Decide what you are hiring for
Most fractional AI hires fail at this step. If what you want is an AI strategy, a governance policy and a board deck, most people selling the title will deliver that well. If what you want is work leaving your team’s hands this quarter, you need someone who builds, and that is a smaller pool. Write down which one you want before you talk to anyone, because the two kinds of practitioner sound identical on a first call.
Where to look
Fractional executive directories such as Fractionus and fractionalcaio.org list people by title. LinkedIn search for “fractional Chief AI Officer” returns the people who use the title publicly, and their posts show whether they talk about systems they built or frameworks they recommend. Upwork is where many fractional AI people hold a verified track record, with client reviews and a job success score you cannot edit. Referrals from other owners in your industry are the most reliable source and the slowest.
The seven questions to ask
- What did you build yourself in the last six months?
You want a specific system, the tool it runs on and the number it changed. A list of clients advised is a different answer. - How do you measure return?
Hours removed, errors removed and hires avoided are checkable. Anything that cannot be counted is not a measure. - How many clients do you carry?
Above six, the days per week on each engagement stop adding up. Four is a full roster for anyone who builds. - Do you resell or take commission on any tool you recommend?
The honest answer is a flat no. Reseller margin decides roadmaps quietly. - Who builds, and who owns the vendors?
If the answer is your team, price their time into the decision. If the answer is the candidate’s own engineers, ask to see their reviews. - What does day 90 look like?
You want something running in production with your team trained on it, and a first monthly report. A roadmap alone at day 90 is a consulting project. - What happens to the systems if we stop?
Everything should live in your accounts, be documented and be handed over in the final month.
Red flags
- They quote a price before they have asked what still runs on you personally.
- They have never run a business, so the conversation starts at tools rather than at margin and headcount.
- They cannot name a client outcome with a number next to it.
- They propose a dashboard or a chatbot in the first meeting, before any data has been cleaned.
- They want a twelve-month commitment up front. Three months is enough to show a measured result.
What to pay
Retainers run $5,000 to $30,000 a month in 2026, and the spread is almost entirely about whether building is included. The full breakdown, including my own two rates and the comparison with a full-time hire, is on the pricing page. What matters at the hiring stage is that you agree the 90-day outcome before you agree the number.
Start with an audit rather than a retainer. Two weeks on one part of the business produces a map of where the hours go and a ranked roadmap, and shows you how the person thinks before you commit. Mine is $4,500 and credited in full against either retainer within 60 days. That is the automation audit.
What the first 90 days should produce
Days 1 to 14 should be an audit where the person sits with your team and watches how the work actually happens, inventories every tool you pay for, and writes down where the hours go. Days 15 to 30 should produce a ranked roadmap with an estimated saving on each item and agreement on the first three. Days 31 to 90 should end with those three running in production, your team trained on them, and a first monthly report showing what changed. If a candidate cannot describe their first 90 days in roughly those terms, they are selling something other than ownership.
If you would rather talk to me
I am a Fractional Chief AI Officer for US service businesses between $1M and $50M, I founded an automation agency and co-founded an AI-native SEO agency, and I answer every question above on a 30-minute call, including the one about what I built last.