From Raw Numbers to Board-Ready Commentary with AI

By Christian Matumona, Kyootek AI Financial Controller

The board pack is where every other piece of the close converges into a single narrative, written for an audience that wants the conclusion before the detail. It's also one of the most templated documents a finance function produces — the structure barely changes month to month, only the numbers and the events behind them do. That combination, consistent structure plus variable content, is exactly what AI handles well.

This is the deep dive referenced in the Complete Guide to AI for Financial Controllers: how to use AI to draft the narrative sections of a board pack without losing the judgment that makes it credible.

What a board narrative actually needs to do

Before drafting anything, it's worth being explicit about the job a board narrative does. It is not a restatement of the financials — the board already has the schedules. It is an interpretation: what happened, why it matters, and what it implies going forward. AI is genuinely useful for the first part — restating what happened clearly and consistently — and far less useful for the second and third, which require knowing what the board actually cares about this quarter.

That split determines where AI belongs in the workflow.

The structure that works

Most board narratives follow a predictable shape, and giving AI that shape explicitly produces dramatically better drafts than asking for "a summary."

1. Headline — one sentence, the single most important fact this period.
2. Performance against plan — revenue, margin, and EBITDA versus budget and prior period, in plain language.
3. What drove the result — the three to five events that explain the headline number, in order of materiality.
4. Cash and liquidity position — current position, trend, and any covenant or runway implication.
5. Outlook — what the board should expect next period, flagged as a forecast, not a fact.

Feed AI this structure explicitly and it will follow it. Ask for "board commentary" without structure and you'll get generic prose that reads well and says little.

The prompt

You are drafting the narrative section of a board pack for a [INDUSTRY] company.

Summary financials this period vs. budget and prior period:
[PASTE: Revenue, Gross Margin, EBITDA, Cash Position — current, budget, prior period]

Material events this period, in order of importance:
[LIST 3-5 EVENTS — one line each, e.g. "Lost largest customer in EMEA, ~8% of ARR" or "Closed Series B, $12M, extends runway to Q3 2027"]

Example of prior board narrative tone (match this style and length):
[PASTE A PRIOR EXAMPLE]

Write the narrative using this structure: Headline (1 sentence) / Performance against plan (2-3 sentences) / What drove the result (1 sentence per event, ordered by materiality) / Cash and liquidity position (2-3 sentences) / Outlook (2-3 sentences, explicitly framed as forward-looking).

Constraints: total length 250-350 words. Do not state any figure not provided above. Do not speculate on causes not listed in the material events. Flag the outlook section clearly as forward-looking.

Run this with last quarter's actual numbers and events first. Compare the draft to what was actually presented to the board. The gap — almost always in the outlook section, where AI tends to be either too vague or too confident — tells you exactly what guidance to add for next time.

The review checklist

Before any AI-drafted narrative goes into a board pack, run it against four checks:

• Every figure traces to source. No number in the narrative should be one AI inferred or rounded differently than your schedules.
• The headline matches what you'd actually lead with. AI sometimes picks the most dramatic number rather than the most relevant one — these aren't always the same thing.
• The outlook is appropriately hedged. Forward-looking language should read as an assessment, not a commitment.
• Nothing material is missing. AI only knows the events you gave it. If something happened that you didn't think to mention, it won't appear — and a board narrative that omits a material event is a bigger problem than one that's slightly verbose.

This checklist takes under ten minutes once it's a habit, and it's the difference between a draft you can present and a draft you have to rewrite.

Where this fits in the close

Board commentary typically lands on Days 7-9 of the close, alongside variance commentary and the CFO status update — see the full sequencing in the Month-End Close Playbook. The same underlying discipline applies: AI produces a structured first draft from data you control, and a qualified professional reviews every figure and claim before it goes to a stakeholder who can't see the work behind it.

Controllers who reach this level of comfort with AI-assisted reporting — trusting the draft, owning the review — are practicing what the AI-Augmented Financial Controller framework describes: judgment retained, assembly delegated.