gitgud retweeted
I want to be shipping at frontier speed, running a software factory or 24/7 persistent agent swarm or whatever, and getting the benefits of everyone else's work stacking up together.
But it feels like the ecosystem is increasingly fragmenting. Everything is customizable, everyone is building the same thing but in different ways that don't add up. Where is the schelling point?
What is the best one thing I should lock into and just use? How can I connect and interoperate all my subscriptions and apps and tools instead of silo them? Do I really have to build this myself?
gitgud retweeted
This shows why adding in-app AI agents to tools like Notion was a waste of effort, time, and money.
This is the future: getting closer to a proper workspace, no matter which tools you use, with AI as a layer on top.
Anthropic is paving the way.
Replying to @claudeai
Also new: with interactive connectors, you can now open the full version of a tool inside Claude. Work in Adobe, Atlassian, and Box beside your conversation. Make changes yourself, or ask Claude to make them with you.
Rolling out on web and desktop.
gitgud retweeted
Hear me out:
We hated "waterfall" because after Big Upfront Planning, implementation took FOREVER, and by the time we were done, the plan was out of date
But with AI, implementation is rapid (100x faster, or more, than before)
So Big Upfront Planning can now make sense?
gitgud retweeted
burying fermat's last theorem in the fifth post of the announcement thread, on a model release that doesn't even merit a whole number
Replying to @claudeai
A team of 1,000 Fable 5.2 agents fully formalized Fermat’s Last Theorem in 30 hours, translating the proof into a language a computer can check step by step.
With the same setup, Opus 5.5 completed only 26% in 18 hours, and barely made any more progress after 30 hours.
Replying to @diagonalized_
security. if you have to rewrite all human written code from scratch anyway (and you do), might as well make some changes along the way
gitgud retweeted
I wonder how long itll take AI labs to solve the pink elephant problem in LLMs. So annoying
This is strange because I did this 8 months ago and I decided it's completely useless for my own thinking. Yes there's a lot information in there, and maybe some of it is actually useful. But until I engage with it there's no added benefit for my own cognition
gitgud retweeted
The first post-social media election also happened to be the first election in years where moderates in both parties overperformed. You should take this question seriously.
gitgud retweeted
everyone is doing a pronouns/land acknowledgement bit but actually the fucking awesome thing is that a large part of why this sweep happened is we had a lot of really strong candidates that didn't lean into the woke 1.0 nonsense
10,000 years of bill clinton neoliberal abundance
gitgud retweeted
If President Trump loses control of Congress, “he will want to punish a nation that, in his view, has ungratefully rejected him,” @davidfrum argues: theatlantic.com/ideas/2026/1…
Replying to @TheStalwart
trump is too feeble/old/unpopular to hold the coalition together. theyre going throw him under the bus but keep all his worst tendencies
gitgud retweeted
Drag-and-drop makes the digital graspable, but building it well is fiddly enough that many don’t bother. Declarative Dragging (with Josh Horowitz & Jeff Heer, #uist2026 best paper award), lets you make an element draggable just by listing the states dragging it can reach.
gitgud retweeted
We've reached 3:30 eastern and there is no blue wave, much less a blue tsunami. It's going to be very close in many states. Get out and vote if you haven't already!




























