Why a client studio builds its own products
Our internal products are not side projects — they are how we test our research honestly before we ask a client to bet on it.
Lab Notes
Notes from the lab floor
Development logs, research explanations, experiments that worked, approaches that failed, and observations from building AI products. Less polished than the research section, on purpose.
Notifications are the wrong abstraction for agent work
Why the notification — the atomic unit of every alerting system — fails as an interface primitive once AI agents work for hours at a time, and what we're building instead.