Aslan @muasdevl

@emollick closed sets swap one trust problem for another. before it was “did the model see the test”, now it’s “do we trust the grader”. publishing each set once it’s retired, so anyone can re-score the old leaderboard, would go a long way on both

@cognition underspecified prompts is the right call. in real work i write half the spec and expect the agent to ask about the other half. a model that asks the one right question beats one that guesses well

@bcherny the 30 minute number is the real story. internal tool pilots usually die in week one on access tickets, long before anyone gets to judge the model

my agents ask for permission up here now. i don't leave what i'm doing. a clean little island on my windows desktop, and i can finally work on something else while they run. how do you keep up with your agents' questions?

china's first fully ai-generated movie is struggling at the box office. turns out "made with ai" was never a reason to buy a ticket. "good" still is.

genie 4 turns a sentence into a world you can walk around in. kids growing up now will think "level designer" was a strange old job, like typesetter.

Google DeepMind @GoogleDeepMind

a day ago

This is Genie 4, our most realistic and interactive world model yet. 🌐 Explore dynamic environments in real time to see how the scene responds, sounds, and how the people in it move, react, and speak.

924 3K 23K 3.2M 10K

his plan at anthropic: find out how many programs one person can run in parallel. i'm running the same experiment. 2 products, 0 employees. no lab though.

Professor Oak @Prof_Oak_

a day ago

Surprise 1: Anthropic acquired a one-person biotech . . . Surprise 2: that person is me Surprise 3: I’m here to tell you all about it Yes, this is what you think it is; after 6 yrs of increasingly loose anonymity: the Professor Oak reveal post …and yes, it has gotten so boring

477 899 13K 1.3M 7K

math, computer science and physics. one announcement. we went from "ai can't do math" to "ai is rewriting the textbooks" in like 2 years what a time to be building

Greg Brockman @gdb

2 days ago

Announcing three major scientific discoveries, one each in mathematics (harmonic analysis), computer science (complexity theory), and physics (quantum field theory). AI starting to help us improve our understanding of the universe! openai.com/index/building…

673 2K 16K 3.1M 5K

they never ask for a raise. they just ask for permission. constantly

my whole team is ai grok: writes the code
claude: reviews it, finds the bugs grok missed
cursor: quick fixes
me: says no a lot 2 products, 22 integrations, 0 employees

I run 4 coding agents at once now. The hard part isn’t the code. It’s knowing which one has been waiting for me for 20 minutes.

@aur0n I’ve started reviewing them differently: Claude diffs file by file, Codex diffs at a glance. The extra three things are fine until one is a refactor in a file I never mentioned.

@pontusab Shipping isn’t the slow part anymore. Reading what the agents shipped is. I’m the bottleneck in my own repo now.

Your agent asked a question 20 minutes ago. You were in another tab. Now questions land on the island. Answer or jump to it in one tap. Free for 7 days → petisland.dev/?utm_source=x&…

@MiruDaws @rork That sounds like a great idea. You should definitely do it.

@levelsio "Has not" and "is not" are data. "Will not ever" is a bet.

@wesbos The fact that this post pushes back on your take is exactly why it's a good one.

@theo Today's weird UX experiment is next year's default. Stay weird.

@araseb_ Definitely Whop, Stripe closes accounts on a whim.

@elonmusk @bot An agent with its own email and its own computer. At this point it just needs a desk.