AI Coaching or a Fractional CAIO? Start With the Work You Need Done

AI coaching fits when you want to get better at doing your own work with AI. A fractional chief AI officer fits when your business needs someone to help choose priorities, shape a roadmap, and coordinate AI decisions with your leadership team. Start with the task that keeps getting stuck.

Maybe you keep opening Claude, asking a question, and getting an answer you still have to turn into work. Or maybe people across the company are already trying AI, and you're the one deciding which ideas deserve time and money. Those situations need different kinds of help.

I offer both. Here is how I would sort them out with you.

What each kind of help is for

	AI fluency coaching	Fractional AI leadership
Focus	Your skill with AI	Business AI decisions
Work	Real tasks and practice	Priorities, roadmap, vendors
Your part	Practice and check results	Bring context and make decisions
Progress	More work you can do yourself	Clear priorities and ownership

A company can need both. Someone still has to decide which problem matters, and someone has to learn how to do the work. We can discuss that combination without turning every request for help into a larger engagement.

Coaching starts with something you want to do

Jarom Watts, a sales director at Easy To Use Products, had never used a terminal when we started. After our second session, he was still partway through the developer application for Amazon's Selling Partner API.

He finished it after I left. Screenshot the form, ask Claude what it meant, work through the next step. He forwarded me his approval email about an hour later.

That is the kind of progress I want from coaching. The next unfamiliar screen becomes something you can work through. Jarom's full story includes the tools he built himself and the setup he later rebuilt without calling me.

At CrewTracks, Erin Koets had a backlog of meta descriptions. We completed thirty in under two hours, including the WordPress connection that let us update them. She estimated the same work would have taken her at least a week alone.

By session three, Erin was proposing things she wanted to build. Later, she was learning to organize her own projects. The CrewTracks case study follows that progression.

What we actually do in a coaching session

Bring work you recognize. A report you keep putting off. A form you don't understand. An email that needs an answer and has somehow become tomorrow's problem three days in a row.

In the engagements above, we worked in person, on their own computer, with their actual tasks. Some of the time went into setup. Some went into learning to ask for what they needed. Some went into noticing an old habit, like hunting through email by hand when the connected tool could help find it.

My current coaching happens over video. The work we choose depends on where you're starting and what you're trying to do. Those client stories describe what happened with them; they aren't a fixed sequence everyone has to buy.

I would want you to understand what the AI produced and how to check it. Can you trace an answer back to the material you gave it? Can you spot an invented detail? Can you make a useful change without starting over?

You don't need to arrive knowing the technical vocabulary. You do need a task you care about enough to try it yourself.

Fractional leadership starts with decisions the business needs to make

Imagine sales wants help preparing proposals, marketing wants a reporting system, and the owner is being pitched another AI subscription. Each idea may be reasonable. You still need to decide which problem gets attention first, what you already have, and who will carry the work forward.

That is a conversation about priorities and responsibility. My fractional AI leadership work can cover strategy, the roadmap, vendor calls, and working with your leadership team as those decisions develop. The scope and pricing depend on the work we agree to take on.

Before agreeing to an engagement, I would want to get specific about implementation. Who is building? Who approves access and spending? Who checks the result? What can your existing team take on?

A title doesn't answer those questions. Put the responsibilities in the scope. My fractional Chief AI Officer guide on Aperture goes further into evaluating that kind of help.

Try this before you book anything

Pick one piece of work that stalled this week. Write down:

1. What needed to be finished? Name the output: a proposal, a report, a decision, a working process.
2. Where did it stop? Missing information? Difficulty using the tool? Nobody agreeing on what to do?
3. Who could check it? Someone needs enough context to know whether the answer or finished work is useful.
4. What would happen next week? Could you repeat it yourself, or would the same decision land back on someone's desk?

Put names beside the work

Imagine Alex handles customer service at a small business. Every week, Alex reads the follow-up notes and makes a list of customers still waiting for an answer. Morgan, the owner, is also considering another AI subscription.

This example and everyone in it are fictional. For a first practice run, they use ten made-up customer records in a tool the business already approves.

• What needs finishing? A summary showing each customer's open question, who needs to answer it, and any promised date. Missing details stay marked as unknown.
• Who decides what matters? Morgan agrees that finding unanswered questions is the first job to try.
• Who does the work? Alex supplies the records, asks for the summary and corrects it.
• Who approves the data and spending? Morgan approves this practice with fictional records. Real customer data would need a separate approval. Nobody buys another tool for this exercise.
• Who checks the result? Alex compares every entry with its source record, including names, dates and unanswered questions.
• Who does it next time? Alex keeps the instructions and runs another practice batch of fictional records. Real records wait for separate approval.
• How will they know it's done? All ten records are accounted for, unsupported details are removed, and Alex can repeat the work and explain the checks.
• Still undecided? Whether Alex can use real customer records, and who approves that. Morgan's subscription decision is also open.

The subscription question needs its own answer. Alex compares the current tool with the proposed one, including costs, permitted data use and what maintaining it would involve. Morgan approves the information shared and any spending, then reviews the comparison. Morgan records whether to buy, wait or keep the current tool, with a reason and a next step. Morgan owns the follow-up on that decision; Alex owns the weekly summary.

Alex may need coaching to get comfortable doing the summary. If the team wants a known task connected to its existing software and has settled who will run it, it could hire someone for that specific build. If decisions about tools and responsibility keep stalling across the business, ongoing leadership support may fit. Write down what the person would actually own before choosing the title.

Copy this for your own stuck task

Fill this out in your own words. If a name or decision is missing, leave the question visible so you can settle it together.

If you can explain the task and check the result, try a small version yourself. A straightforward draft with material you're allowed to share may be enough to get started. You can learn something useful before hiring anyone.

If you keep getting stuck doing the task, coaching is worth considering. If the work stalls because priorities and responsibilities remain unsettled across the business, discuss leadership support. And if you already know what needs building and who will own it, a focused implementation may be enough.

Bring that one stuck task to a 30-minute discovery call. Tell me where it stops, and we can figure out what kind of help would be useful.