Full North Korea 🇰🇵😂

Google is back.

I think it's generally a good thing that this technology become investable by the average person. ft.com/content/8d9e51…

Loading fully autonomous coding agent: 85% openai.com/index/introduc…

Testing out the new Codex coding agent from OpenAI. See below. Small task, but still a bit magical that it happens async while I do other things.

Testing out the new coding agent, Codex, from OpenAI. Got it working with Rails! The magical thing is that it works async while I do other things.

How the hell did they get til $6.5B? nytimes.com/2025/05/21/tec…

The AI layoffs have begun

Corrected title: Sam & Jony declares their love for each other

At Tau, we develop world-model based reinforcement learning algorithms to learn reliable policies and improve beyond human performance. Blog about our first world model: tau-robotics.com/blog/world-mod…

General-purpose robots will have to learn from trillions of trials. We have developed a new world model that is optimized for reinforcement learning in imagination. We scale it up to 1B parameters and train it on a large dataset of human and multi-camera robot data.

The problem with Dia is that even if it gets as good as Arc, the chances of it generating any meaningful revenue for TBC are very slim. Unless the traction justifies another round of funding, I can't see how Dia is going to have a different fate than Arc.

I want this...

Happy to contribute 🙌

RubyLLM from Carmine Paolino is a better tool than any other Python lib I used before. I really not understand how it’s not more hype…

I made this entire 3d computer thing with Claude Code 100% on the server in just a few hours today I've never been so fast, no deployment, no Git push and pull, I'm not even deploying to production (like I normally do), I have Claude ON production now It's vibecoding on

Super interesting analogy for AI in programming: "I believe with Claude Code, we are at the 'introduction of photography' period of programming. Painting by hand just doesn’t have the same appeal anymore when a single concept can just appear and you shape it into the thing you

Is this the beginning of personal generative interfaces? Today, it takes a few hours, but in a year, maybe just one prompt?