I thought this was a joke. Meta now has made 30-50% of software engineers on core teams become data labelers.
Their job is "giving human feedback on AI-generated Github repos" in an org called Agent Data Optimization.
Maybe we are all training data generators after all.
Our comment sections are leveling up.
Reddit users across all platforms will soon be able to post, see, and reply to video comments. This is huge.
Redditors will be able to upload or record short-form video replies in eligible communities. Think tutorials, AMAs, live moments - all directly in the comment threads. This isn't just a new format, it's a new way to engage.
Read more: https://t.co/3ogr6x29Xi
More of the iOS app loop, now inside Codex.
The Build iOS Apps plugin lets Codex view and test your iOS app in the in-app browser, open SwiftUI previews, and hot reload edits without leaving Codex.
I tried whoop for a couple weeks and couldnโt see any difference between the data my Apple Watch was giving me (for no added cost) using tools like Bevel/Athlytic.
Can anyone who swears by whoop tell me how the monthly fee has made a better impact on your health than just your standard wearable tracker like Apple Watch/garmin etc? At least Oura ring is an entirely different form factor ๐ง
Am I missing something impactful here or is it just personal preference of what a consumer wants to wear?
My top 5 takeaways from @ryolu_, Head of Design at Cursor:
1. At AI-native companies, roles are muddy. At Cursor, โroles between designers, PMs, engineers are muddy. We just do what it takes based each personโs unique strengths and use the AI agent to tie everything together.โ
2. Start small and let one app evolve into a complete OS. Ryo started ryOS with just a soundboard. โI asked the agent, can you give a more retro Mac OS look? And it did.โ He then added more features over time. P.S. You can fork ryOS on Github and modify it to be your personal OS.
3. Prototype directly in Cursor to interact with live app states. Ryo uses Figma mostly for initial exploration. โWe prototype in Cursor because that lets us interact with live states. It feels a lot more real than pictures.โ All the back-and-forth of โ50 pixels offโ handoffs goes away.
4. Release to different sets of users in stages to polish features. From Ryo: โFirst, we ship to staff. Then we ship new nightly builds to our Cursor ambassadors. Next we ship to consumers and finally to enterprise.โ Each stage gives the Cursor team an opportunity to polish the product.
5. Specs become more important as AI models improve, not less. Ryo predicts: โAs models get better, theyโll be really good at implementing exact specifications. So the quality of your spec will determine the quality of output.