A dated snapshot of what I’m building, thinking about, writing and still trying to understand.
Last updated 31 Oct 2026 from Sydney, Australia.
I’m exploring how Bob, my AI collaborator, could become part of the space around me: a small physical presence on the desk, or a spatial interface in Apple Vision Pro. I’m working through how it should feel to talk, listen and share attention, with clear privacy cues and an easy way to stop. It’s early design work; the experience is still to be built and tested.
I’m pressure-testing a simple idea: a clear outcome, a small harness and good evidence can be more useful than an elaborate agent team. Complexity still earns its place when responsibilities, context or controls genuinely demand it.
I’m exploring what one agent should require before it accepts another agent’s identity, delegated authority or result. The current work is research, trust-boundary models and prototypes—not a claim that the problem is solved.
I’m turning a growing pile of experiments, notes and working positions into a clearer public record. The current threads are hybrid organisations, making better use of the AI I already pay for, and keeping AI work grounded in relevant context and evidence.
Projects, experiments and the evidence that changes my mind.
Visit the workbenchFamily, retro consoles, gaming and the occasional track out of town.
A bit more about me