Start the conversation somewhere else. Not with what AI can do, and not with what could be automated. Start with what each person on your team is genuinely good at.
That sounds soft. It’s the most practical move available, because the people most worried about AI are very often your best people, and they’re worried for a reasonable reason.
Why is my team worried in the first place?
Because the way AI usually arrives at work sounds like it’s about them.
Someone senior announces an initiative. A tool appears. Nobody asks the person who actually does the job what they think. From where they sit, that looks a lot like the first step toward not needing them, and the more skilled they are, the more they have to lose.
They also know things about the work that nobody else does. Where the hours really go, which steps matter, which shortcuts would quietly break something. When they push back, some of that is fear and some of it is expertise. Treat it as information.
What conversation should I have first?
Sit down with the person who owns the work, one at a time if you can, and ask in this order:
- What are you best at? What part of this job do you do better than anyone?
- What do you love doing? What would you never want to give up?
- What do you hate? What’s the time suck that drains you and isn’t why you took the job?
- What’s repetitive? What do you do the same way every week?
- If a tool could hand you exactly what you needed, when you needed it, what would that be?
The order matters. Asking what can be automated first tells people you’re looking for things to remove. Asking what they’re best at first tells them you’re looking for what to protect. Nobody loves every part of their job, and once people feel safe, the list of things they’d happily hand off tends to be long and very specific.
What should we build first?
The thing they said they hate.
Not the most impressive use case, and not the one a vendor demo made look good. The dull, repetitive task that eats someone’s Tuesday. When the first thing AI takes away is a job nobody wanted, people see what it’s for.
Look at the workarounds your team has already built, too. The spreadsheet with the formulas only one person understands. The checklist taped to a monitor. Those are their expertise, written down, and they’re often the bones of the system you should build. Building on their logic, rather than replacing it, is how someone goes from guarded to co-designing the thing. We wrote more about that in turning a spreadsheet into an app.
One real example: a small music studio founder went from six hours a week on Instagram to about 45 minutes of review. The system plans and drafts in the founder’s voice. The founder still decides what goes out. The hours that disappeared were the ones spent staring at a blank caption.
What do I say when someone asks about their job?
Be honest, and be specific.
Don’t promise that nothing will ever change, because work does change. What you can say truthfully is what you’re setting out to do: take the repetitive and administrative weight off people so they spend more time on the work that needs them.
It also helps to say how the tools actually work. AI is a magnifier. Put average input in and you get average output, faster. Put an expert’s judgement in and you get something nobody else could produce with the same software. The people who know the work are what make the tool worth anything. They are still the differentiator.
Do we need rules before anyone starts?
Yes, and short ones. People are more willing to try something when they know where the lines are.
- Which tools are approved, so nobody is guessing.
- What data can go in, and what never does.
- Where a human signs off before anything reaches a client or goes public.
- Who to ask when something isn’t covered.
A page is enough to start. If you want a template to adapt, ours is here.
What does it look like when it works?
People stop asking whether AI will replace them and start asking whether it can take the next thing off their plate. That shift usually happens with the first tool that saves someone real time on work they disliked, not with a presentation.
It also spreads on its own. A person who got their Friday afternoon back tells the rest of the team, and that does more than any rollout plan.
Where to start this week
Pick one team, one person, and one conversation. Ask the five questions above and write down what they’d hand off first. That list is your first project.
If you’d like help running those conversations across a team, or turning the answers into something people will actually use, book a free readiness call. Twenty minutes, no pitch. We also run hands-on AI workshops built around your team’s real work rather than generic demos.