Board meeting is in 9 days. The deck is 40 slides covering financials, KPI dashboards, product roadmap, hiring, and the risk register. What’s missing is the one page directors actually reread on the plane: the narrative tying last quarter’s results to next quarter’s focus. That page is where board members form their take, which means it decides what gets debated and what gets nodded through. Most operators distribute the deck 5-7 days before the meeting, so the narrative is also the first thing a director reads about your quarter — before any context from you in the room.
This is a writing-and-compression job, not analysis. Your numbers are already reconciled; the AI’s job is to find the angle and hold the structure, not to add up the financials again.
TL;DR
- The 1-page narrative is the most-reread page in the packet. Write it first, then build the deck to support it — not the other way around.
- Feed the AI reconciled numbers + variance vs plan, the 1-2 surprises the board hasn’t seen, your 2-3 bets (each with a metric you’ll be held to), the one tradeoff, and last quarter’s narrative for tone match.
- A long-context model handles the whole job in one pass. As of June 2026, Claude Sonnet 4.6 and Gemini 3.1 Pro carry a 1M-token window standard, so you can paste the prior narrative, the board’s reactions, and the raw KPI dump together. ChatGPT Plus reads ~320 pages in-app (full 1M only on the $200 Pro tier).
- Board financials are confidential. Consumer ChatGPT Plus and Claude Pro may train on your chats unless you opt out — use a Team/Enterprise plan that contractually excludes training, or redact before pasting.
- Strip every hedge (“we may,” “are exploring”). The narrative is the founder/CFO’s read, in a confident voice, with numbers carrying the weight.
The task
Write a 1-page narrative (350-450 words) that opens the board deck. It connects what just happened (results, surprises, misses) to what comes next (the 2-3 bets and the single tradeoff). It is not a recap. It is the founder/CFO’s read on the quarter.
When this is the right job for AI
- Numbers are final and reconciled. AI is not auditing; it is narrating.
- You can give it the 3 things you wish the board would focus on (and the 1-2 things you would rather they not derail on).
- You have the prior quarter’s narrative for tone consistency.
- You want compression and angle, not raw analysis.
Which model, and the confidentiality line you should not cross
You do not need a frontier reasoning model for this — you need a long context window and a steady, plain voice. As of June 2026:
| Model | Context window | Why it fits this job |
|---|---|---|
| Claude Sonnet 4.6 | 1M tokens (standard) | Strongest at holding a consistent tone across a long prior-quarter paste; the default workhorse on Claude Pro ($20/mo) |
| Gemini 3.1 Pro | 1M tokens | Good when the deck and KPI export live in Google Workspace; included with Google AI Pro ($19.99/mo) |
| ChatGPT (GPT-5.5) | ~320 pages in-app on Plus ($20/mo); full 1M only on Pro ($200/mo) | Fine for the narrative alone; tight if you also paste a long KPI dump |
The longer windows matter because you can paste the prior narrative, the board’s last-meeting reactions, and the raw KPI export in one prompt instead of summarizing first — and summarizing first is exactly where tone and nuance leak out.
The privacy line is non-negotiable: board financials are material, confidential data. On consumer ChatGPT Plus and Claude Pro, your conversations may be used for model training unless you turn that off in settings. For real board numbers, use a plan that excludes training by contract — ChatGPT Team/Enterprise or Claude Team/Enterprise — or replace exact figures with [placeholders] before you paste and drop the real numbers back in afterward.
What to feed the AI
- The quarter’s headline numbers (revenue, growth, margin, burn, runway) + variance vs plan
- The 1-2 surprises (good or bad) that the board has not seen yet
- The 2-3 bets for next quarter, each with the metric you will hold yourself to
- The single tradeoff you made and want the board to understand
- Prior quarter narrative (so tone matches; board members notice tone shifts)
- The dynamic on the board: who pushes on growth, who pushes on burn, who pushes on team
Copy-ready prompt
Write the 1-page narrative that opens the board deck.
Last quarter results (vs plan):
- Revenue: <e.g. $4.2M vs $4.0M plan, +12% QoQ>
- Gross margin: <e.g. 68% vs 70% plan>
- Burn: <e.g. $1.1M/mo vs $0.9M/mo plan>
- Runway: <e.g. 19 months>
- 1-2 surprises board has not seen: <list>
Next quarter focus:
- Bet 1: <e.g. close 3 enterprise logos, target $1.2M ACV>
- Bet 2: <e.g. ship usage-based pricing in month 2, hold churn at 2.4%>
- Bet 3: <optional>
The tradeoff: <e.g. we are deliberately overhiring eng by 2 heads in Q3 to ship before competitor>
Board dynamic: <e.g. director A pushes on burn; director B pushes on enterprise traction>
Prior quarter narrative (for tone):
"""
<paste>
"""
Return:
1. A 350-450 word narrative with: paragraph 1 — the quarter in one read (what happened, what surprised), paragraph 2 — the 2-3 bets and what you will hold yourself to, paragraph 3 — the single tradeoff and why it is the right call.
2. A 1-sentence headline at the top — the version a board member would text to their partner.
3. Three predicted questions from the board, with the prepared 2-sentence answer for each.
Tone: confident, specific, no hedging. "We chose to" not "we may consider." Numbers carry the narrative — do not generalize.Sample output structure
Headline: Enterprise pull is real; we are doubling down on it in Q3 while accepting a 2-month margin dip.
The quarter: Revenue $4.2M (vs $4.0M plan) on the back of two unplanned enterprise renewals pulled forward into Q2. Gross margin compressed to 68%; supplier B’s renewal landed harder than modeled, and we absorbed it rather than passing through. Burn ran $200k/mo over plan because we accelerated three eng hires; runway still sits at 19 months. The surprise the board has not seen: 41% of new pipeline is now enterprise, up from 22% last quarter.
The bets: (1) Close three enterprise logos at $1.2M combined ACV by end of Q3; we will hold ourselves to two minimum. (2) Ship usage-based pricing by month 2 and hold logo churn at 2.4%. (3) Hire the VP Sales by mid-Q3; the search has two finalists.
The tradeoff: We overhired eng by 2 heads relative to the original plan. The reason: the competitive window on the integration platform closes in Q4, and shipping late costs us the enterprise pipeline that just opened. The cost is a 1-2 month margin dip in Q3 before the new ACV lands.
How to refine
- AI hedges. Strip every “we may,” “could,” “are exploring.” Replace with “we will” or “we chose to.”
- Narrative reads like a press release. Ask for the founder voice: “Rewrite paragraph 2 the way the CEO would email it to a trusted board member at 11 PM Sunday.”
- Numbers but no angle. Push back: “What is the take? If the board only remembers one sentence, what is it?”
- Three bets is too many; most boards retain two. Ask: “Cut to two bets; the third becomes a ‘we are watching’ line.”
Common mistakes
- Treating the narrative as a recap. Recap belongs in the appendix; the narrative is the founder’s read
- Burying the tradeoff. Every quarter has one; if you do not name it, the board will name a worse one for you
- Skipping the prior-quarter tone match. Board members notice when narrative tone swings from “cautious” to “confident” with no event to justify it
- Letting AI invent numbers. Every figure must come from the reconciled financials; no rounding “for narrative flow”
- Writing the narrative last. By then the deck has set the frame; write the narrative first, then build the deck to support it
FAQ
- What if the quarter was bad? Lead with the miss in paragraph 1, name the cause specifically, and use paragraph 2 to show what changes. Soft-pedaling a bad quarter erodes more trust than the miss itself.
- Should the narrative include forward financials? A single line: runway and the one number you will be measured on next quarter. Full forecast belongs in the financials section.
- Can I reuse this format every quarter? Yes. The structure is durable. Refresh the inputs, keep the three-paragraph shape, and the board learns the cadence. Consistency quarter over quarter is itself a board-reporting best practice — directors compare you against your own prior page, not a fresh template each time.
- Is it safe to paste the financials into ChatGPT or Claude? Not on the consumer tiers by default. ChatGPT Plus and Claude Pro may use your chats for training unless you opt out, so for actual board numbers use a Team/Enterprise plan that contractually excludes training, or write the narrative against
[placeholder]figures and substitute the real numbers locally. - Which AI should I use? As of June 2026, Claude Sonnet 4.6 (on Claude Pro, $20/mo) holds tone best across a long prior-quarter paste; Gemini 3.1 Pro fits if your deck lives in Google Workspace; ChatGPT on Plus works for the narrative alone but its ~320-page in-app limit gets tight once you also paste a KPI dump.
For the broader frame — why the executive summary is page one and what belongs on it — Bain Capital Ventures’ guide to creating an effective board deck is a solid operator reference.
Related
- AI exec summary — same compression pattern for management reporting
- AI investor update — the monthly version
- AI budget narrative — when the bets need budget support
- AI financial forecast — the forecast behind the bets
- AI KPI weekly report — track the bets between board meetings
- Decision Summary Prompts — adjacent decision-memo templates
- AI Pricing Strategy Memo: Tier Design, Anchor, Discount Floor
- AI Unit Economics Calc Sheet Narrative: CAC, LTV, Payback