% intervention as eval?

AI as infrastructure, not “tool” Your best people will produce great results. But org productivity is defined by the median, not your best. I’ve seen many companies treat Claude like it’s another tool when they should be harnessing it like this for the long-tail to deliver

everyone is trying to build async agents that work when they sleep but all they really need are Australians

Building with CC as someone who isn’t technical means I’m naively blind to the “impossible”. (I’m not complaining)

AI won't kill fundamental investing because more information doesn't kill alpha. We have decades of priors here (Excel, Bloomberg, alt data...all democratized analysis & information gathering, and didn't kill alpha). As measured by factor volatility, stocks are less efficient and

CEOs who continue to hand wave will soon have the world wave its hands at them

The only moat left is being a hardcore motherfucker longer than other guy.

Context engineering somehow turned into a DIY major in cognitive science Such a humbling experience once you try make implicit reasoning explicit So much happens to fire in the brain between input and action

AI washing is the new green washing…except Wall Street actively rewards the “efficiency” narrative, giving execs both a stock bump and political cover.

Of course it’s Bret Taylor that ushers in the era of “headless” applications. “But if you’ve ever built an agent before, you know the hard part isn’t going from zero to one, it’s improving it over time.” Have suspected that this is one of the most powerful capabilities a company

x.com/i/article/2036…

Very bullish on open source and local models Imagine running near-Opus-level model locally on that $600, 16GB Mac Mini you bought last month This 27B Qwen3.5 distill was trained on Claude 4.6 Opus reasoning traces and is putting up real numbers: - beats Claude Sonnet 4.5 on