Articles
Clawblins gives the system personality. Underneath are specialised roles, curated context, deterministic controls and human authority. These articles examine how that architecture works in practice, with UnHodl as one consequential proving ground.
Context and model routing
The next AI outage may look like a passport check.
Consider which model should see which context—and how deliberate routing can reduce dependency without pretending every task is interchangeable.
4 min read · Read article → · Published
Personal AI in practice
A real day with Clawblins, my AI crew.
Think through where a personal AI crew helps, where it creates friction, and what the ordinary working day reveals about role design.
7 min read · Read article → · Published
Roles and trust
I do not want more AI agents. I want a crew I can trust.
A practical case for smaller roles, curated knowledge, visible boundaries, and a human who remains responsible for the decision.
8 min read · Read article → · Published