Agent Reliability: Your Harness Is the Product
Why long-running agents fail, and the verification loops that keep them honest.
Finding signal in the age of AI
Research notes, reading seeds, soundscapes, and photographs. A garden tended by a builder.
Deep dives on verification and reliability, plus where capital actually meets constraint.
Why long-running agents fail, and the verification loops that keep them honest.
What LLM-as-judge gets right, where it quietly fails, and how to calibrate it before you trust it.
A framework for where the real constraints sit: compute, data, energy, or trust.
A living dashboard of AI infra signals: capex, capacity, and constraint moves that matter.
Notes on AI, growth, and humanity. Every week.
The best builders in the AI age aren't the fastest. They're the ones who know which signal to follow.
Three ways the work leaves the garden: events, consulting, and the founder stories I'm collecting.
Hosting, moderating, and curating AI events — panels, fireside chats, and founder dinners in the Valley.
AI strategy, evals, and agent reliability. From 'what should we build' to 'how do we know it works.'
A living archive of the people building AI — the decisions, the doubts, and the moments before it worked.
Start a conversationAlbums from trails, studios, and things that bloomed.
AI scientist, full-stack builder, and connector in the Silicon Valley AI ecosystem. I've spent years in data science and LLM/agent research, and I care most about one question: how do we know what's actually working?
I write about verification, reliability, and what stays human in the age of agents. This garden is where the signals I find get planted, watered, and occasionally bloom.
Follow @ClorisSignal on XA collaboration, a correction, a good song. Plant it here and I'll water it.