The most useful AI habit I have seen in finance is simple: write down a good prompt once and reuse it. A shared prompt turns a personal trick into a team process. These five patterns are a good place to start. Use them only with data your company has approved for that tool.
1. First-draft variance commentary
"Here is a table of actuals against forecast by market. Write a three-sentence summary for a regional leader. Use only these figures. Name the two biggest drivers, and list anything you cannot explain from the data."
2. Risks and opportunities
"From this forecast summary, list the three biggest risks and opportunities against plan. For each, give the size from the table and the assumption it depends on. Do not estimate any figure that is not in the table."
3. Challenge my assumptions
"Here is my forecast logic. Act as a sceptical CFO. Ask the five hardest questions you would raise in a review, and say which assumption has the most influence on the result."
4. Explain it to the business
"Rewrite this finance note for a sales director with no accounting background. Keep every number. Explain each finance term in plain words the first time it appears."
5. Check the logic
"Review this bridge. Confirm that the steps add up to the total change. Flag any step whose label does not match its calculation, and any figure that looks inconsistent."
Four habits that matter more than the wording
- Give it the source. Paste the data and tell it to use only that data.
- Ask it to say what it does not know. This is the easiest way to reduce invented detail.
- Check every number. Treat the result as a draft from a new analyst.
- Share what works. Keep a team library, with a named owner for each prompt.
Good prompts rarely need clever wording. They need the same clarity you would give a colleague: what the task is, who it is for, what to use, and what a good answer looks like.
Newer: What a fast close really depends onOlder: AI in finance needs controls too