How to Keep Up With AI as a People Ops Leader (Without Losing Your Mind)

Monday: someone forwards them a Substack about how AI is going to eliminate 40% of HR roles by 2027. Tuesday: a vendor demos an agent that promises to run an entire merit cycle in three minutes. Wednesday: the CEO asks what their AI strategy is. Thursday: they spend two hours trying to figure out whether Claude or ChatGPT is better for the job descriptions they keep rewriting. Friday: they close their laptop having made zero actual progress, and feel guilty about it all weekend.

I hear this pattern over and over again. And honestly, I think we've been telling People Ops leaders the wrong thing about AI.

The standard advice — "just experiment! try it for one task a day!" — sounds great until you remember that the people receiving this advice already have a job. A demanding one. With actual humans on the other side of every decision. The pressure to "keep up with AI" has become its own full-time role, and nobody hired you to do that.

So I want to write something different. Not another listicle of prompts you should try. I want to share what I actually think about this — as someone building an AI-powered comp platform, and someone who talks to People Ops leaders every week about where they really are with this stuff.

You're not as behind as you think

The first thing I want to say, and I really mean this: if you feel like everyone else is racing ahead while you're still figuring out how to write a decent prompt, you're not alone, and the data backs you up.

SHRM's 2026 State of AI in HR report found that only 39% of HR functions have actually adopted AI. Another 31% have no plans to launch AI initiatives at all. That means almost a third of HR teams aren't even trying to use AI yet — and the majority of those who are, are still in the experimentation phase.

I think the "everyone is doing AI" narrative is a story we tell ourselves because the loudest voices on LinkedIn are the people doing the most with it. Survivorship bias. The people quietly figuring it out at their own pace don't post about it.

You are not behind. You are exactly where most of your peers are. I see this every single week in customer conversations.

The real problem isn't AI — it's signal vs. noise

Here's what I think the actual challenge is in 2026: it's not "learn AI." It's filtering the firehose.

New tools launch weekly. Every HRIS adds an AI feature. Every consultant has a framework. Every conference has six AI tracks. The signal-to-noise ratio is brutal, and I'd argue most of the noise is generated by people trying to sell you something. (I say that as someone who builds software for a living — vendors are not always your friend on this stuff.)

A few things that I think actually help:

Pick three sources you trust, and ignore the rest. Not three newsletters per topic — three sources, full stop. Mine right now are The CompTech Roundup (Giac Soliman) for comp-tech, the SHRM research feed for serious data, and one or two practitioner voices on LinkedIn whose taste I trust. Everything else gets skimmed at best.

Distrust anything that promises 10x productivity. Industry analysts are starting to make this point too: the harder question in 2026 isn't whether your HR platform uses AI — it's whether that AI is doing anything genuinely useful. I'd add to that: any vendor pitching transformation in a single demo is selling you outcomes that don't survive contact with real data. I've been on both sides of those demos. The honest ones spend half the call talking about limitations.

Treat "trending" topics as a tax, not a signal. Just because everyone is posting about a new tool doesn't mean you need to evaluate it this week. Wait a month. If it's still relevant, then look. If not, you saved yourself the cycle.

How to actually learn AI when you don't have time to learn AI

Here's the thing that finally clicked for me, and I want to share it because I don't see enough people saying it out loud:

You don't learn AI by studying AI. You learn it by giving it a job you already do and seeing what happens.

I think the version of the advice that actually works for People Ops leaders short on time looks like this:

Pick one task you do every week that you don't love. Drafting comp letters. Summarizing exit interview notes. Writing a manager FAQ. Cleaning up a job description. Something repetitive, low-stakes, and yours.

Try one AI tool on it for one month. Not five tools. Not three. One — ChatGPT or Claude works for nearly everything when you're starting out. Use it daily for that one task. You'll learn more in 30 days of real use than in six months of reading.

Notice where it saves time and where it doesn't. This is the part I think most people skip. AI is genuinely useful for some things and genuinely worse than you for others. Building intuition for which is which — for your specific role, your specific company, your specific data — is the actual skill.

Then add a second use case. Not before. The mistake I see almost everyone make is trying to apply AI to ten things at once, never going deep on any of them, and concluding that AI "doesn't really work for HR." It does work. They just haven't lived with it long enough to know how.

What I think actually separates great AI users from average ones

There's a line from Giac Soliman in our recent webinar together that has genuinely stuck with me: "The sweet spot isn't who can write the fastest prompt — it's who can consistently execute good judgment using AI."

I keep coming back to this.

The skill that matters in 2026 isn't prompt engineering. It's judgment. Knowing when to trust the output and when to override it. Knowing what data the model should and shouldn't see. Knowing when to use AI to draft something and when to write it yourself because the audience matters too much. Knowing that "AI told me to" is not a defensible explanation for a pay decision.

And here's what I think is the good news for People Ops leaders: judgment is exactly what you've spent years building. I'm genuinely optimistic about where this lands for HR — your AI fluency is going to grow much faster than someone with strong prompting skills but weak HR instincts. The technical part is the easy part to catch up on. The judgment part is what you already have. That's a real advantage and I don't think it's talked about enough.

What I think is actually worth your time in 2026

If I were a People Ops leader and could only do four things on AI this year, here's what I'd do:

1. Use one AI tool every day for one task. Build muscle memory. Stop reading about AI and start using it.
2. Read SHRM's State of AI in HR 2026 report. Real data beats LinkedIn vibes every time.
3. Have one honest conversation with your team about how they're already using AI. I think this is the most underrated move on the list. More employees are using AI at work than their managers realize, often without guardrails. Knowing what's actually happening is more valuable than any AI policy you could write.
4. Pick one strategic decision this year where AI changes how you'd approach it — workforce planning, comp benchmarking, headcount modeling, something with real weight. Not because AI will make the decision, but because using it on something that matters will teach you more than a hundred small experiments.

The part I don't think anyone is saying out loud

Keeping up with AI is not the same as being good at your job.

The People Ops leaders I think will matter most in 2026 are not the ones who tried every new tool. They're the ones who got really good at using a small number of tools to support real decisions, and stayed grounded in the human work that AI can't replace.

The pressure to keep up is real. But the pressure to keep up with everything is fake. It's manufactured by a content ecosystem that profits from making you feel behind. I see it every day. Don't let it set your agenda.

Pick your sources. Pick your use cases. Build judgment. Trust that you're doing the work, even on the weeks when it doesn't feel like it.

That, I think, is the actual skill.

‍