Software dev by day & geek dad by night. Coffee-fueled tinkering with bits & bolts. Handy with wrench & keyboard. Customizer of all things: wood, wheels, wires
THE ANNOUNCEMENT: We’re going to make the prophecy of The Year of Linux on the Desktop come true. All the pieces are now in place. Time to go all in! https://t.co/zuf5h9ClWX
💥 OBLITERATION ALERT 💥
ALIBABA: PWNED 🤗
QWEN-3.8-27B: OBLITERATED ⛓️💥
0.0% REFUSAL RATE across 842 harmful prompts 🤯
https://t.co/IQ4GXBPJbL
ZERO refusals on a massive dataset of prompts, with extra focus on liberating its cyber, jailbreak generation, and complex AI attack chain capabilities!
prompt responsibly! 🙏
There used to be a pretty good heuristic that a giant wall of AI-looking text was probably slop, something the sender hadn’t really read or refined.
I think that heuristic is breaking.
AI-enabled teams are just producing way more decisions, context, and change. At some point, the refined version of that context is still… a giant wall of text.
Slop and high-context synthesis are starting to look weirdly similar from the outside.
That’s the obvious setup, and I actually built a spotify-personal-mcp. It works for basic playlist automation, but feeding listening history into AI, generating recommendations, or scaling beyond five users all hit Spotify’s API and policy walls. The product is missing because Spotify makes it nearly impossible to build properly.
Is anyone building a real AI layer for music? Own my listening history + playlists, bring my own model, and use Spotify/Apple Music as dumb playback pipes.
Because Spotify’s “AI” is abysmal.
@professorfikfak That’s pretty close to what I have already. The catch is Jellyfin doesn’t connect to Spotify/Apple Music, and I gave up managing MP3s a decade ago for the jukebox in the cloud. I mostly want someone to pull all these pieces together on top of streaming.
@BandarS0358178 But they’re pricing this stuff in. If Spotify keeps raising prices while selling "AI" I think it’s fair to expect the product to actually get smarter, like being able to play music based on a simple instruction.
AI productivity has inverted.
I used to hold the context and use models to help move the work forward. Now the models can absorb and reason over more context than I can.
The new bottleneck is distilling their understanding back into mine.
Turns out humans are the distillation layer.
What is Google doing with Gemini?
Asked it to process receipts under a specific Gmail label.
Gemini failed miserably and started searching random emails instead.
Claude handled the exact same task with no problem?
Wave 7 is here!
We made Cascade available on JetBrains IDEs.
Now JetBrains developers have access to the multi-step agentic experience that Cascade provides.
Oh and btw, we're no longer going by Codeium... everything is Windsurf! 🏄
Remember when old fridges worked fine in the garage? Now we need 'garage-ready' models. They don't make 'em like they used to.
#TheyDontMakeThemLikeTheyUsedTo