A governed multi-agent AI operating office, built to run a multi-venture portfolio with the same discipline as the pricing and growth work.
Work is organised Portfolio → Strategic Direction → Project → Task — a fixed hierarchy that keeps every active initiative traceable to one owner and one goal. Two AI coordinators operate as peers under a human owner, with complex or high-stakes work escalated to a specialist model rather than handled ad hoc.
An append-only decision register records every material call — fourteen numbered decisions on record since July 2026 — so direction changes are traceable rather than silently overwritten. A read-order operating manual and an escalation ladder keep the system legible to a human owner at any time, not just to the agents running it.
Very few non-technical operators have built anything comparable. It is direct evidence of the same systems-thinking claim made everywhere else on this site — and of real, hands-on AI fluency rather than familiarity with a chat interface.
The architecture is described here; its contents are not. No decision text, credential-vault paths, membership numbers or business addresses are published.