What Are Employers Actually Hiring For Right Now?

Agentic AI didn't just change what companies buy. It changed who they hire, and it changed fast enough that most job descriptions were obsolete before HR finished posting them. Every senior leader I've talked to this year, across ad agencies, B2B media and information companies, marketing services firms, technology CMOs, investors, and sales leaders, keeps circling back to the same thing: the attributes that get someone hired and how to coach, and manage them are being rewritten in real time.

Working inside my own advisory clients’ hiring and employee development plans is what made the impact of that shift impossible to miss.

Alex King, a recruiter who places talent across SaaS, marketing services, and tech startups, already had a name for what I was seeing: two hiring waves running at once. One builds AI into the product. The other, almost invisible by comparison, builds AI capability into the people already on staff.

I spent the next few weeks testing that distinction against a dozen more conversations: a CMO advising Norwest portfolio companies, an HR leader who’d just led her own company through the same transition, a CRO, a venture partner, and a longtime demand-gen operator who sits on several boards in the space. Nobody disputed Alex’s read. Each conversation sharpened it from a different seat: marketing, HR, revenue, capital, the boardroom.

Companies are hiring for AI in the product. Almost nobody is hiring for AI in the workforce

Four hiring criteria showed up in nearly every conversation, regardless of function, stage, or operating model:

* A baseline expectation of demonstrable AI use, regardless of the role

* Little tolerance for a long onboarding or ramp period: less a preference than a reflection of how fast the ground is moving

* Technical skill in the function usually outranking domain expertise, which slid from requirement to nice-to-have

* Demonstrated skill and mindset prized over years of experience

The gap nobody’s staffing

Every company I talk to has someone building AI into what they sell. Almost none of them have someone whose job is making the people already on payroll good at using it. I named this gap back in “The AI Layoff Myth” . The new roles wouldn't be AI research jobs. They'd be built to sit between agentic systems and the humans still doing the judgment work.

Alex put a name on one of them: Enterprise AI Enablement Engineer. Sits between business stakeholders and AI systems, builds internal workflows in real time. Most companies aren’t hiring for it yet. The ones that are can’t find enough people.

That gap shows up at the operating layer first, in titles like GTM Engineer, Director of Revenue Intelligence, and Revenue Analytics: roles built to optimize AI across go-to-market, not just deploy it.

A CMO I spoke with called it the biggest shift he's seeing right now, concentrated almost entirely inside marketing and revenue operations.Then it shows up a level up. The same pressure producing GTM Engineers below the C-suite is producing Chief Growth Officer and Chief Commercial Officer roles above it.

Marketing’s standalone impact is declining, and the gap between marketing and sales has gotten too costly to manage as two functions reporting to two different people. Same root cause, playing out at a different point in the hierarchy.

A title relabeling what a strong CRO already does, or a real structural shift? When I put that question to a CRO I spoke with, he waved it off almost entirely, and the dismissal turned out to be its own answer. He’s not thinking about titles or org charts. He’s thinking about workflows, one at a time.

The pattern underneath both: prompt engineering was a role built around a skill, writing good input for a model. What’s showing up in job titles now is a role built around owning a workflow, the loop the model sits inside, the same shift I described in Loop Engineering: The Bridge to Workflow Transformation: the premium moving from working the tool to owning the standard it runs on. The org chart is one step behind the work.

Solving it one workflow at a time

Not everyone I spoke to is building a company-wide agentic AI strategy. The CRO told me his team took a different tack: instead of a strategy, they take workflow problems one at a time and ask a single question. What can agentic AI do to improve or transform this specific thing?

The example he gave me: the performance decks his Customer Experience team builds for clients, pulling data from display, lead generation, and lead-level detail. These used to take several days. They now takes minutes, and the team added something it never had time to do before: a specific recommendation for each individual lead. The result shows up on every call. Reps walk in with something actionable to say, not just a status update. He credits that shift with the team’s best Q2 in three years.

His hiring bar has moved with it. He’s looking for sales reps who are, in his words, “entrepreneurial with AI,” people who can point to what they’re already experimenting with rather than waiting to be shown. A recent hire who beat quota in his first ninety days, something he said he hadn’t seen before, fit that description exactly.

He’s watching for the opposite signal too. A few of his reps still ask about AI enablement sessions. He told me that’s the wrong question: it means the rep is waiting to be told rather than engaging with the technology directly. Companies need to own the infrastructure for AI enablement; the gap above makes that case. The individual people worth hiring into that infrastructure are the ones who don’t wait for it.

I also asked whether management layers are compressing inside revenue organizations the way they are elsewhere. He didn’t frame it as a headcount question. His team isn’t removing layers. They’re removing manual work inside the roles that already exist, and redirecting the time toward the parts of the job only a person can do.

What’s actually changing

Evaluation criteria are shifting from track record to demonstrated adaptability. That’s what I was seeing across my own client roster, and the HR leader I spoke with confirmed it from the inside, then put herself forward as the example: hired without the full slate of HR credentials, on the strength of a mindset that proved out once she was in seat. She’s rewriting job architecture the same way, testing for how fast someone learns rather than what they’ve already done. It’s a harder interview question to evaluate. It’s also, by her account, how you find the people who turn into your best hires.

None of this surprised me. Based on input from clients and experts in the field, I’ve built two web-based tools for hiring, coaching, and managing sales and marketing talent: Sales Talent Matrix™ and Marketing Talent Matrix™. Practical AI Fluency has been a formal attribute in the sales version for months, added at Alex’s suggestion after he kept seeing it separate strong candidates from weak ones. A CMO I work with helped shaped the marketing version the same way. Neither of us called it a hiring trend at the time. It just seemed like the obvious thing to measure. It turned out to be both.

Depth and originality are gaining value, not losing it. The same advisor watching the GTM Engineer role emerge is also watching content teams shrink and refocus: less volume, more original research and point-of-view work, built to stand out against a market flooded with generic AI output. Trust and differentiation are scarce again, and scarce is valuable.

Management layers are compressing; scope is redistributing, not eliminated. The HR leader I spoke with rebuilt her company's management cadence from scratch once managers started carrying more people: structured 1:1 templates, a realistic meeting rhythm, because no manager with twenty direct reports can run the old playbook.

What isn’t changing

* Comp still matters. It’s just not the only lever, and for the candidates worth competing for, it’s stopped being the deciding one.

* Expertise still matters. What’s changing is the packaging: raw credentials carry less weight than proof you can apply what you know, fast, in an environment that won’t sit still long enough to hand you a training manual.

* Management still exists. It’s leaner, and the people left in it are doing more strategic, less administrative work, but the function hasn’t disappeared. It’s being redesigned around fewer, better conversations.

The tension nobody’s resolving

Alex’s read: the best AI-native candidates are choosing speed and autonomy over money, consistently, because they want to work somewhere that moves the way they think.

A quieter, more transactional read sits right next to that one. Call it skills-maxxing: people optimizing their current forty hours for what they can carry into the next opportunity, faster, rather than committing to a mission. Professional development interest, largely dormant since 2023, is back, driven in part by people wanting to bank AI experience they can take with them.

Both readings are probably true of different people in the same market. I’m not going to resolve that tension here, and I’d be suspicious of anyone who claims they can.

What this means over the next 24 months

The companies staffing both sides of this wave, the product-facing AI team and the workforce-facing one, will out-execute the ones only staffing the visible half. Right now, most companies are only staffing the visible half. That’s the opportunity, and it’s closing faster than most org charts are moving to catch it.

A venture partner I spoke with, who has 35 active investments across marketing and technology, sees the same thing from the capital side. What he’s underwriting now isn’t proficiency with the technology itself. It’s the ability to apply agentic AI to an existing business to drive scale and revenue growth. That’s a different bar than knowing how to use the tools, and it’s the one investors are actually funding.

For CEOs

* Name a single owner for internal AI enablement now, even fractional or borrowed from another function. The Enterprise AI Enablement Engineer gap described earlier doesn’t close itself, and unowned work never gets done by consensus.

* Rewrite job architecture to test for adaptability as a discipline, not a talking point. The HR leader in this piece was hired on that basis herself, then rebuilt her own company’s job descriptions around the same standard once she saw it work.

* Before restoring a management layer, make someone justify the span of control with a real cadence, not a headcount instinct left over from the org chart you had two years ago. That instinct is exactly what keeps companies staffing only the visible half of this wave.

For Operators

* Skip the company-wide AI strategy document. The CRO in this piece produced his best Q2 in three years by taking one workflow, the client performance deck, and asking a single question: what can agentic AI change about this specific thing. Pick one workflow and answer that question before writing a strategy anyone has to sign off on.

* In interviews, listen for whether a candidate is already experimenting with AI unprompted, on their own time, versus asking when the enablement training starts. That distinction predicted a rep who beat quota in his first ninety days. It reads faster than anything on a resume.

* When you compress a layer, name exactly where the freed time goes. Cutting a layer without redirecting the time toward judgment work only leaves you thinner, not sharper.

For Investors

* Stop underwriting proficiency with the tools themselves. The venture partner in this piece, with 35 active investments across marketing and technology, has moved past that question: he’s underwriting whether a team can apply agentic AI to an existing business to actually grow revenue, a materially different bar.

* In diligence, ask whose judgment is actually encoded in a target’s AI-driven workflows, and whether that judgment survives the departure of the one person who built it. An undocumented standard held together by one employee is a retention risk wearing an AI capability as a costume.

* Read “AI-driven” cost cuts skeptically. A round of layoffs with an AI label attached is not evidence of anything by itself. Ask what got rebuilt, not just what got removed.

Three different vantage points, the same blind spot underneath: everyone has gotten comfortable evaluating AI in what a company builds or sells. Almost nobody has built the discipline to evaluate AI in who a company hires and how it develops them. That gap is where the next eighteen months of B2B leadership actually gets decided.

My thanks to everyone who shared their insights and feedback: Alex King, Scott Vaughan, Emily Hayes, Scott Anderson, Craig Johnson, Tom Click, Phil Brannon, Tonia Ries, Sam Thompson, and to the Uphoff Advisory clients whose hiring and talent conversations sparked this piece in the first place.

The views expressed in Uphoff on Media are entirely my own. They don’t represent the opinions of any company I’ve led, any board I’ve sat on, or any investor who’s had the pleasure of debating strategy with me over the years. If something I write here sounds brilliant, I’ll take full credit. If it turns out to be wrong, I was clearly misquoted by myself.

“Uphoff on Media” is published by Tony Uphoff, Founder and Managing Partner of Uphoff Advisory, LLC: a strategic advisory practice for founders, CEOs, and investors in B2B information, marketing, and technology. The businesses that drive business.