If you run finance, operations or the whole company for several clients a few days a month, you have had this conversation at least once this quarter. An owner leans across the table and says “we should be doing something with AI”, and looks at you, because you are the most senior person in the room who is not them.
You did not sign up to be the AI expert. Most fractional executives I talk to are honest about that. They have deep experience in their function and a good read on their client’s business, and they are being asked to have an opinion on a technology that changed twice since they last looked.
This page is about what your clients are actually asking, the answer you can give them this week, and how to bring in someone who has done the work, without handing over the relationship.
What your client is actually asking
“What should we do about AI” is three plainer questions, and none of them needs you to understand how the technology works.
- Where would this save us real money or time? Not “transform the business”. Hours a week, on a named task, with a number beside it.
- What could go wrong? A customer’s data pasted into a free tool. A wrong figure nobody caught for a month. A tool someone in accounts built that the whole company now relies on and nobody else understands.
- Who owns it? They are hoping the answer is not a new hire.
You can answer all three for a client in a month, because they are questions about the business, and you know the business. What you may not have is the second half of each answer: what the tools can actually do, what breaks, what a professional requires before a business relies on software, and what a vendor is leaving out of the demo. That half is a specialty. It is mine.
What I actually do
Most people you could bring in for AI are advisers. They hand your client a deck and leave. I am not that, and the difference matters for the kind of client you have.
I build the thing. Since 2021 I have built and run the systems small businesses run on: CRMs, ticketing, client portals, quoting and invoicing, inventory, booking, and the AI features inside them. Most of it built with AI tools, all of it kept running afterwards. So when the answer to your client’s question is “build it”, the person who gave the answer is the person who can. And the advice comes from someone who has to live with the consequences.
I can look at what your client already has and tell you if it holds up. The most common thing a fractional CFO or COO inherits is not a blank page. It is a tool someone in operations built over a weekend, a vendor’s “AI feature” nobody evaluated, a spreadsheet that became a system. Send me the thing. I read the code, run the security pass, and tell you what would break and what I would require before the business relies on it. In writing, in plain English, under your cover.
One answer you can use at every client. You do not want five bespoke engagements. The Ready Test, the one-page AI policy, the six questions before relying on an app and the two-week payoff assessment are the same at every client you have. They can be part of your standard offer, with me behind them.
No commissions, no affiliations. I take nothing from any software vendor or platform, and I am not tied to one. When I recommend a tool, or tell your client not to buy one, that is the whole reason.
Your language, and theirs. I spent most of my career in sales, management and operations before I wrote a line of code. You can put me in front of an owner without translating, and I will not embarrass you with jargon or a pitch.
What you can bring me in for
Every one of these has a written result your client can read without a translator, and a page on this site that shows the method before you ask.
Bounded pieces. Fixed price, days not months.
- Accounts, access and what happens when someone leaves. The most basic review and the one most often failed. Which AI platforms and tools the business is actually on, who holds the admin on each account, whose card it bills to, where the keys and passwords live, and who else could get in tomorrow if that person did not come back. If the answer is one name, that is the finding, and fixing it is usually an afternoon. See where it runs and who can get in.
- AI use review and blind-spot report. What staff are actually doing with AI versus what management thinks, which data is going where, what nobody approved. The gap is usually bigger than the owner expects. See your staff already has an AI plan and the twelve questions.
- Review of something the business already relies on. A staffer’s weekend app, a spreadsheet that became a system, a vendor’s AI feature. I read it, run the six questions and the infrastructure checklist, and give a verdict: rely on it, fix these first, or do not.
- Security pass on the platforms they chose. A static review of who can see what, what the system will fetch on a user’s behalf, where the keys live, and what happens when a setting is missing. Scheduled scanning set up to keep running after I leave. See the security pass that runs anyway.
- Vendor second opinion. Before your client signs for an AI feature or platform, I ask the vendor the seven questions on your behalf and write up what the answers mean. I take nothing from any vendor, which is the point of asking me. If the client is hiring a developer or agency to build something rather than buying a product, the questions are different, and they are here.
- The one-page AI policy and a half-day briefing. What staff may use, what never gets pasted in, who to ask. Written to be followed, briefed in person. Increasingly required of small suppliers by their larger customers. See the playbook.
- Where AI pays off, in two weeks. Map the workflows, score them, hand back a ranked list with a number beside each. Four hours of the owner’s time. The right first engagement when the client’s question is vague. See the method.
Ongoing. A few days a month, alongside you.
- AI strategy and roadmap. What a plan contains: five short things with owners, in the order to do them, and the rules while you do.
- The AI lead seat. The AI side owned: roadmap, rules, review of anything built, vendors vetted, and a leadership briefing when it is on the agenda.
- Is it still working. A monthly pass over the apps and automations the business relies on: checks against real data, errors tagged by version, the security scans, a backup someone has actually restored from. See the app is the visible part. Nobody else offers this to a firm this size.
Continuity. The one nobody asks for until it is too late.
- Vault and handover setup. A knowledge base the AI tools read every session, the coding standard as a file in the repo, and a handoff pack, so that when the person doing the AI work leaves, the next one picks up where they left off. See what a company keeps when the coder leaves and MyVault.
- The cold-pickup drill. Once a quarter, someone who did not build it runs it from the written material. It is the only proof the handover works.
Building.
- Build it. A prototype in a week, the workflow automation, the AI feature inside an existing tool, or the move off a no-code platform that hit its ceiling. Most engagements do not start here. It matters that the person advising your client can do it.
The three shapes this takes
You do not have to become the AI person. You have to have one. In practice it takes one of three shapes, and clients move between them.
A call before the meeting. You have a board meeting on Thursday and AI is on the agenda. An hour with me beforehand and you walk in with the three questions above answered for that client, the four things a plan contains, and a clear view of which vendor claims to push back on. You present it. I am not in the room unless you want me there.
A specialist, per topic. The client wants a policy for what staff may paste into AI tools. Or someone built a quoting tool over a weekend and the client wants to know if it is safe to rely on. Or a vendor is selling an “AI feature” and the client wants a second opinion before signing. Each of those is a bounded piece of work with a fixed price and a written result you hand to your client under your own cover.
An AI lead alongside you. Some clients are past the first conversation and need the AI side owned: a roadmap, the rules, the review of anything built, the monthly check that it is all still working. That is a fractional CIO seat, filled the same way you fill yours, a few days a month, sitting alongside you rather than replacing anyone.
What I bring, and what I do not
I have sat with hundreds of small businesses working out what they actually needed, and then built it. What I do not do is compete for your seat. Three commitments:
- The relationship stays yours. I work through you. Your client hears from me when you decide they should.
- Nothing gets pitched behind you. No software, no retainer, no “while I’m here” to your client that you have not heard first.
- Your client’s language, not mine. Every written result is meant to be forwarded. If it needs a translator, I have failed.
What to send your client today
You can make the next conversation concrete before we have spoken.
- The Ready Test. Three minutes, ten yes-or-no questions, and it tells the owner which of four steps their business is on. Send it with “take this before Thursday”.
- Is your business ready for AI? The five signs of a business that is ready, in plain English, and the twelve questions a leadership team can answer in one meeting.
- A chatbot for everyone is not a strategy. For the client who bought licences last year and is wondering why nothing changed.
- The employee who can now build apps. For the client whose ops manager built something the company now depends on.
How it starts
Tell me which client and which of the three questions is loudest. Say hello. A first conversation is a conversation, not a proposal, and if the honest answer is that your client does not need me yet, I will say so and point you at the pages above.