Sneak peek at @o_openit 's current build. We'll soon be shipping the first version, fully #opensource. Knowledge platforms should be owned and governed by the community that builds them, not centralized, not gatekept, and fully agentic-friendly by design. The goal: a dynamic knowledge tree that's open by nature, one that invites you to contribute and watch it blossom. Still plenty to build and iterate on, but the first commit of this idea will soon be live for anyone who wants to contribute and help shape what it becomes. #oopenit

really hard to believe this new

u onfire sir!!

Haiku 5.5 from @AnthropicAI is a really nice, efficient, fast, and cheap model to handle daily basic tasks, docs, wire-frames designs and few advanced front-end jobs.. so happy with its performance, and i would tell you it's 20x times better tan any grok model for the same tasks. and 50x cheaper, even more cheaper than mimo models... bravo for claude fam

dear @benjitaylor !! that text box in the DMs section makes it extremely difficult to properly review what I’ve written. Before I send it, the window is only one line high. I’m using Mozilla on a Mac, i have to scroll up / down with just one line for the review, quite uncomfortable, I think the box should be much larger, and adjustable so we can write and review our messages more comfortably thx for your att

We need more people tweeting REAL sht in tech X, no bulasht

check this out if you are using @opencode

That “ tesla roadster ” is close to useless for people. It’s not safe for human transportation, and it never will be with those kinds of spins and stunts, they’re only good for a new version of *Knight Rider* film. But @grok it could be pretty useful for transporting goods, deliveries, and things like that…

cookin styles with grok

im cookin

Intelligence is about getting the maximum benefit from the natural abilities and resources you have

Superintelligence isn’t about doing everything. It’s about developing your natural strengths. A monkey and an elephant aren’t expected to compete at climbing trees; AI models shouldn’t be expected to excel at every task, either. Using a model with billions of parameters trained for "neuroscience" or programing languages to build a frontend isn’t superintelligence, it’s a forced fit, much like the one-size-fits-all approach that has long been used in education. Today’s models aren’t equally good at everything. As a user, I want professional-grade models for specific tasks, graphic design, writing, coding, and clear distinctions between those tasks and their costs, idontwannaallofthemmixedasafkinpack......

This is a thin line when it comes to evangelizing, im not fan of this, they can ban you without any reasons… Besides, they’re going to miss out on valuable feedback, because most of the time, when someone bullies an AI, it’s because the AI is behaving absurdly. I think this is exactly when an automatic feedback menu should pop up, encouraging the user to type **/clear** and reset the context a little.

Pastel green color schemes are great for reducing screen glare. I’ve added them to @lore_div_