“Will our order arrive Thursday?”
A reasonable question. You work here. You should be able to answer it.
The order is in one system. Stock is in another. The warehouse sent an update yesterday, but it went to someone else.
Ask Sam.
Sam is at lunch.
You’ve learned the systems. Done the training. You know where to click, which search works, and which status doesn’t mean what it appears to mean.
You are proficient in all the tools.
The customer is still waiting.
Knowing how to use everything hasn’t made everything work together.
You find the pieces, check that they belong together, and work out which source to trust when two of them disagree.
Then you answer the question.
Now you have AI.
Another window. Another login. Somewhere else to put the information you’ve just spent twenty minutes finding.
It writes a lovely email.
Thank you. But Thursday?
This is where I’d like us to pause before assigning everyone another training module.
People have already put considerable effort into learning the tools you gave them. If they still spend their days carrying information between those tools, what exactly are we asking them to get better at?
Some of that carrying is easy to overlook. The record says “responded to customer.” It doesn’t say “investigated whether Thursday still means Thursday.”
And because someone keeps finding the answer, the arrangement appears to work.
It does work. It has a person holding it together.
That person might have twelve other things to do.
Your people have become the adapters. Will AI help with that, or give them something else to adapt to?
Follow one ordinary question from beginning to end. Ask the person answering it to show you where they look, who they ask, and what they have to check again.
Pay attention to the pauses.
“Oh, I always have to…”
There it is.
Perhaps the warehouse update could appear alongside the order. Perhaps a changed delivery date needs to reach everyone who speaks to the customer. Perhaps two systems use different names for the same status, and everyone has learned to translate.
Some gaps need a technical connection. Others need a decision.
Does “available” mean sitting on a shelf, or already promised to someone else? Who can confirm a delivery date? When does an estimate become a commitment?
AI won’t settle those questions by being present. It might simply give us a very confident Thursday.
But an assistant with access to reliable sources could bring the pieces together, show where the information came from, and flag what still needs a human decision.
Now the person helping the customer has something useful to work with.
If Thursday is impossible, perhaps there’s another product, a partial delivery, or a conversation worth having before the customer discovers the problem alone.
That is useful work to have time for.
Check whether the new arrangement actually gives them that time. Has a step disappeared? Can they answer without chasing Sam? Or do they now check three systems and the assistant?
Your people learned the tools. They’ve been making the connections.
If we’re giving them AI, let’s give some thought to what we can finally take off their hands.