AI for FP&A Capstone Exercise — AI for FP&A

What you'll learn

Apply skills from across this course to a single realistic FP&A scenario under deadline pressure
Produce a concrete, role-relevant AI-assisted deliverable — a rolling forecast update or a board-pack narrative — for a board pack scenario
Self-assess your output against the verification standard covered in the governance lesson

You have worked through planning and budgeting automation, AI-powered rolling forecasts, variance analysis, scenario modeling, board pack generation, and the governance framework that ties them together. The capstone brings those skills into a single high-stakes scenario: a rolling forecast update that has to be defensible in a board pack, prepared under a tight deadline with imperfect data.

FP&A work rarely happens with clean data and unlimited time. The test of everything covered in this course is not how it performs on a textbook example, but how it holds up when the forecast has to move by tomorrow morning, one input is uncertain, and the number has to survive a board member's question. This exercise places you in exactly that situation.

Capstone Exercise

Q3 Rolling Forecast Update: From Statistical Baseline to Board-Ready Narrative

Context

You are an FP&A manager at a 250-person B2B software company. It is Thursday afternoon. The board meeting is Monday morning, and the CFO needs the updated Q3-Q4 rolling revenue forecast in the board pack by end of day Friday. Your planning platform has generated a statistical baseline forecast from the last eight quarters of data. You know two things the model does not: the company changed its standard contract term from monthly to annual billing five months ago, which shifts renewal timing in a way the baseline may not reflect, and the sales team closed two unusually large enterprise deals last month that are not expected to recur at that scale next quarter. The CFO has asked for the updated forecast number, a short board-ready narrative explaining the change from the prior forecast, and a note on the confidence level of the number.

Your Task

Draft a four-step prompt sequence you would use to guide an AI assistant through this workflow, plus one verification annotation for each step. Step 1: a prompt to review the statistical baseline output and identify where it may be extrapolating patterns that predate the contract-term change or the large enterprise deals. Step 2: a prompt to draft a business-adjusted forecast figure that accounts for both known factors, alongside the unadjusted baseline for comparison. Step 3: a prompt to draft a board-ready narrative (approximately 150 words) explaining the change from the prior forecast, in a register appropriate for a non-executive director audience. Step 4: a prompt to draft a one-sentence confidence disclosure that accurately reflects the uncertainty in the forecast without either overstating or needlessly undermining it. For each of the four steps, write one verification annotation describing the specific check a competent FP&A manager must perform before relying on that output.

Your notes (optional)

Deliverable

A four-step prompt sequence with a brief description of the expected AI output at each step, plus a verification annotation for each step explaining the specific check an FP&A manager must perform before the output is used in the board pack. The sequence should demonstrate the seasonality-break check from the rolling forecasts lesson, the baseline-versus-adjusted transparency principle from the scenario modeling lesson, and the certainty-overstatement check from the board pack lesson, applied together in one realistic scenario.

Quick check

Step 2 asks for the statistical baseline and the business-adjusted forecast side by side rather than one blended figure. What does the course say a blended figure costs you?

Select one answer.

Key takeaways

A defensible rolling forecast update depends on explicitly telling the AI what it cannot know from historical data alone — a contract-term change or an unusual one-time deal will not be visible to a statistical baseline unless the FP&A professional supplies that context directly. At a public company, the same forecast also feeds GAAP or IFRS-basis reporting and, where it informs external guidance, the forward-looking disclosure obligations securities law imposes — making an unverified number a compliance risk, not just an internal one.
Presenting the statistical baseline alongside the business-adjusted figure, rather than a single blended number, gives the board and the CFO the transparency needed to evaluate the judgment call rather than simply trust it.
Board-ready narrative quality depends on an explicit brief about audience, tone, and the specific instruction to avoid certainty-overstating language — the same discipline covered in the board pack lesson applies directly to forecast narratives.
Verification annotations are the FP&A professional's accountability record: they document what human judgment was applied to each AI-assisted output before it reached a board-facing deliverable.
Every skill in this course converges in the same place — AI accelerates the production of a forecast, a variance explanation, or a board narrative, but the FP&A professional remains the one who has to be able to defend the number in the room.