Finance & FP&A

How to build a monthly board financial pack with AI

Finance & FP&A3 AI tools7 steps6 friction points

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

A monthly board financial pack is a structured summary of your company's financial position — revenue, burn, runway, cash balance, budget-versus-actuals, and forward projections — packaged into something a board member can read in ten minutes and act on. For most operators, assembling it means pulling from three or four disconnected systems, manually reconciling numbers that never quite match, and reformatting everything into slides or a PDF before the board call. It's not intellectually hard. It is relentlessly time-consuming.

The workflow feels like an AI job because so much of it is pattern-matching on structured data: summarize this month's P&L, compare actuals to budget, write a narrative around the numbers, flag anything that looks off. If you've ever looked at a spreadsheet export and thought 'I know exactly what story this tells — I just need to write it up,' that instinct is basically right. Language models are genuinely good at interpretation, summarization, and formatting once the data is in front of them.

ChatGPT, Claude, and Gemini can all help meaningfully with this workflow today. They'll draft a board-ready narrative from a pasted P&L, suggest which metrics to highlight, reformat a table into a slide structure, or write the commentary paragraph explaining why burn spiked in month three. Where they fall short is everything upstream of the prose: connecting to live financial systems, pulling consistent data, and producing a result you can reuse next month without rebuilding the whole thing from scratch.

Skip the manual work → See the Starch version

Finance & FP&A3 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 income statement and cash flow data from QuickBooks, NetSuite, or your bank feed as a CSV or copy a table from the relevant report view. This is the manual step — no LLM replaces it.

2 Paste the raw data into Claude or ChatGPT with a prompt that specifies your board pack structure: revenue, gross margin, operating burn, net burn, and ending cash balance. Ask the model to extract and organize those numbers into a clean table.

3 Paste last month's budget or forecast alongside actuals and ask the model to calculate variance by line item and flag anything more than 10% off-plan. Review the output manually — models occasionally miscalculate, especially across multiple columns.

4 Write a context prompt telling the model what happened this month — a large one-time cost, a hiring pause, an unexpectedly strong revenue month — and ask it to incorporate that narrative into a two- to three-paragraph executive summary framing the numbers.

5 Ask the model to suggest the three to five forward-looking metrics or risks a board will likely ask about, given the financials you've shared. This is where LLMs are genuinely useful — pattern-matching against what boards typically focus on.

6 Use ChatGPT or Gemini to draft slide-by-slide talking points or a PowerPoint outline. Paste the structured output into Google Slides or PowerPoint manually; no LLM sends the deck directly.

7 Save your best prompt chain in a document. Next month, you'll repeat every step from the top — the model retains nothing between sessions.

Prompts you can copy

Here is our P&L for March [paste data]. Extract revenue, gross margin, operating expenses by category, net burn, and ending cash. Format as a clean table. Flag any line item that increased more than 15% month-over-month.

Here are our March actuals and our March budget [paste both]. Calculate variance by line item in dollars and percent. Write a two-sentence explanation for each variance over $10,000.

We had a $120k one-time legal settlement in March that inflated burn. Write a 150-word executive summary for our board pack that contextualizes this month's burn and explains what normalized burn looks like.

Based on these financials [paste summary], list the five questions a Series A board is most likely to ask at our next board meeting, and draft a one-paragraph answer to each.

Format the following financial data as a seven-slide board deck outline: cover, key metrics, P&L summary, cash and runway, budget vs actuals, risks and mitigants, asks. Write speaker notes for each slide. [paste data]

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 connection to your financial systems — every run starts with a manual export from QuickBooks or NetSuite and a copy-paste into the chat window.

Model context limits cap how much transaction detail you can include; paste a full month of line-item data and you'll often see truncation or hallucinated totals mid-table.

Formatting drifts between runs — the clean table structure you carefully prompted in February isn't reliably what you get in March, so you're re-prompting the structure every month.

Nothing persists — the model has no memory of last month's board pack, your preferred narrative tone, or the budget assumptions you've been working against all quarter.

Variance analysis requires you to manually supply both actuals and budget in the same session; if they come from different systems, reconciling them before pasting is on you.

You can draft the pack with an LLM, but distributing it — emailing the board, posting to a shared drive, updating a board portal — is entirely manual; the model stops at the text.

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 that handles this workflow against your live financial data, so assembling the board pack next month takes a fraction of the time the first one did.

Connect QuickBooks, Stripe, and Plaid once. Starch syncs your actual invoices, transactions, and balances on a schedule — so when the board pack agent runs, it's pulling real current numbers, not last month's manual export.

The Investor Reporting starter app pulls live burn rate, MRR, and runway from your connected Stripe and Plaid data, adds competitive context researched on the fly, and drafts the full narrative update — without you assembling it from scratch each time.

The Runway Analysis app gives your board a live view of burn and cash runway — calculated from real Stripe revenue and Plaid bank data, updated daily, with six-month history and 24-month projections — so the numbers in the pack are never stale.

Describe the exact pack structure you want in plain English — 'a seven-section board financial pack covering P&L, budget vs actuals, runway, and top risks, emailed as a PDF to my board list on the last Friday of every month' — and Starch builds and schedules that automation.

The output is a persistent app, not a one-off chat. Next month the same agent runs against fresh data with the same structure, the same narrative tone, and the same distribution list — with no prompt-rebuilding required from you.

Get closed-beta access →

Toolkit