A fractional Chief AI Officer who becomes permanent has failed, however good the invoices look. The mandate should be designed to end.
Fractional executive roles have a structural temptation. The work is genuine, the relationship becomes comfortable, and after a year nobody remembers what the engagement was supposed to achieve. The retainer continues because stopping it would be awkward. That is a bad outcome for the client and, less obviously, a bad outcome for the advisor: the moment your income depends on staying, your judgment is compromised in a way you will not notice.
Four things a good mandate has
- An exit date, written on day one. Six to nine months, with an explicit description of what the organization must be able to do without me by then.
- A stop mandate. The standing authority to recommend killing an initiative, exercised in front of the people who approved it. If that authority is not granted, the role is decorative.
- Presence where decisions are made. Not a parallel workshop track. The investment committee, the board session, the meeting where the vendor contract gets signed.
- A decision record, not a roadmap. Two or three pages a month: what was decided, by whom, against which number, and what would make you reverse it. Roadmaps age badly. Decision records get reread.
What the exit actually requires
Handing over is not a document. It is three capabilities that have to exist in named people before I go: someone who can run the gate between proof of concept, pilot and production without me in the room; someone in finance who can read the cost of an AI system at volume rather than at pilot scale; and a governance routine that survives a busy quarter. If those three are not in place, the engagement should be extended honestly rather than drifting into permanence.
The uncomfortable part
Designing the mandate this way costs money. A well-run exit is revenue you deliberately do not earn. But it is also the only version of the role that a CEO can defend to a board without embarrassment, and it produces the referral that pays for the next two mandates. Independence is not a virtue claim here. It is the asset.
More on how the mandate is structured on the advisory page.