Advisory

When the initiative is too consequential to improvise.

Organizations bring me in when an AI system will write to systems of record, operate inside regulated workflows, or act where failure carries consequences beyond the enterprise — the seam where complex AI initiatives actually fail, between executive intent and engineering reality.

The governance architecture I design — control planes for agentic systems, plus the operating models I’ve built for AI governance boards and Centers of Excellence — is the same work I implement, not a deck I hand off.

Engagements Three ways to work together

Governance Architecture

Design & stand-up

The structures that make AI adoption defensible at scale: governance boards that operate as clearance systems, Centers of Excellence that operate as transformation engines, and control planes that keep human authority over agentic systems enforceable and auditable.

AI Initiative Leadership

Interim & embedded

Leadership of a consequential AI program with governance, security, and evidence standards built in from the start — and the program still run end to end: strategy, data architecture, governance, delivery. For organizations standing up a VP of AI mandate or a Chief AI Officer charter that need the function built before — or instead of — the full-time hire.

Executive Advisory

Standing counsel

Standing counsel for COOs, division presidents, CIOs, and Chief AI Officers navigating the agentic transition — AI procurement against clear evidence standards, regulatory readiness, and the institutional design of human authority over autonomous action.

Proof of demand

The market already pays for this judgment — in defense, in regulated enterprise, and to U.S. government leadership.

I’ve stood up the governance function for consequential AI mandates inside high-assurance environments — programs covering multi-hundred-person organizations and agency-wide deployment scope — and I advise U.S. government leadership on AI policy and enterprise adoption. The pattern repeats across sectors: a hospital system deploying clinical decision support, a bank operationalizing model risk, a federal program fielding agentic tooling — the governance problem is identical in shape.

The judgment is earned on the build side too: a large-scale NLP and data-science capability stood up from zero inside a nine-figure program, and a data and AI practice scaled from low single-digit millions to $25M with margins moving from ~30% to ~50%. I design oversight the way I do because I’ve carried the P&L on the other side of it.

What I bring is the discipline that travels: decision rights, evidence standards, and a testable line for when a system has earned the authority to act.

How I work Structure first, then speed

Decision rights

Every engagement begins with who decides what, under which constraints, and with what authority to say no — before a line of policy is written.

Evidence standards

What a system must be able to produce — intent, inputs, constraints, an action preview — before it earns authority to act, or it stays advisory.

Testable criteria for done

Governance that reads as a checklist fails. The deliverable is criteria you can actually evaluate against, and that a practitioner can act on.

Governance isn’t a brake on adoption. It’s the architecture that makes adoption defensible at scale — and defensibility is what lets an organization move quickly without betting the enterprise. Proper use, not permission theater.

Have an initiative that can’t afford to be wrong?

Include your organization, the initiative, and where it stands today.

Start a conversation →