Introducing the AI COEE

The AI Center of Excellence is a familiar concept. Most large enterprises have one, or are building one. The problem is that traditional Centers of Excellence are advisory: they publish guidance, they review proposals, they get cc'd on rollouts. They don't actually execute.

The AI COEE — Center of Excellence and Execution — is the construct I've been developing to fix that gap. It's not just a policy body. It's the operating unit that holds the AI program together end-to-end: governance, training, vendor management, integration, measurement, and continuous improvement. It reports to the CEO, not buried two levels deep in HR or IT, because AI cuts across both — and because the consequences of getting it wrong (compliance exposure, candidate experience damage, productivity loss) sit at the CEO level.

This chapter explains what the AI COEE is, who staffs it, and what it produces. The rest of the playbook is the playbook the AI COEE actually runs.

What the AI COEE is

Who staffs the AI COEE

The AI COEE is a small, senior, cross-functional team. Not a department of 30. A working group of 5–9 people with executive-level authority and dedicated capacity. The composition I recommend:

• AI COEE Lead — full-time, reports to CEO. The single accountable owner. Typically someone with enterprise tech program experience.
• VP/Director, IT or CTO designee — owns integrations, security, licensing.
• VP/Director, HR or CHRO designee — owns employee-facing policy and change management.
• Head of Recruiting or VP TA — owns the TA use cases, KPIs, and rollout sequence.
• Legal / Compliance — owns vendor due diligence, candidate disclosures, AI Act readiness.
• Frontline operator — a recruiter or HR business partner who actually uses the tools. Rotates every 12 months.
• Data / Analytics partner — owns the measurement framework.
• Communications partner — owns the narrative, internal change comms, and CEO updates.

What the AI COEE produces

The AI COEE is judged on outputs, not meetings. The core deliverables on a 12-month cycle:

Quarterly

• CEO-level adoption & ROI dashboard (one page, five metrics)
• Vendor portfolio review (active contracts, renewals, retirements)
• One AI hackathon (cross-functional, problem-led)
• Risk & incident report (anything candidate-facing or compliance-relevant)

Semi-annually

• Employee AI utilization survey (org-wide, anonymous)
• AI literacy curriculum refresh
• Bias audit on all candidate-facing screening tools

Annually

• AI policy refresh (principle-based, not tool-specific)
• Workshop / training calendar for the next year
• Compliance posture review (NYC, CO, EU, applicable state laws)
• Board-level state-of-AI presentation

Why "and Execution" matters

The most common failure mode of a traditional Center of Excellence is that it becomes a governance bottleneck without owning execution. Teams either route around it (shadow IT) or get stuck waiting for approvals. The AI COEE solves this by holding the budget, the licenses, and the integration partnerships itself. When a recruiting team needs a new AI sourcing tool, they don't write a memo and wait — they bring the use case to the AI COEE intake, and the AI COEE either ships it or runs the pilot inside its own portfolio.

That changes the dynamic. The AI COEE becomes the place teams want to go, not the place they avoid. And because it has CEO-level air cover, it can make decisions at speed.

The AI COEE is what gets the rest of the chassis underneath the engine. Without it, even the best individual AI deployments will hit a ceiling. With it, you have the scaffolding to scale.

How the AI COEE shows up in the rest of this book

Every phase from here forward names the AI COEE explicitly. Phase 1 (Assess) is the AI COEE's first deliverable. Phase 2 is the AI COEE charter itself. Phases 3 through 8 are the AI COEE's operating cadence. Think of this book as the AI COEE's first-year operating plan, told in eight chapters.

Let's start with what the AI COEE does on Day 1.