π€ AI Tech Consulting at Every
π 2x O'Reilly author on Prompt Engineering & DSPy
π 350k students on AI Udemy course
π Built a 50 person growth agency
I just saw Codex leak a thinking trace that might explain why it is more token efficient. Small sample:
"Or if no org scope, keep legacy-only? But then write path not semantic. Could create report? No. Need ask user."
Codex thinks in grug brain to save tokens.
Microsoft leaked the training FLOPS for Claude Mythos
based on their slide Claude Mythos used:
6.1*10^27 FLOPs
(with 95% CI at 5.3*10^27 and 7.1*10^27, assuming 1 px measurement error)
This has been hard to write, but after four years, I've said goodbye to @Every
I joined four years ago thinking I knew how a company was supposed to operate, and I'm leaving with my whole worldview reset.
Every taught me that you don't need a ton of fundraising or a huge workforce in order to make serious products. You just need a few good people, some agents, and a lot of imagination.
There are very few places that shape who you end up becoming, and Every 100% did that for me.
Got to watch us go from a 4-person team building a scrappy newsletter business into dozens of people now growing a full AI-native product studio and media ecosystem, and somehow I never once got bored. Anyone who's had a job knows what a gift that is :)
I've never been surrounded by so much taste and human intelligence in one place. Thank you to @danshipper@katelaurielee@kieranklaassen@naveennaidu_m@poojary_yash@bran_don_gell@darustudio@kplikethebird@NataliaZarina@andreygalko for always pushing me to do better, I honestly think I got to make some of the best work of my career along you guys.
I'll be sharing more soon about what's next for me, but in the meantime, my DMs are open. Come say hi.
One observation is that when I train people one level behind they're bored, one level ahead they're thankful, two levels ahead they're scared, and three levels ahead they're angry.
Mike and @NataliaZarina are the two most plugged in people in New York City (maybe everywhere that's not OpenAI or Anthropic) on how AI is being adopted inside of large tech companies, PE firms, and hedge funds.
Their level of access is insane. Turns out most of the world has never heard of Claude Code or Codex and still things making GPTs is state of the art.