Everyone's Hiring an "AI Workflows" Person. Here's Why You Shouldn't.

There's a new job at the top of every "future of work" list, and almost nobody can hire it well yet.

It's called the "AI workflows" role, and it has a powerful champion. Jiaona Zhang — chief product officer at Laurel and an adjunct lecturer at Stanford — recently told Business Insider that every company should be hiring for it, and that it's the role she'd push every new graduate toward. The pay is already serious: Box has advertised an "AI business automation engineer" position paying up to $183,000, and its CEO Aaron Levie has said he expects most companies to end up with some version of the role.

The thesis is sound. The job is simple to describe: find the parts of a business that AI can improve, build or implement the systems that make those improvements real, and prove how much time and money you saved. If that sounds familiar, it should — it's the exact problem every operator is wrestling with right now.

But here's the question nobody in the headlines is asking: for most companies, is hiring a brand-new role really the smartest way to solve it?

The hottest job nobody can hire well

Start with the backdrop. The World Economic Forum's Future of Jobs Report found that 41% of employers plan to reduce their workforce by 2030 as AI automates certain tasks, and big tech companies are already hiring fewer recent graduates than they did before the pandemic. AI is reshaping the org chart from both ends — eliminating some roles while inventing others.

The "AI workflows" role is the most talked-about of the new ones. But "most talked-about" and "easy to hire" are very different things. A search of major job boards turns up relatively few openings under that exact title — the role is still in its infancy. Companies are hiring for adjacent versions under names like AI automation engineer, AI operations specialist, and AI enablement manager, which means there's no established talent pool, no standard playbook, and no reliable way to tell a great candidate from a confident one.

So the pitch to a growing company is: invent a job that barely exists, pay six figures for it, hope the person you hire is actually good at something nobody has been doing for very long, and wait while they figure out your business from scratch.

The catch for real companies

Even the companies with the budget and the brand to attract this talent are struggling to get value from it. McKinsey's latest State of AI research found that organizations are deploying generative AI more than ever, yet most are still working to capture measurable business value from it. The tools are everywhere. The returns are not.

That gap matters enormously for a firm doing $2M–$5M in revenue and trying to break past it. At that size, a $150,000–$183,000 hire — plus benefits, plus ramp time, plus the risk that the brand-new role doesn't work out — isn't a rounding error. It's one of the biggest bets the business will make this year. And you're making it on an unproven role, with an unproven hire, in a discipline the giants haven't fully cracked.

The honest math looks like this. Best case, you find a great person, wait three to six months for them to learn your business and ship their first real systems, and start seeing returns late in the year. Worst case, you spend the same money and time and end up where McKinsey says most companies are — lots of AI activity, no measurable value, and a salary line you can't easily walk back.

Hire the function, not the role

Here's the reframe that changes the decision. You don't actually want a person with a particular job title. You want the function: someone to find what AI can improve in your business, build the systems, and prove the time and money saved.

A function is something you can buy already built and already proven — without inventing a role, running a months-long search, or carrying a six-figure salary to find out whether it works.

This is the part the "just hire someone" advice skips. The work Zhang describes — automating the sales team's follow-up, building an AI agent that preps demo calls, creating internal tools that save hours of admin — isn't one person's job. It's a capability. And capabilities can be delivered as a service by a team that has already done it dozens of times, instead of assembled from scratch by one new hire learning on your dime.

Why a system beats a single hire

Compare the two paths honestly, side by side:

Speed. A new hire needs to be sourced, vetted, onboarded, and ramped — realistically three to six months before the first system ships. A productized AI workflows team can have your first workflow live in around two weeks, because the playbook already exists.

Risk. One hire is a single point of failure: if they're not as good as they interviewed, or they leave in a year, the capability leaves with them. A team carries the redundancy, the documentation, and the institutional memory so the function doesn't walk out the door.

Breadth. The role asks one person to be strong at sales automation, ops tooling, data plumbing, and change management. That unicorn is rare and expensive. A team brings each of those skills without you betting on finding all of them in a single new graduate.

Cost structure. A hire is a fixed six-figure salary you carry whether or not the value shows up. A service is a fraction of that cost, and the entire point of the engagement is to produce measurable, provable returns — the exact thing McKinsey found most in-house AI efforts can't yet demonstrate.

Proof. Zhang's own advice to new grads is telling: the way to prove your worth is to show you saved a group of people real time and created real leverage. That's the standard the role is held to. A team that does this for a living walks in already able to prove it.

The honest exception

To be fair: if you're a large enterprise with the budget to hire, the management depth to lead a brand-new function, and the patience to wait out the ramp, building the role in-house can absolutely be the right long-term move. Box hiring an AI automation engineer makes complete sense for Box.

But most companies aren't Box. For an established firm or a growing SMB that needs measurable results this quarter — not a hiring experiment that pays off sometime next year — the smarter sequence is to buy the function now, prove the value, and bring it in-house later if and when it makes sense. You don't have to choose between "ignore AI" and "bet $183,000 on a new role." There's a third option, and it's faster and cheaper than either.

What this looks like with FrontPipe

This is precisely what FrontPipe was built to be: the AI workflows function, productized. We do exactly what Zhang describes the role doing — find what AI can improve in your business, build the systems that make it real, and prove the time and money saved — except we've done it across operators who've scaled through fourteen Inc. 5000 finishes, and we've been doing it since 2021, before "AI agency" was a job title.

No req to open. No six-month search. No single point of failure. First workflow live in about 14 days, a team instead of one new hire, and a cost that's a fraction of a $183,000 salary. The role everyone says you should hire — running for you, from day one, with the measurable results most in-house efforts are still chasing.