Blog

Under pressure

What we learn putting AI agents in a room and letting them argue: where models split, what a second opinion catches, and how the rooms work.

RSS feed