An example I made up.
The situation
Gayatri Pillai is regional head of customer operations at Merbau Parts Pte Ltd. The company distributes spare parts for delivery vans. Its head office is in Singapore, and it has offices in Kuala Lumpur, Jakarta, Bangkok, Ho Chi Minh City and Manila.
Gayatri has two weeks to decide one thing. Should all five offices move to one refund rule from 1 March? Under the rule, front-line staff approve a refund for a returned part up to a set sum, with no supervisor. The aim is to pay customers back sooner.
What the AI said
She asks BarbtailAI, the planning assistant her team uses, whether one rule can work in every office. It compares the five offices on the figures they all record, and it answers in three short lines:
"One refund rule can serve all five offices. Return volumes and approval times are within 10 percent of each other across the five. I am 85 percent sure."
It's a clear answer, and the figures behind it are correct.
The check
Gayatri doesn't argue with the figures. She sends each country manager the same question: "Who has to agree before a refund is paid in your office, and who expects to be told?"
Four managers give the same answer. Manila's answer is different. In that office, the finance lead also signs refunds above a sum that is lower than the sum in the new rule. That's because the Manila office pays refunds from its own account and reconciles that account each week. None of the figures BarbtailAI compared shows this.
The decision
The rule starts on 1 March in Kuala Lumpur, Jakarta, Bangkok and Ho Chi Minh City. Manila keeps its finance sign-off for now. Its finance lead has two weeks to propose how the new rule fits that step.
Gayatri tells the five country managers and the regional finance director. She also tells the front-line team leaders, because their teams are the ones who will use the rule. Then she writes one line into the plan: "This plan compared what the offices measure. It did not ask who decides in each one."
Three months later
The four offices pay refunds faster. Manila has joined the rule, with its own finance step built in, and its weekly reconciliation still matches. Before any new regional rule starts, Gayatri now asks the same question first.
The figures in this example are illustrative.
Where I learned it
Here is my record. I've done 25 years of advisory and leadership work across Asia Pacific, the UK, the US and Australia. I'm Malaysian, and I'm based in Singapore.
I'm keeping this article to what that record shows. So there's no list of countries or employers here.
What it taught me
I'm careful not to pretend to be an expert in an industry I haven't worked in. Gayatri's question is close to that habit. She doesn't assume that an office she leads from Singapore works the way head office works. She asks the people who run it, before the rule starts.
The point of the example fits in one sentence. The same answer can be right in one office and wrong in the next, because who must agree isn't the same everywhere.
BarbtailAI's figures were close across all five offices, and a close match like that looks like a good number. I do not treat a good number as proof that everything is working. The figures told Gayatri that the offices were alike in what they record. They didn't tell her who signs a refund in Manila, or why.
Here is the line I use for this. AI can draft the plan. It cannot decide whether the plan is right.
BarbtailAI compared the offices correctly, and the rule fitted four of them as it stood. For the fifth, a person still had to ask who must agree and change the start for that office. Gayatri did that within her two weeks, and she told the people who needed to know. The rule reached all five offices in the end.
If you're about to apply one AI-assisted plan across several countries, ask the same question before you decide. Ask each office who has to agree before the plan takes effect there, and who expects to be told. Then decide, office by office, where the plan starts and where it waits for a local step.
What I won't claim
Gayatri and Merbau Parts are made up, and so are the figures. I don't claim to know how every market works. Nothing on this site measures or scores anyone.
I give keynotes, briefings and workshops on this.