§ Exploration
AI
Practical AI systems, agent workflows, and how to keep them useful without making them brittle.
I'm most interested in AI when it helps real systems move faster without turning them opaque. That means agent tooling, verification loops, and workflows that stay understandable by humans.
The recurring question is not whether a model can do something impressive once, but whether the surrounding system makes that result repeatable, reviewable, and worth trusting.
§ Related builds
BUILD
Dispatcher
A practical orchestration layer for triaging inbound tasks, generating issues, and handing implementation off to the appropriate agent or toolchain.