Watching videos at 1.5x speed is the best habit I picked up. Now live lectures and videos feel painfully slow and I zone out every 5 minutes.
We need a speed slider for real people.
Evernote has become unusable at this point. Subscription popups every damm hour like I owe them money.
Switching to @obsdmd has been the best decision.
Since LLMs are trained on already existing code that means LLMs aren't writing insecure code. They are copying ours.
Maybe vibe coding didn't break security, it just exposed how much of it was already.
Scribbble new version is out, with some crazy updates:
📸 New Snapshot tool to capture your annotations
↔ Switch toolbar position left/right
🔉 Sounds effects!
🤩 Lovely custom cursors based on selected tool
Update now!
2% left of what exactly?
Codex should provide a bit more transparency into how usage limits are calculated so users can better optimise and manage their remaining quota.
💫 @appjsconf is all about knowledge sharing and networking. We're back after the break – next on the agenda:
👉 @iwoplaza now on stage, talking about composable WebGPU shaders for React Native and beyond
👉 Yefim Vedernikoff on how they actually ship such cool features at @partiful
👉 @ziusko on portals (or actually the illusion of portals) in React Native
👉 @Nklockiewicz on the common challenges in building React Native apps with on-device AI
Hot take: always convert an image into text before submitting it to an LLM for analysis.
This way, you can double-check exactly what data is being fed into the model, instead of trusting OCR and interpretation in one step without verification.