Build on finance judgement you can show: explaining a number, knowing where a process needs a decision, and being able to say why you accepted a result.
Isn’t learning to prompt enough?
It helps with asking. The harder part is knowing whether the answer is right.
1. Explain the number, then question it.
Practise. Define one familiar metric: period, population, currency, calculation and source.
Test. Reconcile a fictional result to its inputs, and investigate an unexpected change.
Show. Explain what the result supports and what it doesn’t establish.
Practical check: An attractive chart doesn’t supply its own business context.
2. Know where the work needs a decision.
Practise. Map one recurring report or reconciliation from input to accepted result.
Distinguish. A fixed rule, a missing input, a judgement and an authorized action.
Show. Recommend what to remove, simplify, automate or keep with a person.
Practical check: Choosing not to use AI can be a well-supported decision.
3. Explain why you accepted the result.
Practise. Use normal cases, ambiguous records and a case with missing data.
Check. Figures, references, omissions, code behaviour and exception handling.
Show. What you corrected or rejected, and how another reviewer can repeat the checks.
Practical check: Don’t present a practice example as a measured business result.
4. Leave a process someone else can run.
Handover. Explain the inputs, steps, review points, owner and fallback.
Learning support. Managers provide suitable time, tools and supervision, not just an expectation to use AI.
Your next example. One problem, your contribution, the checks, the result and the limits.
Read more: AI Skills Beyond Prompting: What a Business Team Needs · Using AI on Finance Spreadsheets Without Sharing Financial Data
This is general guidance, not a qualification or a promise about employment.