Investor Relations

How to build a board meeting deck with AI

Investor Relations3 AI tools7 steps6 friction points

Want an agent to handle this on your live data? See the Starch version →

A board meeting deck is one of those deliverables that looks like a design problem but is actually a data problem. You need burn rate, runway, MRR growth, pipeline, headcount, and key risks — all current, all consistent, all telling a coherent story. Most operators are pulling from four or five different sources, reconciling numbers in a spreadsheet, and doing a final pass in Google Slides the night before the meeting. It takes a full day when you count everything.

The workflow feels like AI should be able to help because so much of it is structured synthesis: take a set of numbers, add narrative context, format it for an audience that wants signal over detail. That's exactly the kind of task where GPT-4 or Claude can draft something useful in minutes rather than hours — especially the prose sections, the framing of wins and risks, and the executive summary that nobody wants to write at 11pm.

ChatGPT, Claude, and Gemini can each contribute meaningfully here. They're good at turning a bullet-point data dump into polished board-ready language, suggesting slide structure, writing commentary on metrics you paste in, and doing a gut-check on whether your narrative hangs together. What they can't do is reach into Stripe or Plaid themselves — you bring the data to them, not the other way around.

Skip the manual work → See the Starch version

Investor Relations3 AI tools7 steps6 friction points

AI walkthrough

How to do it with AI today

A practical walkthrough using ChatGPT, Claude, and other off-the-shelf LLMs — what they're good at, what you'll have to do by hand.

Tools that work for this

ChatGPTClaudeGemini

Step-by-step

1 Export your financial data manually: pull a CSV from Stripe for MRR and new customers, log into Plaid or your bank to get cash balance and recent transactions, and export last quarter's actuals from QuickBooks or your accounting tool.

2 Open Claude or ChatGPT and paste in your raw numbers with a framing prompt — tell it you're preparing a board meeting deck, what stage the company is at, and what the key metrics are. Give it the numbers directly in the message.

3 Ask the model to draft a slide-by-slide outline for the deck: opening summary, financial snapshot, growth metrics, pipeline or sales update, headcount and burn, key wins and risks, and asks. Use the output as a structural skeleton, not a final product.

4 For each section, feed the model the relevant data and ask it to write the talking-point narrative — what the number means, what trend it shows, and what you want the board to take away. Do this section by section rather than all at once to keep the context tight.

5 Use Perplexity or a separate Claude session to pull in any market context you want — competitor moves, industry benchmarks, recent funding rounds in your space — and paste that into your deck narrative manually.

6 Take the drafted text into Google Slides or PowerPoint and build the visual deck yourself. The AI has given you the words and structure; the layout, charts, and formatting are still manual work.

7 Do a final review pass: ask the model to read back your full narrative and flag any inconsistencies between the numbers you gave it earlier and the conclusions in the text. This catches the drift that happens across a long session.

Prompts you can copy

I'm preparing a Q2 board deck for a Series A SaaS company. Here are our key metrics: MRR $280k, MoM growth 6%, net burn $180k/month, runway 14 months, ARR $3.36M. Draft a 10-slide board deck outline with one sentence describing what each slide should contain.

Write the financial snapshot section of our board deck. Tone should be direct and confident. Numbers: cash $2.5M, net burn $180k/month, runway 14 months, MRR $280k up from $264k last month, 3 new enterprise logos closed this quarter.

Here is our risks section from last quarter's board deck: [paste]. We now have these new risks to add: [paste]. Rewrite the risks section so it's honest but not alarming, and ties each risk to a specific mitigation we're taking.

We beat MRR target by 12% this quarter but missed pipeline coverage ratio. Write a 3-sentence executive summary that leads with the win, addresses the miss directly, and ends with what we're doing about it. No spin.

Given these headcount and payroll numbers [paste], write 2-3 sentences explaining our burn trend to a board that knows we hired 4 people last quarter and is expecting burn to be higher than it is.

Reality check

Where this gets hard

The walkthrough above works — until your numbers change, the LLM hallucinates, or you have to re-paste everything next month.

No live data connection — every session starts with a manual export from Stripe, Plaid, QuickBooks, and wherever else your numbers live, then a copy-paste into the chat window.

Context doesn't persist between sessions. The careful framing and tone you established last quarter's board prep is gone; you're re-explaining your business and formatting preferences from scratch each time.

The model can't cross-check its own outputs. If the burn number in your financial slide contradicts the runway figure in your executive summary, you're the one who catches it — or doesn't.

Formatting and charts are entirely on you. The LLM hands you words; getting those words into slides with proper data visualizations is still manual work in Slides or PowerPoint.

Numbers drift if your session gets long. Pasting data early and asking follow-up questions 20 messages later often produces narrative that quietly misremembers the figures you gave it.

Nothing is automated. Next quarter you run the exact same prompt chain again — same exports, same paste, same back-and-forth — with no compounding benefit from the work you already did.

Tired of the friction?

Starch runs the whole workflow on live data — no copy-paste, no hallucinated numbers, no re-prompting next month.

See the Starch version →

Starch alternative

The same workflow on Starch

Starch is an agentic operating system — it builds and runs the persistent software your board prep actually depends on, connected to your live financial data. Instead of re-running prompts against last week's CSV exports, the underlying numbers are always current.

Connect Stripe and Plaid once — Starch syncs your MRR, cash balance, burn, and transaction data on a schedule so every time you open the Investor Reporting app, the numbers reflect what's actually in your accounts today, not what you exported last Tuesday.

The Investor Reporting app pulls live financial metrics, formats them into a structured update with narrative summaries and charts, and lets you add your own commentary on wins and risks — dramatically cutting the time from 'numbers are ready' to 'deck is ready.'

The Runway Analysis app gives you a live burn and runway dashboard that combines Stripe revenue with Plaid expense data — so the runway figure you put in your board deck is the same number you're managing the business against, not a separate calculation you reconciled by hand.

Connect QuickBooks or NetSuite from Starch's integration catalog and the agent can pull actuals, bills, and journal entries into your reporting surface — no manual accounting export needed before each board cycle.

Describe what you want in plain English and Starch builds it: 'Give me a quarterly board summary view that shows MRR growth, net burn, runway, top three wins, and open risks, refreshed from Stripe and Plaid every morning.' The agent builds that app; it runs continuously, not just when you remember to prompt it.

Presentation Agent — currently in development, request beta access — will take your Starch financial data and generate a complete slide deck from a text description, closing the last manual gap between live data and a formatted board-ready presentation.

Get closed-beta access →

Toolkit