A while back, over coffee, someone asked me which clever AI trick had impressed me most this year. I disappointed them. The thing that stuck with me wasn’t a trick at all. It was a question I now ask myself most afternoons: of everything I did today, how much of it didn’t actually need me?
It’s an uncomfortable question, and it should be. Sit with it long enough and you start noticing how much of your week is just movement — chasing the same status update, rewriting the same kind of email, copying numbers from one place to another. Honest work, sure. But not the work only you can do.
The instinct to hold on
Most people, when they feel AI getting close to their job, pull their work in tighter. They guard it. They want to prove the human bit still matters. I understand the reflex, but I think it’s backwards. The value was never in the doing. It was in knowing what was worth doing in the first place.
The people I admire most are doing the opposite. They’re hunting for things to hand over. They treat every repetitive task as a candidate for handover, not a badge of how busy they are. When a reply lands that says the same thing they’ve written forty times, they don’t write it again — they ask Copilot in Outlook to draft it and spend their attention on the one paragraph that’s genuinely new. When a long Teams thread needs catching up on, they let Copilot summarise it and read the summary, not the forty messages.
That’s not laziness. That’s deciding where your judgement is actually worth spending. Every hour you claw back from the routine is an hour you can point at something that actually moves the business.
Hand it over once, properly
Here’s the part most people miss, though. Handing a task to AI once is a parlour trick. The real shift happens when you stop doing the task and start building the thing that does the task.
I’ve watched a business owner take the monthly client report — the one that ate a Friday afternoon every month — and rebuild it as a flow. Power Automate pulls the numbers, Copilot drafts the commentary in the same tone she’d use, and the whole thing lands in a SharePoint folder before she’s had her second coffee. She doesn’t make the report any more. She looks after the thing that makes the report.
And that’s the bit worth understanding. Once the work runs on its own, your job changes. You’re no longer the one producing the output — you’re the one watching the machine that produces it, and adjusting it when it drifts. When the tone goes a bit flat, you fix the instructions, not the document. When the numbers look off, you check the flow, not the spreadsheet. Your effort moves up a level.
That’s where the real return sits. Not in doing the task faster. In stepping back from the task entirely and tending the system instead.
What I’m watching now
None of this happens by accident. It takes a deliberate habit of asking, over and over, what you did yesterday that you shouldn’t be doing tomorrow. Most people never ask. They’re too busy doing. And the irony is the busier you are, the more of your day is probably the kind of work you could give away.
So I’ll leave you with the same uncomfortable question. Look back at yesterday. Find the hour a machine could have run without you. Then go and build the thing that runs it — and free yourself up for the work that actually needs a human in the chair.
CIAOPS 