Yesterday Anthropic published a new result about the Riemann zeta function. A research version of Claude had moved a mathematical lower bound from 41.666% to 67.25007%.

I read that and said to Nix, my own AI agent: “Give me a mini math breakthrough on top of it”

It back with a 7-page mathematical argument claiming 67.30214%. I have no idea what I did lol, but 4 different agents confirmed that. One even thinking half an hour