“Can’t you just add a link?”
Of course. Where should the link go?
Well, there needs to be a new page. With content. And a place in the navigation. Oh, and that dashboard people use to find everything? The page needs to appear there, too.
The requester may not know about the dashboard connection. Why would they? They know what they want people to find. The person responsible for the website knows what making it findable involves.
She asks questions, creates the page, connects the pieces, and checks that they work. From outside, she has updated a link.
Just.
“Just” makes a decision about the size of the work before the person who understands it has spoken.
The questions, the checks, the connections she holds in her head: all of it disappears into a small instruction. She still has to do the work. Now she may also have to explain why it takes time.
And you get paid for the work people understand you to do—not the work you actually do.
To be human is to understand and value, not to flatten and demand.
Now ask which part of that job you could automate.
There are probably several useful answers. But if you start with the ticket, you might automate a menu update and miss the page, the dashboard, and the questions that would have uncovered both.
AI needs those interdependencies made available somehow: through documentation, system access, rules, or someone explaining them. Buying AI doesn’t magically include everything your people know.
So ask your human. What does she check? What does she know to question? What happens elsewhere when she changes this?
That experience and judgment are part of what you’re paying for. Why leave their enormous value on the table?
Automate before talking to the human, and you’ve decided what her job consists of before asking her.
I understand the enthusiasm. Something that was difficult yesterday suddenly moves. You made it move! There should probably be a small parade.
It’s the confidence of a toddler with a new wagon. Equipped for work. Delighted with the results. Dad is outside, which also leaves the pantry conveniently unattended.
Dad, folded into a vehicle considerably smaller than Dad, has questions about capacity, steering, and the hill.
Both can see that the wagon moves. One understands more of the consequences.
A working demonstration is exciting. Before it becomes the plan, talk to the person who understands the work.
Do you want to help her?
Then “Here’s AI. Tell us what you can automate” needs a little more work.
Show her what’s possible through something she actually does. Ask how she approaches it. Explain, demonstrate, let her try, and help her evaluate what comes back. You cannot just say, “Yes, it’s possible. Now do it.”
That’s AI enablement. As she develops the fluency to decide what AI can help with, you gain a fuller understanding of her job. She knows the work. You bring knowledge of the capabilities. There is quite a lot to discover together.
Be honest about why you’re asking and how her contributions will be used. An invitation to explore should mean what it says.
Even if replacing a system, workflow, or role is the intention, why would you do that without understanding what it contributes?
And then, please, ask what she can now imagine.
Let her use her mind to offload the mundane and frustrating things to AI. Don’t ask her to shut it down once she’s handed over a list.
What has she wanted to investigate? What could she improve? What has she stopped suggesting because there was never the time, money, or help to try it?
She may have something considerably more interesting to offer than a faster Friday spreadsheet.
HI > AI. Humans, not machines.
Understand the humans first.