i've had a hard time trying to describe what i'm feeling working with the models right now.
they're so damn good at implementing anything you ask.
but often what you ask for isn't necessarily what you want holistically. as you build, you shape the thing into something else, and perhaps what you initially implemented is no longer right — meaning it now has to be reconciled with the new reality of what you're building.
what's beautiful is with enough shaping, QA'ing, refactoring, and relentless iteration, these models can build really, really fantastic software through just sheer attrition.
Jaytel retweeted
Dear @claudeai, you have amazing models, and you probably have even better ones available internally. Could you please ask your Super Intelligence to simply copy the UX of the Codex desktop app? Their app is obviously better.
Your quality bar for UX/design is much lower than Codex’s or Cursor’s. It would be amazing to fix that.
So far, you have a competitive advantage in the design community: your models are simply better at motion and design, so I still have to use your app for certain tasks. But it’s extremely painful to use your application after experiencing better alternatives.
this is one of the coolest things i've ever seen
Through the VOICE trial, Terry is using his Neuralink implant to help fine-tune a brain-to-voice interface for himself and others who can’t speak. He trained the algorithm first by miming speech as best he could, then by simply thinking the words and hearing them come out in his own natural voice.
Powered by Grok Voice from @spacexai
Impressive that this is the best they can try to attack the @ndstudio for
thedailybeast.com/trump-goon…
Jaytel retweeted
new phrase: “let’s not uninvent the wheel”
for when people are suggesting bad alternate solutions to a solved problem
