How to Use AI for Board Paper Preparation
I used to spend 3 hours preparing a single board paper.
Research. Structure. Draft. Rewrite. Format. Check the numbers. Rewrite again. By the time it was done, I'd burned half a day on a document that would get 15 minutes of discussion.
Now it takes 20–30 minutes. Same quality. Often better.
That's not a pitch. That's what happened when I started using AI properly — not as a toy, but as a working tool with structure behind it.
Here's exactly how I do it.
Why Most Executives Get AI Wrong for Board Work
Most executives try AI once. They open ChatGPT, type "write me a board paper about Q3 performance," get back something generic and unusable, and conclude AI isn't ready for serious work.
They're right that the output was rubbish. They're wrong about why.
The problem isn't the AI. The problem is the prompt. A one-line instruction produces a one-dimensional output. Board papers aren't one-dimensional. They have structure, context, tone, audience expectations, and institutional memory that a generic prompt can't capture.
The solution is system prompts — detailed instruction sets that encode how YOUR board papers should look, sound, and function.
The System Prompt Approach
A system prompt is a reusable set of instructions that tells the AI exactly how to behave. Think of it as a brief for a very capable but very literal junior analyst.
Mine covers:
Document structure. The exact sections my board expects — executive summary, background, analysis, options, recommendation, financial impact, risk assessment.
Tone and register. Board-level language. Not academic. Not casual. Precise, direct, and confident. No hedging unless the uncertainty is genuine.
Formatting conventions. How we handle tables, how financial figures are presented, what goes in appendices vs. the main body.
Institutional context. Company strategy, current priorities, recent board decisions that inform the paper. This is the part most people miss — and it's the part that makes the output actually useful.
Anti-patterns. What NOT to do. No buzzwords. No passive voice. No burying the recommendation at the end. No 40-page papers when 8 will do.
I keep this as a markdown file. About 90 lines. I update it every quarter as board expectations shift.
Step-by-Step: Drafting a Board Paper with AI
Here's my actual workflow.
Step 1: Gather Your Inputs
Before you touch AI, assemble:
- The raw data or source material (financial reports, operational data, market research)
- Previous board papers on related topics (for context and consistency)
- The specific decision or information the board needs
- Any constraints (timeline, budget, regulatory)
This takes 10–15 minutes. You'd do this anyway.
Step 2: Load Your System Prompt
Paste your system prompt into the AI session. This sets the context for everything that follows.
If you're using Claude, this goes at the top of the conversation. The AI now knows your document structure, tone, formatting rules, and institutional context.
Step 3: Feed the Source Material
Paste in your raw data, previous papers, and any relevant context. Be specific about what you want the paper to cover.
I'll say something like:
"Using the attached Q1 financials and the operational update from the GM, draft a board paper recommending we proceed with the fleet expansion. The board approved the capital allocation framework in December — reference that decision. Key concerns from the last meeting were timeline risk and labour availability."
Step 4: Review and Iterate
The first draft will be 70–80% right. That's the point. You're not expecting perfection. You're expecting a structured, well-reasoned draft that would have taken you 2 hours to produce from scratch.
Now you edit. This is where your expertise matters. You know the politics. You know which board member will push back on the timeline. You know the CEO wants the financial impact front and centre.
Step 5: Final Polish
Run the near-final version back through the AI with specific instructions: "Check this for consistency in financial figures. Flag any unsupported claims. Tighten the executive summary to under 200 words."
Total time: 20–30 minutes. Down from 3 hours.
What AI Actually Does Well in Board Work
Structure and formatting. AI is excellent at organising information into the right sections with consistent formatting. It won't forget the risk assessment section.
First drafts. Getting from blank page to structured draft is the hardest part of writing. AI eliminates the blank page problem entirely.
Consistency. If you tell it your conventions, it follows them every time. No more fighting with inconsistent table formats or varying heading styles.
Research synthesis. Feed it 50 pages of source material and ask for a summary of the three most important points. It'll do in 30 seconds what would take you 20 minutes.
Quantitative analysis narrative. Give it the numbers and ask it to write the story. "Revenue was up 12% driven primarily by the commercial segment, which grew 18% on the back of three new contracts signed in Q4."
What AI Cannot Do (Yet)
Political judgement. It doesn't know that the CFO and COO disagree on capex timing, or that the chair has a particular sensitivity to environmental risk.
Strategic intuition. It can structure an argument, but it can't tell you which argument to make. That's your job.
Relationship context. It doesn't know that the last three papers on this topic were rejected, or why.
Accountability. Your name is on the paper. You own the recommendation. AI helps you get there faster — it doesn't replace the judgement that makes the recommendation yours.
This is the key insight most people miss. AI doesn't replace you. It replaces the 80% of the work that doesn't require your expertise — the structuring, the formatting, the first-draft writing — so you can focus on the 20% that does.
Beyond Board Papers: Other Executive Documents
The same approach works for:
Committee reports. Audit committee, risk committee, remuneration committee. Each has its own system prompt with specific structure and focus areas.
Executive memos. Shorter, more direct. I have a separate system prompt optimised for 1–2 page decision memos.
Investor updates. Quarterly updates to investors. Consistent structure, data-driven, forward-looking.
Due diligence summaries. When evaluating acquisitions, AI can synthesise financial data, market research, and operational assessments into a structured DD summary in minutes.
Meeting preparation briefs. A one-pager before every board meeting summarising key issues, likely questions, and your recommended positions.
The Compound Effect
Here's what surprised me most.
The time saving on a single board paper is meaningful but not transformative. 3 hours to 30 minutes — nice, but it's one document.
The compound effect is what changes everything.
When you can produce board-quality documents in 30 minutes instead of 3 hours, you produce more of them. Your communication gets better. You have time to review other people's papers and provide feedback. You prepare more thoroughly for meetings.
The executives I've worked with who adopt this approach don't just save time. They become more effective. Better prepared. More responsive. More thorough.
That's the real return.
Getting Started
You don't need to build a 90-line system prompt on day one.
Start with one board paper. Write your usual brief to yourself about what the paper needs to cover. Feed that to the AI along with the source material. See what comes back.
Then iterate. Each time you use it, note what worked and what you had to fix. After 3–4 papers, you'll have a clear picture of what your system prompt needs to include.
Within a month, you'll have a reusable system that cuts your prep time by more than half.
Within a quarter, you won't remember how you worked without it.
I don't say that lightly. I've been doing corporate advisory for 20 years. This is the largest productivity improvement I've experienced in my career.
You feel it. I feel it.
I teach executives exactly this — the system prompts, workflows, and skills to use AI for real board-level work. Join the next cohort — 4 weeks, step-by-step, free.