Chapter 05 · Phase 2 of 8

Charter the AI COEE

Phase 1 told you where you are. Phase 2 builds the team that will get you where you need to go — and the operating model that keeps governance and execution under one roof.

1 · Assess 2 · Charter AI COEE 3 · Buy-in 4 · Pilot 5 · Deploy 6 · Train 7 · Govern 8 · Scale

The charter is the document that gives the AI COEE its authority. It names the team, defines the scope, sets the cadence, and most importantly, establishes the reporting line to the CEO. Without that final piece, the AI COEE is just another committee.

The charter has six sections

1. Mandate

One paragraph. What does the AI COEE exist to do, and what does it explicitly not do? Mine looks like this:

"The AI Center of Excellence and Execution is the cross-functional team accountable for the responsible adoption, governance, and measurable outcomes of artificial intelligence across [Company]. The AI COEE owns AI policy, vendor selection, integration architecture, workforce training, and adoption measurement. It does not replace functional ownership of specific AI use cases — those remain with the business units. It serves as the central authority, the central capability builder, and the central reporting line for AI."

2. Composition & roles

Name the seats. (See Chapter 3.) For each seat, define: full-time vs. part-time, decision rights, and time commitment. The AI COEE Lead is full-time. Everyone else is typically 20–40% of capacity. Total commitment across the team is usually 2–3 FTEs, not 8.

3. Decision rights

This is where most COE charters fail — they don't say who can decide what. Be explicit:

AI COEE decides: AI policy, vendor approval, tool selection above $25K, integration architecture, training curriculum, audit cadence.
Functional teams decide: Which approved tools to deploy in their workflow, day-to-day usage, configuration.
CEO approves: Annual budget, policy changes that affect candidates or employees, public disclosures.

4. Cadence

Weekly working sessions for the AI COEE Lead and core staff. Bi-weekly full AI COEE meetings. Monthly readouts to executive leadership. Quarterly board-level reports. Annual policy refresh. This is the rhythm. It should be on every member's calendar before Phase 2 ends.

5. Budget & portfolio authority

The AI COEE holds budget for: AI tooling above the per-seat license threshold, training, vendor due diligence, third-party audits, and external consulting. This is what separates the AI COEE from an advisory body. When the recruiting team needs a new sourcing tool, the AI COEE writes the check.

6. Success metrics

The AI COEE itself is measured. Set targets for Year 1 in five categories:

Adoption — % of recruiters actively using approved AI tools (target: 80%+ within 12 months)
Productivity — measurable hours saved per recruiter per week (target: 5+)
Quality — time-to-fill and offer acceptance rate vs. baseline
Risk — zero compliance incidents; 100% of candidate-facing tools bias-audited annually
Capability — % of recruiters who complete AI literacy certification (target: 100% within 6 months)

From Jonathan's Playbook

The CEO reporting line is non-negotiable

In every enterprise rollout I've watched succeed, the AI program leader had a direct line to the CEO. In every one I watched stall, the program leader reported up through a function head whose own priorities competed with the AI program. At Wells Fargo and Kaiser, the equivalent programs lived inside business technology — and even then, the successful ones had a steering committee with C-level air cover. Don't compromise on this. If your CEO won't sponsor the AI COEE, you're not ready to scale AI.

The IT partnership is half the AI COEE

One of the most common failure modes is treating AI as an "HR thing" or a "recruiting thing." It isn't. AI cuts across identity management, data security, licensing, integration architecture, and incident response — all of which sit in IT. The AI COEE charter has to name IT as a co-owner from Day 1.

What the IT seat owns inside the AI COEE:

Single-sign-on and identity for all AI tools
Data residency and privacy posture (especially for tools that touch candidate data)
Integration architecture — how AI tools connect to ATS, CRM, HRIS, calendar, email
License consolidation and seat utilization audits
Tier-1 tech support for end users
Vendor security review

I don't know if CTOs are listening to this conversation, but they need to be aware of it — because in many instances, technology is being purchased that the CTO is left out of the conversation, and it's driven by the heads of recruiting.

Brian Fink · Managing Partner, The Rework Group Listen →

The first deliverable: a one-page operating model

Before the AI COEE moves on to Phase 3, it should publish a one-page document — for the org, not just the executive team. The page answers six questions:

Who is on the AI COEE? (Names, photos, faces — make it real.)
What does the AI COEE do?
How do I submit an AI use case for AI COEE review?
How do I get trained?
How do I report a concern or an incident?
Where do I find the current list of approved AI tools?

Publish it on the intranet. Pin it in Slack or Teams. Print it. The visibility is the message: this team is real, it has authority, and it's the place to go.

Watch out

If you build the AI COEE without authority, it dies inside 90 days

The first time someone bypasses the AI COEE with no consequences, the program is over. The CEO has to back the AI COEE publicly the first time someone tries to deploy a tool outside the process. That moment is the test. Plan for it.

Phase 2 is short — usually 2–3 weeks. The charter is approved, the team is named, the operating model is published. Now the AI COEE goes to the executive team to lock in the business case.