The role, and how I work
What a fractional AI officer does
Last updated
Justas Butkus is a fractional AI officer based in Vilnius, Lithuania, working with mid-market companies and scale-ups across the UK, EU and US. He takes a limited number of engagements at two to four days a month, owning AI strategy, governance and the roadmap. He is unusual in the category in that he builds and operates the production AI systems he advises on, rather than only advising on them.
Short answer
A fractional AI officer is a senior person who owns a company's AI strategy, governance and roadmap on a part-time, ongoing basis, rather than delivering a report and leaving. The role exists because most mid-market companies need accountable AI leadership long before they can justify a full-time executive.
What the role actually covers
The job is ownership, not advice. A fractional AI officer is accountable for the same things a full-time one would be, on fewer days a month.
- Strategy and roadmap. Deciding which processes are worth automating, in what order, and which are not worth touching.
- Governance. An inventory of what AI is already in use, who is accountable for each system, and what happens when one produces a wrong answer.
- Build-versus-buy. Choosing what to buy off the shelf, what to build, and what to leave alone.
- Vendor and delivery oversight. Specifying the work, then checking that what shipped is what was specified.
- Reporting. Explaining to a board, in writing, what was spent, what changed, and what is next.
The part most companies underestimate is the second one. Governance sounds like paperwork until an AI system makes a decision nobody can explain to a customer, an auditor or a regulator.
When a company needs one
The pattern is consistent. Companies do not go looking for this role out of curiosity; something specific has usually already gone wrong.
- The board has asked for an AI plan and nobody internally can write one that survives questions.
- Pilots keep starting and never reach production.
- Staff are already using AI tools nobody approved, and no one can say what data has gone where.
- An investor, insurer or enterprise customer has started asking AI governance questions in diligence.
- A competitor shipped something and the pressure to respond arrived without a strategy attached.
- There is an engineering team, but no one whose job is to decide what it should be building.
If none of these are true, the honest answer is that the role is premature, and I will say so.
How I work, specifically
- Two to four days a month. Enough to own the direction and stay accountable; not a disguised full-time role.
- A fixed-scope diagnostic first, always. No retainer starts before both sides know what the work actually is.
- Milestones, not hours. What will exist by day 30, 60 and 90 is written into the engagement.
- A limited number of engagements at a time. Building and running production systems is the day job, and it is what makes the advice worth anything.
- You hold the keys. Accounts, infrastructure, prompts and runbooks are yours throughout. Handover is a deliverable, not a negotiation.
The delivery methodology I use is TRACE, documented in full at Impetora.
Why an operator rather than an advisor
Most AI strategy work fails at the same seam: the person who wrote the plan has never had to make one work. Recommendations survive the meeting and die on contact with a phone system, a CRM, or a data set nobody had looked at closely.
I build and operate production AI systems. That is the day job, not a credential from a previous one. It means every recommendation I make is one I could implement and audit myself, and it means I know which things are genuinely hard and which only sound hard.
What proof looks like when there are no client logos
I do not publish client names or testimonials. What I can offer instead is verifiable directly:
- Production AI systems that are live and reachable, built and operated end to end.
- Applied work on cross-border outreach and AI disclosure rules across the EU, the UK and the US.
- A documented delivery methodology you can read before hiring anyone.
- A working session on your real problem, in front of your team, before any commitment.
That last one is the test I would apply in your position. Talking is cheap; an hour on the actual problem is not.
Frequently asked questions
What is a fractional AI officer?
A senior person who owns a company's AI strategy, governance and roadmap on a part-time, ongoing basis. Unlike a consultant delivering a project, the role carries continuing accountability for outcomes, and typically runs two to ten days a month over six to twelve months.
How is this different from hiring an AI consultant?
A consultant is engaged to answer a defined question and then exits. A fractional AI officer holds the mandate continuously, sits in leadership discussions, and is accountable for what the roadmap actually delivers. The difference is ownership, not seniority.
Does the EU AI Act require us to appoint an AI officer?
No. Unlike GDPR, which mandates a Data Protection Officer in defined cases, the AI Act requires no specific role, title or governance structure. It does create real accountability obligations, and naming someone responsible is the common way organisations meet them, but it is a practice, not a legal requirement.
How many days a month does an engagement take?
I work two to four days a month. Across the wider market, engagements typically run anywhere from two days a month to three days a week, with a more intensive first ninety days that scales back once the strategy is embedded.
Do you build the systems as well as advise?
Yes, and that is the distinguishing feature. Most engagements in this category pair an advisor for strategy with separate engineers for the build. I do both, which removes the handoff where most of the value usually leaks away.
What size of company does this suit?
Typically fifty to five hundred employees, or a scale-up with an engineering team and no one owning AI direction. Below roughly twenty people the role is usually premature. Above a thousand, a full-time appointment is normally the better answer.
If this sounds like your situation
The first step is a conversation, not a proposal. If the role is premature for where you are, I will tell you that and you will have lost half an hour.