Recent Writings
-
Hi all, Today I’m sharing a new working paper, Human Review Is Not a Safeguard. It looks at why an approved or flagged AI output doesn’t show that the system behind it was governed, and proposes a four-plane framework for checking where enforcement actually operates. It’s an exploratory paper, not a validated methodology, so I’d…
-
At ServiceNow Knowledge 2026, the focus was on the need for clean data to ensure effective AI governance. The new RaptorDB and AI Control Tower were introduced, but they rely on pre-existing data quality. StrataLayer addresses this gap by providing essential data governance before AI systems process information. The importance of structured data was emphasized,…
-
30-Second Perspectives A 30-Second Perspective on pre-processing governance and the gap at the bottom of the enterprise AI stack. At Knowledge 2026 this week, ServiceNow is demonstrating what a mature AI governance architecture looks like inside the platform. Four enforcement planes, clearly defined. Model governance, agentic execution, identity and access, policy and accountability, each mapped…
-
The ServiceNow Australia release shifts AI governance from policy to enforcement — blocking unapproved agents at the infrastructure layer, not flagging them after the fact. This decision-grade synthesis covers what materially changes from Zurich, seven critical gaps that remain, and a phased roadmap structured by role for CISOs, architects, AI engineers, and TPMs. GA target:…
