Choosing between a fractional AI lead and a first AI hire

In a survey HFMA published in November 2025, two-thirds of the 125 healthcare CFOs who answered said a chief AI innovation officer will become an important C-suite role in the next three years. In the same survey, CFOs rated their own preparedness for AI deployment at 2.82 out of 5. An organization of 100 to 1,000 people has three options for closing that gap.

Three options

Option one is a full-time AI or data leader. A 2026 salary guide from the staffing firm KORE1 puts a chief AI officer's base salary at $280,000 to $400,000 at companies with $50 million to $500 million in revenue, and $400,000 to $550,000 at companies with $500 million to $5 billion, before bonus and equity. If that salary is more than the budget allows, look at option two.

Option two is a first data scientist or "AI person," hired into an analyst or manager band. The BLS median wage for data scientists was $120,230 in May 2025. Benefits, including payroll taxes, add about 45 percent to wages for professional jobs in private industry, according to the Bureau of Labor Statistics, so a data scientist at the median wage costs about $174,000 a year before tooling and the recruiting fee. The catch is how much one person can cover. A single mid-level hire can build models, but setting the rules for which tools may use which data, negotiating vendor terms, briefing the board and running a build are a lot to add to one person's job. A new hire also has to learn what data, systems and contracts the organization already has before building much.

Option three is a fractional AI lead: a senior person for about four working days a month on a fixed retainer. Kent Applied's version starts at $6,000 a month, and larger engagements get their own quote. Twelve months at $6,000 is $72,000.

Year-one costs, calculated from the figures above rather than measured:

• Full-time leader: $280,000 to $400,000 base before bonus, benefits and equity, and a search of 8 to 12 weeks once the job description and pay band are agreed.
• First data scientist: about $174,000 a year with benefits and payroll taxes at the median wage, and 7 to 12 weeks to hire on your own.
• Fractional lead: $72,000 at the starting fee, a fixed-price proposal within 48 hours of the first call, and a roadmap and 90-day plan by the end of week four.

The fractional option costs less than half as much as the first data scientist in year one, and about a quarter or less of the executive's base salary alone. Its larger advantage is speed.

Time to fill

A staffing-firm benchmark for 2026 says that when a company recruits a data scientist on its own, without an outside recruiter, the search takes 7 to 12 weeks. A senior one takes 9 to 13 weeks, and roles that require experience running generative AI in production take 14 to 20 weeks or more. SHRM's 2025 benchmark, as KORE1 reports it, puts the all-jobs US average at 44 days.

Vendor pitches keep arriving during those weeks, and whoever has time makes the decisions. Part of the cost of a slow hire can be the contracts signed while the seat is empty.

What a fractional lead should own in 90 days

By day 90, a fractional lead on retainer should have produced these six things.

1. A baseline on three workflows. Volume, touch time, cycle time, and error rate for the three processes leadership complains about most, measured from data the organization already has. Later work is measured against them.
2. A data posture memo. This is one page, signed by compliance, that says which tools may use which data, which version of each tool is allowed (consumer, enterprise or API) and under which contract.
3. A costed roadmap. Which workflow first, what it costs today, what a fix costs to build or buy, and what will be measured 90 days after go-live. Each line says whether the time saved is at the bottleneck.
4. One working proof of concept. A tool that runs on the organization's own de-identified cases, with a named person checking its output.
5. Vendor terms on any purchase in progress. These cover limits on how the vendor may use your data, a clause keeping your data out of model training, exit rights, and the list of other companies that will handle the data. A fractional lead works for the buyer and takes no resale or referral fees.
6. A hiring spec, if a hire is coming. The fractional lead writes the job description for their own replacement and interviews the candidates.

Judge the retainer like any build: what exists now that didn't before, and what it measurably changed.

When to hire full time instead

Two cases.

First, when there are three or more production systems that need daily ownership: a model that drifts, a document pipeline that breaks when a payer changes a form, a reporting stack that feeds a regulator. Four days a month cannot cover daily failures. At that point the fractional lead should hire a replacement, hand off, and end the retainer.

Second, when AI is the product rather than a tool. A digital-health company whose revenue depends on a model needs the person in the building, with equity, every day.

For everyone else, a good sequence is a readiness assessment, then a retainer, then a hire. A two-week readiness assessment shows whether there is enough work to justify a retainer. The retainer produces the roadmap and the first working system. The hire, when it comes, inherits a running program.

The question to answer first

Before comparing candidates or consultants, write down what the person will own on day one. If the answer is "figure out what we should do about AI," that is a project with a start and an end, and a fractional lead will finish it sooner and for less than a new hire. If the answer is a list of named systems already running, hire full time.