You can't lead an AI rollout you haven't used
Leaders who've never handed AI a piece of their own work are judging pilots, budgets and vendors on demos. What changes when they use it, 3 tasks to try this week, and the 2 things that stay with the leader.
Picture a director who signed off the AI rollout. They have a licence, and they’ve used it twice: once to see what the fuss was about, and once to draft a leaving card.
Everything else they know about AI comes from vendor demos, a board paper and what their power users tell them in the corridor.
That director now decides which pilots get money, what the training budget is for, and whether the rollout is working. They’re making those calls without ever having felt the tool fail.
What changes when a leader uses it on real work
Calibration. A demo shows the tool on a task someone picked because it works. Your own work shows you where it works and where it confidently makes things up. You need the second kind of knowledge to judge a pilot.
Better questions. A leader who has watched an AI summary drop the one caveat that mattered asks a vendor different questions. “What happens when it’s wrong, and who notices?” is a question you think to ask after it has happened to you.
Credibility. Teams can tell whether a leader’s enthusiasm comes from use or from a slide. “Use AI more” from someone who doesn’t is a hard instruction to take seriously.
3 tasks to try this week
All 3 on your own real work, none of them a demo.
- A research question you’d normally hand to someone. What changed in a regulation that affects you, or how 4 competitors describe the product you sell. Ask for a source for every claim, then open 2 of the sources and check them.
- A first draft of something you write often, such as a board update or a note to the whole company. Tell it who reads it and what they already know. Then compare it with what you’d have written, line by line.
- A data question about a spreadsheet you know well, so you can tell when the answer is wrong.
The third one teaches the most. Ask about numbers you know by heart, and watch how sure of itself it sounds when it’s off.
What stays with you
Using it yourself still leaves 2 things in your hands: deciding where the effort goes, and judging what comes back.
I wrote about that split for individual work in Most people are using AI back to front. It matters even more for leaders, because a leader’s habits get copied across a whole team.
Teams copy what you do
A rollout email tells people what the organisation wants. What the leader does every day tells them whether the organisation means it.
If you want your teams handing real work to AI, the most useful thing you can do this week is hand it some of yours, and tell people what happened, including the time it got something wrong.
I give talks to leadership teams on this, and on the split between the few people getting real work out of AI and everyone else: talks for leaders.