Skip to content

Notebook · 04 entries · last updated 2026-07-09

Read this first

Four self-directed studies in how AI should behave when real organisations depend on it.

Each works a problem end to end: where the system earns trust, where a person stays in the loop, and what it must refuse to do. Studies rather than shipped products — the reasoning is the artefact.

01 — The entries

Independent studies · not client or production work

01Working note6–8 min read

AI Chief of Staff for Founder-Led Teams

Decision continuity for scaling founder judgement.

  • Decision continuity
  • Founder-led teams
  • Restraint
Read the noteUpdated 2026-05-18
02Design study8–10 min read

AI Operations Copilot

Incident reasoning for high-pressure operational teams.

  • Human-in-the-loop
  • Incident response
  • Bounded learning
Read the noteUpdated 2026-06-02
03Design study10–12 min read

AI Risk & Decision Intelligence System

Surfacing commercial risk before it becomes operational reality.

  • Probabilistic outputs
  • Governance thresholds
  • Calibration
Read the noteUpdated 2026-06-21
04Position note10–12 min read

When AI Becomes Labour, Not Software

Designing AI systems that operate as accountable digital workers.

  • Category shift
  • Labour economics
  • Compliance as product
Read the noteUpdated 2026-07-09