TOP LAUNCHES

Shopping by proxy

TOP LAUNCHES

“Buy for Me” is a new feature that lets Amazon’s AI purchase products from other retailers when it can’t find them on its own site. You never leave the app. Amazon just… handles it.

Recall’s Augmented Browsing adds an overlay to your browser that highlights keywords you’ve already saved in your knowledge base. It surfaces connections to things you’ve seen or noted before—without you needing to search or even remember.

The latest Midjourney update adds better coherence, faster generations with a new Draft Mode, and smarter prompt interpretation. It’s built to be more responsive, more accurate, and cheaper to run when you’re just experimenting.

Jammy Chat recommends tracks based on your current mood using a simple, conversational interface. It’s not pulling from your listening history—it’s reading the room (your room).

oMoo is an AI-powered haptic music player that lets you experience music through touch. It translates rhythm, pulse, and texture into vibrations you can feel in your palm, making music more accessible to deaf and hard-of-hearing users—or anyone curious about a different way to connect with sound.

THE BIG IDEA

OpenAI is open again?

OpenAI’s latest move isn’t a shiny new model or viral image drop—it’s a Google Form. But it might be one of the more interesting things they’ve done this year.

They’re working on a new language model with open weights and, for the first time in a while, asking the public what they want from it. The last time OpenAI released anything truly open was GPT-2 back in 2019. Since then, the ecosystem has shifted. Meta’s LLaMA models are everywhere, Mistral’s picking up steam, and open-source AI has gone from niche to normal. Now OpenAI is stepping back in and asking, “What should we build together?”

The questions they’re asking aren’t small. How should the model be licensed? Who should get to use it? What should the boundaries be? It feels less like a product launch and more like a pulse check. A way to show they’re listening, even after a few years of keeping things mostly locked down.

There’s still a lot we don’t know. What the model will be, how open “open” really means, or when it’s actually coming. But the shift in tone is clear. Less “here’s what we made,” more “help us make the right thing.”

Whether this ends up shaping the future of open AI or just softens the edges of OpenAI’s brand, it’s a reminder that even the biggest players know the conversation is starting to matter just as much as the code.

TRENDING DISCUSSIONS

Are “stupid apps” secretly the future?

That’s what Gabe is betting on. In this thread, he makes the case for “vibeware”—apps that aren’t trying to solve big problems, they just make you smile. Stuff like Klack, Googly Eyes, TabTab. No roadmap, no AI agent, just vibes.

He ties it all to the rise of vibecoding—using tools like Cursor and Replit to ship tiny, weird ideas fast, without pretending they’re the next unicorn. More fun, fewer pitch decks.

If you miss the old internet—or just want to build weird stuff without a strategy doc—this thread’s for you..