssj3goku90.bsky.social · 2mo ago

Manage your own context window wisely

gtmauan.bsky.social@gtmauan.bsky.social · 2mo ago

Underrated productivity hack: clean your desk before you close your laptop, not when you open it. Future-you walks into a workspace that says “sit down and build” instead of one that says “ugh, later.” #buildinpublic #devlife #selftaughtdev

Turns out this isnt as cool as I thought on Anthropic api pricing.

Michiel Berk@michielberk.com · 2mo ago

I am really excited to run some parts of my workflow with Opus 4.8 and some parts of my workflow with the much cheaper GLM 5.2 (which replaced Sonnet 4.5 for me).

And then I - without profanity - filled in an email form. That could not be send. Because of a profanity filter. I am beyond fucking pissed right now. @ralphlaurenpolo.bsky.social here's some profanity for you: get your fucking shit together.

Wait wtf, the @astro.build guys build a framework on top of Pi? The rabbit hole deepens...

OOOOOOOOOOOOOOH