Two years ago, a client hiring me knew what they were getting, and I knew what I was delivering, and neither of us needed to discuss how the work got done. 

That is no longer true. AI is now doing part of the work, and nobody named it at the start. Last week I spoke with another fractional CFO about how our practices have changed, and neither of us could point to a standard for handling it; she had seen exactly one client agreement that mentioned AI at all.

I sit on both sides of this. Companies hire me as a consultant, and as a fractional CFO, I hire consultants.

Today I want to walk through the four aspects that need to be discussed upfront.

The subscriber section has my own answers to all four: how I describe my method to clients, where my data line sits, and what transfers at the end of four standard deliverables, plus a downloadable checklist for your next hiring conversation.

Consulting has always rested on a division nobody had to state: the client buys the output, the consultant keeps the method. That held because the method was not transferable, since you cannot hand over the fifteen years that taught you which line in a variance report to question first.

AI moved into the middle of that division. Part of the work is now done by a tool; the method it follows is written down rather than held in someone's head, and the result can be copied and handed to someone with none of the expertise behind it.

Everyone uses AI now, including the consultant you are about to hire. And almost nobody discusses how.

What matters is specific: which tools, on which parts, where their judgment enters, and how they check the output. A three-statement model drafted by a tool and reviewed by a fifteen-year CFO is a good product. The same model skimmed by anyone is not, and both arrive looking identical.

There is also no shared toolset here: one consultant works in Claude, another in ChatGPT, a third in whatever their last firm paid for, and each handles your data differently.

If you deliver: describe your method before you are asked, have it written down, and if you used AI to build the model, say so.

If you hire: ask what they use, on which parts, and how they verify it. Then say which tools you are fine with, because permission is more useful to them than silence.

A tool is doing part of the work, so there is now a second thing that can change hands, and most engagements never put it on the table.

Say you hire someone to build a board pack. You can ask for the pack, or for the pack plus the project that produces it, with the metric definitions and narrative framing included, so your team can run it themselves every month afterward. Different engagements at different prices, and sometimes the method is much more useful than the deliverable itself.

If you deliver: define what the engagement produces in components. The output goes to the client; the project that reruns it is negotiable and priced; the skill behind it stays with me. Your answers can differ; the position to avoid is a client asking something you have not thought about.

If you hire: name the components in the scope, because if you ask for a model you only get a model.

The confidentiality clause says client information stays confidential and says nothing about whether putting it into an AI tool breaks that.

Some companies allow business tier accounts only, others accept a professional account with data training switched off, and others keep certain information out of AI entirely.

If you deliver: know your own line before a client asks, meaning what you anonymize, what never goes in, and which account the work happens in. Have your AI use policy ready.

If you hire: send your AI use policy to your consultants, ask legal to update the agreements, and put the data handling rules inside them. Be specific about the account tier as well as the data type.

Then settle whose account produces the work. I would never ask a consultant to use my company's Excel license, but if they are building a financial model for me, it makes sense for them to build it in my Claude account rather than theirs.

A consultant delivering a project keeps the method, which still holds.

A fractional CFO on retainer is not delivering a project; they are building capability inside a company, and what they build stays. If I document a close process and train an analyst on variance commentary, all of it belongs to the company.

AI complicates it because the capability now includes the tools the team uses and the instructions that run on them. So what happens when the thing I built is a skill they run every month-end?

By retainer logic, it is theirs; by consulting logic, it is my method, and there is no default to fall back on.

If you deliver: know which engagement you are in before you build anything reusable, and have the offboarding conversation at the start.

If you hire: AI belongs in every function build and every function upgrade. Leave it undiscussed and you get an outdated function that needs rebuilding.

None of these have settled answers, and no professional body is working on one, so the rules get made engagement by engagement, by whoever raises the subject.

That is rarely the client, since most do not know the questions exist, and rarely the consultant, because raising it means volunteering a conversation that could cost you the work.

So it mostly does not happen, and both sides find out where they disagree in the last week.

The section above is the questions; below is how I answer them in my own practice, and my answers are not a standard.

What I tell clients about how I work, where my data line sits, and a component-by-component breakdown of what transfers at the end of four standard deliverables, with a downloadable checklist.

If you are not yet a subscriber, this is a good edition to upgrade on.

None of this is settled, and I do not expect it to be for a while. But these questions get asked eventually, and the difference between asking them in the first week and the last week is considerable. I hope this helped you think it through.

See you next Tuesday!

This newsletter is for you, and we want to make it as valuable as possible. Please reply to this email with your questions, comments, or topics you'd like to see covered in future issues. Your input shapes our content!

Want to dive deeper into balanced AI adoption for your finance team? Or do you want to hire an AI-powered CFO? Book a consultation!

Did you find this newsletter helpful? Forward it to a colleague who might benefit!

Until next Tuesday, keep balancing!

Anna Tiomina
AI-Powered CFO