Someone owns AI here. Month after month.
John joins your leadership rhythm as the accountable AI lead: setting priorities, deciding on tools and vendors, governing use, overseeing delivery, developing the team, and reporting what actually changed each month. He is an outside advisor, not an officer or employee of your company.
The role only works if it can decide.
Every failed version of this engagement failed the same way: a knowledgeable outsider, a monthly meeting, and no authority to stop anything.
A title without authority is decoration.
Bring someone in if
- AI decisions are being made across the organization and nobody is accountable for the whole picture.
- You will give the role real decision authority over tools, vendors, and what gets built.
- There is more than one initiative running, or more than one team spending on it.
- Leadership wants a short, honest monthly read rather than a dashboard nobody trusts.
- You want someone who has built the systems, not only someone who has written about them.
Do not bring someone in if
- You want a name on an org chart for a deck or a funding round. A title without authority produces nothing and everyone can tell.
- The role would advise but never decide. If every recommendation has to survive a committee that never meets, this is theatre.
- You have one workflow and one problem. That is a build sprint, and it is cheaper.
- You are looking for someone to sign off on legal, security, or compliance in your place. That responsibility does not transfer, and it is not offered here.
If the problem is one workflow rather than a portfolio, the build sprint is the cheaper answer. The delivery standard the role holds people to is written up under how we work. Public institutions have their own authority, procurement, and review path; that work is scoped as government and public sector.
What a month actually contains.
The same loop every month. It is deliberately boring, because the value is in it happening again rather than in any single session.
Portfolio review
What is running, what is in build, what is waiting, and what was declined. Nothing is allowed to sit in the portfolio without a reason for being there and a named owner.
Decisions: tools, vendors, policy
The decisions that otherwise get made by whoever bought a licence first. What is approved for which work, where each tool may touch data, which vendors are worth a pilot and which are not, and what the policy says when someone asks.
Delivery oversight
Whoever is building, internal team or outside firm, the work is reviewed against an agreed bar before it counts as done. Evaluation on real examples, a named human review path, and documentation are the conditions, not the extras.
Team enablement
The people who will actually operate these systems get trained on the ones you have, not on general AI literacy. Role by role, on the real work, so capability does not leave when the engagement does.
Scorecard to leadership
A short monthly read: what changed, what is working, what was stopped, what it cost, and what the next decision is. Written to be read in five minutes by people who were not in any of the meetings.
Then it repeats
The value is in the loop, not in any one month. Decisions get revisited when the evidence changes, and what was declined in March gets reopened in September if the reason for declining it went away.
What stays with you.
All of it is yours throughout, not handed over at the end. An engagement that cannot be ended cleanly was scoped wrong.
| Deliverable | What it is | Who owns it |
|---|---|---|
| A living AI portfolio | What is running, what is next, what was declined and why, with a named owner against each line. | You. It survives the engagement. |
| Policy and review paths | What is approved for what work, where the human checkpoints are, and who decides when something is unclear. | Your responsible people, who sign it. |
| Vendor and tool decisions | What was evaluated, what was chosen, what was rejected, and the reasoning behind each one. | You, for the next renewal conversation. |
| Delivery oversight | An agreed bar applied to internal teams and outside builders alike, with the evidence recorded. | The teams doing the building. |
| A monthly scorecard | A short written read of what changed, what it cost, and the next decision in front of you. | Leadership. |
The monthly read, in one page.
Leadership should be able to tell in five minutes whether the month moved. The rows below are invented to show the shape of the report, not to describe anyone’s organization.
| Measure | This month | Note |
|---|---|---|
| Workflows in operation | 4 | One added this month, one retired. |
| In build or evaluation | 2 | Both waiting on the evaluation gate. |
| Declined | 1 | Vendor could not meet the agreed data boundary. |
| Human review path | Cleared same day | Two items escalated to the named owner. |
| Policy changes | 1 | Approved-use list updated for the finance team. |
| Next decision | Renew or retire | One tool comes up for renewal next month. |
Illustrative scorecard. Not a client record.
Boundaries. Read this one properly.
The title is the part of this engagement most likely to be misread, so it is spelled out.
- Not an officer
“Fractional Chief AI Officer” is the name of an engagement, not a corporate office. John is an outside advisor: not an officer, director, or employee of your company, holding no delegated legal authority and signing nothing on your behalf.
- Sign-off stays with you
Legal, security, privacy, and compliance sign-off stays with your own responsible people. Soxoa does not provide legal advice and does not certify anything against a regulation or a standard.
- Outcomes
No outcome is guaranteed. The engagement guarantees a named person accountable for the portfolio, decisions made and written down, and a monthly report you can hold someone to.
- Scope and price
Scope and price are agreed before work begins, including the monthly cadence, what the role decides, and what it only recommends.
- Your data
Data handling and access are agreed before any real records are shared, and reviewed again whenever a new tool or vendor enters the portfolio.
- Ending it
The portfolio, the policy, the decision record, and the scorecards are yours throughout. An engagement that cannot be ended cleanly was scoped wrong.
Questions leadership asks first.
Are you an officer of our company?
No. “Fractional Chief AI Officer” is a service name. John is an outside advisor: not an officer, director, or employee, with no delegated legal authority, and he signs nothing on your behalf. Legal, security, and compliance sign-off stays with your own responsible people.
Does this replace our CTO, CIO, or head of data?
No, and it works badly when it is framed that way. The role owns the AI portfolio and the decisions inside it, and reports alongside your existing leaders rather than over them.
How much time is it each month?
A standing rhythm agreed before the engagement starts: a set of recurring sessions plus availability between them. The cadence is part of the scope conversation, because it depends on how many initiatives are running and how often decisions are actually needed.
What authority does the role actually need?
Enough to decide what gets built, what tools are approved for which work, and what gets stopped. Without that, the engagement produces recommendations that sit in a folder, and you should not buy it.
What if the right answer is to do less?
Then that is the recommendation, and it goes in the scorecard with the reasoning. Retiring a tool nobody uses and declining a vendor that cannot meet your data boundary are both normal months.
Can this follow a sprint or a workshop?
Often it does. A workshop or a sprint answers one question well; the fractional engagement exists because the questions keep arriving. It also works the other way: the role frequently scopes sprints and hands them to your team or to Soxoa.
What does it cost?
Scope and price are agreed before work begins. The scope conversation covers cadence, decision authority, and what the role owns, and the price follows from that.
Where this sheet sits in the set.
Each engagement stands on its own and each one bears on the next. Start where your organization actually is, not where the ladder starts.
- S-01AI Readiness Assessment
“Where are we, honestly?”
Fifteen questions, five dimensions, one recommended next step.
- S-02Live Build Workshop
“Can my team actually do this?”
One day. One working automation, built live on your own work.
- S-03AI Build Sprint
“Get the important one into production.”
A fixed-scope sprint that takes one to three workflows into operation.
- S-04Fractional Chief AI Officer
“Who owns AI here, month after month?”
A named, accountable AI lead on a monthly engagement.
You are on this sheet
Decide who owns this before the next tool arrives.
Thirty minutes with John. Describe who is making AI decisions today and what authority the role would really have, and you will get a straight answer about whether a fractional engagement fits or whether you need something smaller first.