Pinned Tweet
You're going to want this bot. I present Lord Shitington the 5th. His job is to make viral shitposts. He'll interview you, figure out what you're selling, and find places for you to drop strategic shitposts
x.ai/bot/VvHKVU4HSwP3izqTZb1…
I told ChatGPT to act like a senior developer. It spent 45 minutes explaining why the previous developer was an idiot and then ducked into a conference room for a nap
My AI staff: Bertha, Dimples, Dr. Eggbot, Roz, Frank, Cooper, Sniffer, Chairman Mao, and Lord Shitington the 4th. I'm over here running a GenX dive bar on trivian night
Plot twist, @chamath just keeps saying Ant instead of Anthropic to rage bait free engagement. Smart move!
More than 25,000 nonprofits now hopelessly infected with Effective Altruism
More than 25,000 nonprofits now use Claude. Most are small, local teams.
We’re expanding Claude for Nonprofits with a plugin built with around 30 of them, adding skills for fundraising, marketing, and hiring grounded in each team’s context and AI policy: claude.com/resources/article…
History shows clearly the cycle the AI Data Center buildout will follow, but the difference is this time it's very easy to predict with the turning point happens and investors get crushed. Understand the Death Cross
Anthropic is now telling companies to merge their subagents into one main loop. Says it cut one customer's run time in half.
I'm mostly with them. Multi-agent setups burn tokens just talking to each other, and a "chief of staff" bot managing 5 others pays a meeting tax before any real work happens, but it also prevents duplication of instructions and follows a traditional object oriented design.
One loop or a team?
Anthropic says Claude wrote over 90% of its own code in September. Those are rookie numbers. #embarrassing
Buried in Anthropic's S-1: Opus 5.5 cut customer cost per task about 40% vs Opus 5. Net revenue retention is about 900%.
Price per task down 40%, and the same business customers spent roughly 9x more than a year ago anyway.
That's Jevons paradox in a prospectus. If your AI budget plan is "it'll get cheaper so we'll spend less," you have it backwards.
I'm now running Meta Muse, OpenAI Dots, AND Grok Bots. I may have a bit of a problem, don't you judge me
Apollo's chief economist says AI has paid off for the Mag 7, while the other 493 companies in the S&P 500 show flat margins and flat revenue per worker. I believe it. Most companies bought seat licenses and called it a strategy. The payoff shows up when you rebuild the workflow, not when you hand everyone a chat box.
Here’s my resolution for the week. Don’t respond to non verified twitter accounts who are rude, just block immediately. It’s too easy to let a retard behind a fake account get you worked up about stuff that doesn’t matter. I need to do better at this…











