When HR Leaders Stop Talking About AI And Actually Use It

⏱️ Read time: ~2.5 min

My first offline workshop this year took me to Barcelona for the day! Sounds fancy, right?

Here's what I actually saw: the airport, the hotel conference room, and the airport again. Oh, and maybe a 10-second glimpse of the Sagrada Familia during landing if that counts.

Exhausting? Yes. Jet-set glamorous? Not quite.

Worth it? Absolutely.

My participants? The global HR leadership team of a mid-sized company. And what unfolded in that room reminded me why this work matters.

The Shift That Happens When You Stop Talking and Start Doing

We started the day the way most AI conversations start: with hopes and fears laid out on the table.

The hopes? Clarity. Productivity. Automation of the boring stuff. Personal efficiency.

The fears? Data security. Hallucinations. Information overload. Job displacement. "Are we even allowed to use this?"

Classic.

But here's the thing: those fears didn't disappear because I showed a perfect slide deck (though thanks to Gamma.app, my slides were clean and sleek).

They dissolved the moment we rolled up our sleeves and started experimenting.

First challenge (30 minutes): Use Copilot to simulate a recruitment process for an actual role the company is currently hiring for. With incomplete information. Go.

• One team built a competency framework and nailed the output format.
• Another generated job descriptions and interview questions in minutes.
• A third created a 9-day onboarding plan, then hit a wall when they asked AI to identify internal stakeholders (because, spoiler: their current setup doesn't have org chart access).

Second challenge (75 minutes): Simplify your company's HR policies by over 50%. Present your approach.

One team proposed a principle-based global framework with a local chatbot to answer employee questions. Another built a systematic AI agent to evaluate every policy against compliance standards and assign risk scores. The third focused on phased simplification with timelines and compliance checks.

Then I did something fun: I recorded the results presentation and used my own AI agent to score all three approaches and outcomes against predefined criteria.

That got the room buzzing with questions. How this worked, example use cases, and so on.

What Actually Changed

By the end of the day, the reflection session told the story:

• “If AI gets you to 70-80% so fast, that's huge."
• "Setting this (=Agent) up was way easier than I thought."
• "This could actually help us move from operational firefighting to strategic work."

The vibe had shifted. Not because AI became less complex or the risks disappeared, but because they'd used it in a structured but still experimental way. They'd seen what it could do, where it stumbled, and how it fit into their actual work.

Fear didn't vanish. It just stopped being the loudest voice in the room.

Oh, and I ran my own little experiment too: I had Hedy.ai, my super smart meeting assistant, join the session and take notes for us (with my participants' explicit consent, of course!). It's designed for online meetings, but it worked beautifully offline with the right context set ahead of time and a few minor tweaks in my facilitation. My written summary to the client? Sent as soon as I was through airport security with 10 minutes to review Hedy's output. More on Hedy very soon, I think it deserves its own deep dive.

Key Takeaway

You can talk about AI all day. You can read about it, debate it, worry about it.

But the real shift happens when you stop theorizing and start experimenting.

That's when the noise quiets down. That's when the possibilities become real.

What will you experiment with today?

Have fun!

Elena

P.S. I made a mistake. That was me, myself and I... not the AI. Two weeks ago I annouced that my first free session for the year would happen on Tuesday, 27 January. Well, that was a mistake. I am currently working through feedback from the sessions I ran in November and will be announcing what I have planned for this year shortly. Stay tuned.