There’s Room for Both of Us in the Front Seat
My brother (pictured) is three years younger than me. When we were kids, there was nothing more contentious than deciding which one of us got to sit in the front seat. If we were left to figure it out on our own, we would have arrived at our destination with an embarrassing amount of scrapes and bruises.
Between the ages we deeply cared about the front seat and the time we couldn't have cared less, there was plenty of room for both of us. That didn't matter. What mattered was the prestige of getting dropped off at school and emerging from that front seat like royalty; red carpet, white doves imagined.
So, as most parents do, ours developed a system of how we would take turns. My brother would sit up front on the way to school, I would sit up front on the way home, switch, reverse, repeat. There was room for both of us in the front seat, and yet we were obsessed with being first, right, winning. We're also incredibly competitive people.
Sometimes when I sit down to write things, a memory pops up that on the surface has little to do with the topic. If you've been with me long enough, you know that doesn't matter :)
So buckle up for HR and AI. There's room for both of us in the front seat.
Where I've landed on this
I've been writing all year about the gap between 2026 plans and the people infrastructure to back them up. Several of those gaps have been closed with AI. Time will tell if that was the right move.
The tools are real, the speed is undeniable, and the conversation has moved on. A few months in, the question isn't whether AI belongs in the people operations space. The question is where AI does real work, where it stumbles, and what the experienced practitioner brings to the combination. Three things, in that order.
Where AI does real work
AI is clearly earning its keep in specific, repeatable ways: first drafts got a whole lot faster. Policy language, internal announcements, offer letters, board memos, and the pre-read for the leadership offsite.
Research compressed, too. Benefit provider comparisons, comp benchmarking, vendor due diligence, policy research. What was once a week of gathering information now takes an afternoon of prompting and verifying.
These are the right uses, and they’re the floor, not the ceiling.
Where the tools stumble
AI also stumbles in specific, repeatable ways. That's not a reason to stop using the tools. Instead, there are reasons to watch the seams.
Kat Kibben at Three Ears Media ran a head-to-head test of AI-written job posts against ones written by recruiters they had trained. AI scored higher on title analysis, templates, and grammar. Humans scored higher on tone, bullet discipline, and the intake conversations that make a job post worth reading in the first place. Kat's read: AI works best when it's layered into specific stages, not when it's used as a one-click generator. That matches what I've been seeing across HR and people operations more broadly. The tools handle the parts of the work that were already templated, and they struggle with the parts that were never going to template well.
Performance reviews are an obvious one. Employees can clock when a review was generated versus written, and the moment they do, the whole exercise stops doing its job. The em dashes…my goodness, the em dashes are the deadest of giveaways.
Post-acquisition integration is another place where the seams show. AI can build the integration plan, draft the all-hands deck, and map the new org chart. What it can't see is which retained leader is signaling they're going to leave in three months, or which team is about to lose trust in the deal because of how one early decision got communicated.
None of this means AI isn't working. The fix isn't less AI. It's a stronger human layer on top of whatever the tool produces, watching where the seams slip.
Where the draft ends
AI produces a strong draft. What follows is the work, and that work is where everything I've seen earns its keep.
The executive hire is the cleanest example. The tool can scan resumes, draft the job description, and summarize references. What it can't produce is whether this particular leader is going to land in the room with the existing team, or whether the founder's hesitation is a real signal or a passing nerve.
Departure signals are another. Who's about to leave isn't in the data. It's in tone shifts, meeting energy, and small behavioral changes that only read as signals after watching enough people move through enough roles. That's not a prompt.
Then there's everything that happens before an acquisition closes. Data doesn't surface people-side landmines, and diligence tools don't either. The risk hides in what the leadership team isn't saying out loud. Reading it takes two decades in rooms with the same kinds of people.
And the single most valuable skill in this work doesn't fit on a slide or in a prompt: knowing when to push and when to hold. That read is built from watching situations play out and learning which ones need pressure and which ones need patience.
This is the part of the work AI doesn't reach. There's room for both of us in the front seat. Only one of us is reading the road.
How to Get More Out of Both From Here
Here's what I'm watching for in the practices that are running AI well:
A review layer on the highest-stakes AI output. Not to catch mistakes but to ensure the output reflects the decision the team needs to make. The tool drafts and the leaders decide.
Keep the manager muscle warm. At least one nuanced conversation per manager per quarter, completed without an AI assist. Not because AI is bad. Because the skill stays sharp only with use, and it needs to be sharp when the stakes are high.
Personalize where it matters. Senior candidate outreach, top-performer check-ins, and high-trust internal communications lose their value the moment they sound generated. The tools handle the routine while preserving the moments people remember.
Put the voice back in. Redirect job descriptions, handbooks, and internal comms toward something that sounds like the company again. Distinctiveness is easy to restore and immediately visible.
Draw the line explicitly. Commit to and document where AI ends in HR and people ops, and where human experience takes over. Then, share that line with the leadership team. Take it a step further and include the line in the company's AI policy. Deciding on a case-by-case basis is how inconsistency compounds.
What this means for the work
The case for fractional HR and people operations used to be capacity. Senior expertise without the full-time cost. While still true, this is now only a piece of the picture.
The case now is the pairing. AI handles draft and speed while the practitioner handles the parts of the work AI doesn't reach. Specifically: the executive 1:1 nobody else heard, the founder who needs to walk back a decision without losing the room, the acquisition risk hiding in plain sight.
That's what a career spent in the room looks like when it's working with AI, not against it.
Where to go from here
Now that we're a few months into the year, most companies have built something. Some of it is working, and some of it could work a lot better with another set of experienced eyes.
If any of this is landing, the seams are showing up somewhere in real AI investments right now.
The first session is on me. Forty-five minutes on what's been built so far, where the AI investment is paying off, where it needs another layer, and the two or three moves that will produce the most leverage from here.
That's the work I love doing most.
