No AI Team? A Fractional AI Expert, an AI Hire, or a Partner for HR

Your CEO wants to know what HR is doing with AI. You own the HR work but no engineers, and the tech team’s roadmap is full until next year.

So you have three ways to get the capability. You can rent a fractional AI expert for a few days a month. You can hire someone. Or you can bring in a partner that builds the agents and hands them over.

Most pages on this question are written for a CEO picking a Chief AI Officer. We read the five that rank, and HR as a function shows up in none of their page text outside a nav link. This post answers it for HR. We compare the three routes on cost and speed first. The question that decides it is who owns the running agent in month 12, and who answers for it when it touches a decision about a person.

The three routes side by side

	Fractional AI expert	AI hire	Build-and-run partner
What you pay for	Advice and oversight, a set number of hours a month	A salary, benefits and a recruiter fee	A fixed fee per step, with a go or no-go at the end of each
Cost signal	$8,000 to $30,000 a month for a fractional Chief AI Officer retainer (JRG Partners)	$500,000 to $800,000+ in year one for a full-time Chief AI Officer (same source)	Scoped per engagement
Time to start	Days to 2 weeks (JRG)	6 to 9 months for an executive search plus notice period (JRG)	Opens with a strategy step of about a month
Running after month 6	A roadmap and decisions, often no working system	Whatever one person could build alone	Working agents inside your systems
Who can change it	Nobody, once the retainer ends	The hire, while they stay	Your team, with the code and runbooks
Typical failure	Nobody is there to build what was advised	One engineer, no data access, no peers	A system your team can’t run without the partner

A note on the numbers: the JRG figures are for an executive Chief AI Officer, not an AI engineer, and they come from a search firm that sells the fractional model. Treat them as a signal of scale, not a quote.

Route 1: the fractional AI expert

A fractional AI expert is a senior person you rent for part of each month. Most sell it as a fractional Chief AI Officer or a fractional head of AI. They sit in your leadership meetings, set priorities, review vendors and help you answer the board.

The retainer buys judgement, but it usually doesn’t buy building. JRG Partners puts the embedded or intensive tier at $8,000 to $30,000 a month. That buys strategy and oversight for a company in a “$20M to $200M” revenue band.

Who it fits: an HR leader who needs someone to make build-versus-buy calls and stand in front of the board. It works best when engineers or a vendor are already lined up to do the work.

Where it fails: in an HR function with no one to build. The retainer pays for advice, and if nobody on your side can turn that advice into a working agent, you get a good roadmap and nothing running by month 6. That’s the HR case most of the time.

The title is also easy to claim, so test it in the interview. Ask each candidate to walk you through an AI system they took from brief to production, and what that system does today.

Route 2: the AI hire

The second route is to hire. That usually means one AI engineer for HR. It can also mean a senior lead who will hire engineers later.

We won’t quote a single salary for an AI engineer here. Published pay datasets disagree by a wide margin because each one measures a different slice of the market. Price the role with your own compensation team. At the executive end, JRG puts year one for a full-time Chief AI Officer at $500,000 to $800,000+. That total includes a one-time search fee of $100,000 to $150,000+.

Time is the bigger cost. JRG counts 6 to 9 months before a full-time Chief AI Officer starts, once you add the search to the notice period, so a hire can’t give the board an answer this quarter.

The hire works when HR has a long list of agents to build and clean data the hire can reach, plus an IT team willing to give them access and review their work.

The usual failure is the lone engineer. One person lands in HR with no peers to review their code. They wait weeks for access to the HRIS, and nobody in HR can tell good work from bad. The first pilot stalls in that queue of access questions, which we walk through in why HR AI pilots stall. If the hire leaves, the knowledge of how the agent works leaves with them.

Route 3: the build-and-run partner

The third route is a partner that builds working agents in your environment, then hands them to your team. The risk with any outside firm is a stack of reports and no change in how the work gets done. So ask every partner what will be running after they leave, and who will run it.

Culturro is the partner this post describes. Since 2017 we’ve worked with more than 30 organizations on HR strategy, engagement, training and leadership development. We also build and run AI of our own. Merlin, the AI coach in our product Risely, has coached 5,000+ people in 40 languages inside Slack and Teams. Running it taught us what it takes for employees to keep using an agent after launch day.

Our engagement has three steps. Each one is on a fixed fee and ends in a written go or no-go:

1. Strategy, about a month. You get a sequenced roadmap and a business case for each use case in the numbers your CFO tracks. You also get a written list of what AI won’t decide in your function.
2. Proof, a few weeks per use case. It runs in our sandbox on synthetic data, or in your own cloud tenant on synthetic or de-identified data your security team approves. The go or no-go is written against accuracy, failure cases, cost per query and whether your team used it.
3. Build, a production system inside your environment, under your security team’s controls.

Everything we build transfers to you at handover: the code, prompts and evaluation sets, plus the documentation and runbooks. There’s no Culturro platform underneath, and we hold no production credentials after handover. Your own providers bill you for hosting and model usage. Ongoing support is an optional month-to-month retainer, and you don’t need it to keep a delivered system running. If your own team will build, we write the roadmap and the proof so they can.

We also don’t build systems that make decisions about people on their own. That covers hiring and pay, and it also covers promotion, discipline and termination.

Who it fits: an HR function with no engineers that needs working agents in months and wants to own what gets built.

Where it fails: when the partner leaves behind a system only the partner can run. That’s the failure to screen for, with us or anyone else. We compare the kinds of firms in HR AI consulting firms. Whoever you pick, get the handover terms in writing before the build starts.

What each route leaves you owning in month 12

Cost and speed are what you compare in month 1. By month 12 you either have systems running or you don’t, and you answer for them either way.

In month 12	Fractional AI expert	AI hire	Build-and-run partner
Running systems	Whatever someone else built from the roadmap	What the hire built, if they’re still there	Agents in your systems, with code and runbooks
Who answers the board	You, using the expert’s advice	You and the hire	You, with documented evaluation results
Who signs off a decision about a person	A person in HR	A person in HR	A person in HR

The last row is the same in every column on purpose, because no route moves employment-law exposure off the HR function.

Two laws make this concrete. Local Law 144 in New York City requires an employer using an automated employment decision tool to have a bias audit done within a year before use. The employer must also publish a summary and give notice to candidates and employees.

The EU AI Act lists AI used for recruitment and selection as high-risk in Annex III. It does the same for AI used to make decisions on terms of work, promotion and termination. AI that monitors and evaluates workers’ performance is on the list too. Screening, pay and performance tools can all fall inside that line.

So put one question to each route. When a regulator or an employee asks how the agent reached an outcome, who can open the logs and explain it? A fractional expert often leaves the answer to a vendor. A hire makes it depend on one person staying. A partner makes it depend on whether the handover was real.

Matching the routes to the HR Agent Map

The HR Agent Map lists 122 HR agents across 12 HR areas. Each one carries a verdict and a sensitivity rating. The verdicts split like this: 48 Proven, 55 Strong case, 7 Emerging and 12 Human decides. Sensitivity splits into 27 low, 56 medium and 39 high.

Those two fields tell you a lot about which route an agent needs.

Kind of agent on the HR Agent Map	What it needs	Route that fits
Proven, low sensitivity	Someone to build it once and an HR owner to run it	A partner, or software you buy
Strong case, medium sensitivity	A proof on your own data before anyone commits	A partner with a proof step, or a hire with IT support
High sensitivity, any verdict	Evaluation sets, logs, legal review and a named owner in HR	A partner that hands over the evidence, or a hire with peers
Emerging	Someone to watch the market and say “not yet”	A fractional expert
Human decides (12 agents)	A person makes the call; AI only prepares the file	Any route, as long as nothing decides on its own

A fractional expert is most useful at the edges of the HR Agent Map, where the work is judging what’s ready. Most agents sit in the middle, and those need someone to build them, test them on your data and hand them over.

Start with the agents you actually want. The HR Agent Map shows each one’s verdict and sensitivity, and how to prioritize HR AI use cases shows how to pick the first two or three.

A decision framework by company situation

Before you pick a route, check that you’re ready for any of them. We use the 10X HR-AI Framework for this. It scores a use case on three lenses: Impact, Trust to earn and Readiness. Each use case lands in one of four tiers: Build now, Proof first, Build with guardrails or Not yet. A Not yet comes with the one thing to fix first.

Once that check is done, match your situation to a route.

Your situation	Route
You have engineers or IT capacity, but nobody to set direction	A fractional AI expert. You’re buying judgement because someone else can build
You have a long list of Build now agents and clean data, and IT can support two or more years of work	Hire, and plan for more than one person
You have no engineers, the board wants working agents this year, and your top use cases score Proof first or Build with guardrails	A partner that hands over everything
Your top use cases score Not yet	None of the three. Fix the thing the framework names

If you haven’t set a strategy yet, start a step earlier with how to build an AI strategy for HR. The money side is in budgeting for AI in HR.

When none of the three is needed yet

Sometimes the honest answer is no route at all. If your HR data sits in spreadsheets nobody trusts, an expert will advise you to fix it. A hire will spend the first six months fixing it. A partner’s strategy step will tell you the same thing. You can start that work without any of them. Our guide to HR data readiness covers what has to be true first.

The same goes if you can’t name the agent you want. “We need AI in HR” isn’t a use case. Pick one task, with an owner, a volume and a cost today. Come back to this choice once you have that.

Our recommendation

We’d pick the partner route for an HR function with no AI engineers. We’d also make handover the condition of the deal.

Here’s why: a fractional expert is good value when someone else can build, and in most HR functions nobody can. A hire takes 6 to 9 months to arrive at the executive level and leaves one person holding everything. A partner builds inside your systems and hands over the code, prompts, evaluation sets and runbooks. That leaves you with something running in month 12 that your team can change.

We’re a partner, so weigh that. The test we’d want you to apply to us is the same one we’d apply to anyone. Ask what you’ll own when the engagement ends, and get it in writing.

The one exception: if your engineers are already in place and idle, rent a fractional expert to aim them.

---

Talk to us before you pick a route

We’ll spend 30 minutes on the agents you want. We’ll check them against the HR Agent Map and tell you which route fits. If it’s a partner, we’ll show you how we’d build on your own systems and what you’d own at handover. If it’s not us, we’ll say so.

Book a 30-minute call

Frequently asked questions