Building Judgement in People Operations

One of my connections recently asked an AI tool to redesign a senior People & Culture role for 2030. The answer removed or automated a long list of work: administration, reporting, employee queries, recruitment, analytics, and parts of performance management. The skills that became more important were judgement, organization design, commercial thinking, influence, governance, and an understanding of human behavior.

That list matches what I see. It also raises a question the post didn't get to. If judgement is the skill that matters most, and the tasks AI is absorbing are the ones where judgement used to be built, where does the next generation of People leaders get it?

What judgement looks like in People work

Judgement in HR is rarely one big decision. It's a steady stream of smaller ones:

• Deciding whether a performance problem is about the employee, the manager, or a role that was never scoped properly.
• Knowing when a policy should hold firm and when applying it rigidly would cost you a good person for no real gain.
• Noticing that three resignations from one team in a quarter aren't a coincidence, even when each exit interview gives a different reason.
• Telling a founder that the reorg they want will fix their frustration and create a new problem six months out.

None of these have a clean answer in a handbook. They depend on context, pattern recognition, and a sense of how people will react or respond.

Where it has always come from

Most experienced People leaders built that judgement through volume. They answered the same benefits question fifty times and started to hear the anxiety behind it. They ran enough recruitment processes to spot a mis-hire forming in the second interview. They pulled enough reports by hand to notice when a number looked wrong.

That repetitive work was tedious, and it was also the training ground. When AI handles it, efficiency improves and a key learning path quietly disappears.

Building it on purpose

If the reps are going away, teams have to create the conditions for judgement deliberately. A few approaches that work:

1. Make reasoning visible. When a senior People leader makes a call on a sensitive issue, write down why. What did they weigh, what did they rule out, what would have changed their mind? A short decision log becomes a teaching tool.
2. Review cases together. Run regular sessions where the team walks through a real situation (anonymized) and each person says what they would do before hearing what happened. The disagreements are where learning happens.
3. Treat AI output as a draft to question. If a tool flags a flight risk or recommends a rating, ask junior team members to argue for and against it. Checking a recommendation against what you know about the person and the team builds judgement. Accepting it doesn't.
4. Put People staff close to the business. Commercial thinking comes from exposure. Invite HR into pipeline reviews, budget conversations, and product planning so they understand the pressures managers are under.
5. Give people ownership of real decisions early, with a safety net. Judgement develops when you are accountable for an outcome, not when you are observing someone else's.

Back to the design question

The original post ended by asking whether we would design our organizations the same way if we had today's technology from the start. For People teams, I think the answer is that we would design roles around decisions instead of tasks. We'd ask which calls this role is responsible for, what information it needs, and how the person in it gets better at making those calls over time.

That last part is the one most likely to be skipped. Automating the admin is the easy half of the work. The harder half is making sure the people left doing the thinking have had enough practice to do it well.