I build AI dev tools for a living, and the ground under that work keeps moving — what agents can do, what's worth building, what design even means when the doing is delegated. This site is the machine I built so those signals stop slipping past me.
Agents & Humans is my personal research space: a pipeline of Claude Code skills and scheduled scouts that finds, scores, and writes up signals on the future of agentic automation & design. Everything here is my reading of the sources — collected notes, synthesized articles, and working ideas. It represents nobody's strategy but my own curiosity.
The radar is tuned to one question and its fallout: agentic coding and the harnesses around it, MCP and the agent-tooling ecosystem, the UX of dev tools when the "user" is half-human and half-agent, the surfaces where humans direct and supervise agent teams, and the market forming around all of it. Sources that speak to that thesis score high; everything else stays in the library.
The editorial design system and the skill architecture behind this site are inherited, with admiration, from a research-pipeline project by Shelby Hubick. The content here is entirely my own.