Letting AI Do the Work Without Losing Your Judgment
16 Oct 2026
Let AI do the work and keep the judgment.
Let AI do the work and keep the judgment.
Everyone has heard that advice. It has one flaw. Where do you think the judgment came from?
Nobody handed it to you. You built it by doing the work. The drafts, the numbers, the thing you got wrong on a Tuesday and never got wrong again.
That is the work now moving to AI.
Yesterday in Barcelona I put that question to a room of senior marketers from across Europe, at the Travel Summit of Criteo.
What stays human is the part you cannot hand over. Deciding what good looks like. Knowing when a number smells wrong. Saying no to something polished that is still wrong.
The room left with one habit for Monday morning. That one stays in the room.
Thank you to Anna Tamm Ortiz, Riccardo Russo and the Criteo team for the invitation and the wonderful summit experience.
Where did yours come from?
The habit that matters more than the AI tool itself
I did not share the Monday-morning habit publicly, and I will not share it here either. It belongs to the room it was built in. What I can share is the pattern behind it, because it comes up in almost every session I run with marketing and commercial teams right now.
AI is very good at producing something polished. Polished is not the same as right. The marketers in that room could all name a campaign, a deck, or a forecast that looked finished and was still wrong, because the AI tool had no way to know what "right" meant for their specific client or market.
That gap, between polished and right, is exactly where judgment lives. And judgment is built the same way it always was: by making the call, watching what happens, and adjusting. If AI produces the first version of everything, a marketer can go a long time without ever practising that call.
→ Before you approve an AI-drafted piece of work, ask what specifically you checked, not just whether it looked good.
→ Keep one part of the process manual on purpose, even when AI could do it faster.
→ Debrief the misses. A wrong number that got caught and fixed is worth more than ten right ones that were never checked.
I go further into how to build this kind of AI fluency in what senior leaders actually need from AI in 2026.
If you want a practical starting point for your own team, I have a free AI Tools & Tips Guide, the actual tools and workflows I use as a coach and leader. Leave your email here and I will send it over: https://www.thechangerepublic.com/free-resources