Internal Comms & Meetings

How to run a monthly business review with AI

Internal Comms & Meetings3 AI tools7 steps6 friction points

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

A monthly business review is the ritual that keeps leadership, investors, and department heads oriented around the same set of numbers. It typically means pulling revenue, burn, headcount, and pipeline data from a half-dozen sources, synthesizing what happened, identifying what's off-track, and packaging the whole thing into a format a room full of people can actually absorb in forty-five minutes. For a small team, this can easily eat a full day — before the meeting even starts.

The workflow feels tailor-made for AI because most of the heavy lifting is analytical and editorial. You have data; you need narrative. You have bullet points from department heads; you need a coherent story. You have last month's numbers; you need context that explains why they moved. These are exactly the tasks where a capable language model — given the right inputs — can produce a solid first draft in minutes rather than hours.

ChatGPT, Claude, and Gemini can genuinely contribute here. Paste in a CSV of financial data and ask for a burn and revenue summary. Give Claude your department updates and ask it to synthesize themes and flag risks. Ask Gemini to draft the agenda or write the exec summary. None of this requires technical setup, and the output quality is high enough to use as a working draft. The challenge isn't capability — it's the plumbing between your actual data and the model.

Skip the manual work → See the Starch version

Internal Comms & Meetings3 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: download a CSV from QuickBooks or Stripe for the month, pull your bank transaction export from Plaid or your accounting tool, and open the file to check it loaded cleanly before pasting anything into an LLM.

2 Paste your revenue and expense data into Claude or ChatGPT (stay under the context limit — trim to the most recent 60-90 days if needed) and prompt it to calculate net burn, MRR, MoM growth, and runway at current pace.

3 Collect department updates via a shared doc or email thread. Paste the raw notes into Claude and ask it to identify the top two wins, top two risks, and any open decisions that need leadership input.

4 Open ChatGPT or Gemini and prompt it to write an executive summary paragraph that combines the financial snapshot with the department themes — give it explicit word count and tone guidance so it doesn't write a press release.

5 Ask the model to generate a meeting agenda based on the content, with time allocations for each section. Paste the output into your calendar invite or slide deck manually.

6 Use Claude to draft a written update version of the review — the kind you'd send to investors or board members. Paste in your exec summary, metrics, and department notes as inputs; ask it to write in first person and maintain a consistent tone across sections.

7 Review the outputs, correct any numbers the model misread or hallucinated (cross-reference against your original exports), and manually assemble the final deck or doc.

Prompts you can copy

Here is my company's revenue and expense data for March. Calculate net burn, ending MRR, MoM revenue growth, and estimated runway in months at this pace. Show your math.

Here are department updates from Engineering, Sales, and Customer Success for the month of March. Summarize the top 2 wins and top 2 risks across all three, and flag any items that need a leadership decision.

Write a 150-word executive summary for our March business review. Tone: direct, no spin. Include: MRR of $84k, net burn of $62k, runway of 11 months, and the fact that we signed our two largest customers this month but missed our outbound pipeline target.

Draft a monthly investor update email based on the following business review notes. Keep it under 400 words. Be honest about what's behind plan and why. Do not make it sound like a fundraising pitch.

Create a 45-minute meeting agenda for our monthly business review covering financial performance, department updates, risks and blockers, and a decision on our Q3 hiring plan.

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 data — every run starts with a manual export from Stripe, QuickBooks, or Plaid, and if you forget to update the file, the analysis is stale.

Context window limits force you to trim your data before pasting, which means you're making judgment calls about what to cut before the model even sees it.

Outputs are inconsistent month to month — the structure, tone, and metric definitions you carefully prompted in February aren't what you get in March unless you reconstruct the same prompt chain from scratch.

Nothing persists between sessions — the model has no memory of last month's review, so you can't ask 'how does this compare to February?' without pasting February's data back in manually.

Numbers the model calculates from pasted data can drift from your source of truth — small formatting inconsistencies in your export cause silent errors that require manual cross-checking after every run.

Assembling the final output — deck, email, agenda, and written update — still requires you to move between the LLM, a slide tool, and your email client manually, even after the AI does the drafting.

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 persistent software on your live business data. For a monthly business review, that means an agent assembles the apps, dashboards, and automations that pull your actual numbers, draft the update, and send it — without you re-running prompt chains from scratch each month.

Connect Stripe and Plaid once — Starch syncs your revenue and expense data on a schedule, so the Runway Analysis app always reflects real net burn and current runway, not last month's manual export.

The Investor Reporting starter app pulls live MRR from Stripe, burn from Plaid, and formats a polished monthly update with narrative and charts — you answer a few questions about what happened and Starch drafts the rest.

Scenario Analysis lets you model the decisions that come out of the business review — what happens to runway if you delay hiring or miss next quarter's revenue target — using your actual baseline numbers, not spreadsheet assumptions.

Describe the internal review format you actually want in plain English — 'a monthly business review dashboard showing MRR, burn, headcount, and pipeline by department' — and an agent builds it as a persistent app that stays current automatically.

Connect QuickBooks or NetSuite from Starch's integration catalog and the agent can pull entity-level financials — invoices, bills, journal entries — directly into your review without a manual export step.

Automations can be scheduled to run at month-end: pull the latest numbers, generate the draft update, and post a summary to Slack — so the business review prep starts before you even open your laptop.

Get closed-beta access →

Toolkit