Writing
Notes for the executive conversation: production AI, governance, and what a board can actually audit.
Supervisors Are Now Asking Who Owns the System
October 2026 guidance from the FCA, MAS, and HKMA is not a new technology story. It is a test of accountability before an AI system goes live.
On 19 October 2026 the FCA published good and poor practice for AI in financial services. The failure it describes is familiar: governance fitted on after the design was already chosen. The same month, supervisors in Singapore and Hong Kong asked for a named owner.
Read essay →Why the AI Roadmap Never Reaches Production
The handoff between the decision and the system — and the one-team standard that closes it
The roadmap gets approved. The agent does not reach audited production. The failure is the handoff after the decision, not a shortage of advice.
Read essay →Agent Observability: What Boards and Regulators Actually Need
Logs, traces, and evaluations aren't optional for regulated AI — they're how you prove the agent is safe to operate
Boards don't ask whether you have an AI strategy. They ask what the agent did yesterday, whether it was correct, and who is accountable when it isn't. A practical framework for making agent behavior auditable.
Read essay →The Operator Model for Enterprise AI
What a leadership team should be able to demand before an agent is called production
Enterprise AI programs stall between an approved strategy and a system a board can audit. The Operator Model is the standard for closing that gap: one owner, evidence before scale, and a system someone can explain.
Read essay →