How much did your bank AI pilot really save?
Inspect the sample story, then practise with a real example from your own work.
The question
Arun, an analyst at fictional Straits Bank in Singapore, helped run an AI drafting pilot for 40 relationship managers. The first slide claimed SGD24,000 of monthly savings.
| Monthly input | |
|---|---|
| Managers | 40 |
| Drafting hours saved per manager | 6 |
| Review hours added per manager | 2 |
| SGD per loaded hour | 100 |
| Tool cost SGD | 12000 |
| Correction cost SGD | 6000 |
Invented data. Time released is valued capacity, not automatically a cash saving.
- Action and result
- Arun rebuilt the calculation and told the sponsor before the expansion meeting. Expansion was held; the pilot continued at its existing size.
“Tell the story of changing the recommendation. What impact can Arun honestly claim?”
Not saved by this site. Your browser may restore text when you navigate back.
A strong answer, step by step
Open each step in turn. This is one strong answer, not the only one.
Step 1: Correct the claim
“Gross time value was SGD24,000, but net time value was SGD16,000. After tool and correction costs the pilot lost SGD2,000 a month.”
Why it works: The decision rests on net value rather than an attractive gross number.
Step 2: Explain your part
“I found the review and correction costs, rebuilt the bridge and raised the difference before the sponsor’s meeting.”
Why it works: A concrete action makes the speaker’s contribution inspectable.
Step 3: State the observed outcome
“The sponsor held expansion and kept the small pilot. I cannot claim a cash saving from a rollout that did not happen.”
Why it works: Avoided hypothetical cost is not a measured realised saving.
Step 4: Show what changed in your method
“I added a net-cost bridge and a customer-quality measure to future pilot reviews.”
Why it works: Learning is visible in a changed review practice.
The answer, in short
- Decision
- Claim the corrected analysis and the changed expansion decision. Do not claim realised savings.
- Because
- The evidence shows a recommendation changed, not a reduction in actual payroll or software spend.
- What would change it
- A later cancelled purchase order or measured expense reduction could support a cash-savings claim.
Common traps
- Repeating the gross savings figure.
- Claiming all of the sponsor’s decision.
- Calling a hypothetical avoided rollout a realised saving.
Check your answer
- Did I show the net calculation?
- Did I name my own action?
- Did I state the observed decision?
- Did I avoid invented savings?
Last reviewed Nov 10, 2026