How to read your business numbers with AI: a monthly report you can build yourself

The month closes, the profit and loss statement is sitting in QuickBooks or the point-of-sale export is waiting in a folder, and the owner closes the tab. There is no time to dig through the columns, and the numbers stay unread until tax season. The fix is plain: paste the figures into ChatGPT or Claude and ask them to tell the story.

This works because a monthly export is exactly the kind of structured text these tools read well. QuickBooks Online, the cloud version of QuickBooks, brings finances, workflows and reporting together in one place and generates financial reports from them, so the raw material is already there for most owners. No new tool is needed, just an existing AI subscription and about twenty minutes.

What to export

Start with the profit and loss statement for the month. In QuickBooks Online, open the Reports menu, choose the Profit and Loss report, set the date range to the month just ended, and export it as a spreadsheet or copy it straight from the screen. A POS usually has a monthly sales or item summary in its reporting section. Export the same report each month so the categories line up and the trends are real.

If the profit and loss alone feels thin, add a second export: a sales-by-item or sales-by-category report, or a list of job costs if the business runs jobs. Two or three small tables are enough. The point is a comparable snapshot, not a data dump.

What to keep out of the paste

Strip the private parts before pasting. Remove customer names, vendor names, addresses, bank account numbers and anything with a person's identity attached. Category totals and dollar figures by account are all the AI needs to spot a trend. If the export includes full transaction lines, summarize them first or delete the identifying columns.

This habit matters beyond one report. Safe-use ground rules, meaning a short list of what never goes into AI, are the same discipline a whole team should learn before anyone uses these tools on company data.

What to ask for

Paste the tables and ask in plain language for three things: the trends, the warnings, and a one-page summary. A prompt that works looks like this: "Here is my profit and loss for last month. Tell me the three biggest changes from the numbers, flag anything that looks off or worth checking with my bookkeeper, and finish with a one-page summary I could send to my accountant." Asking for a fixed structure keeps the answer usable and comparable month to month.

Treat the answer as a first read, not an audit. The AI is good at noticing that expenses in one category jumped or that revenue dipped in one week. It cannot know that a one-time equipment purchase explains the jump. Every claim worth acting on gets a human check against the source numbers.

Make it a monthly routine

The saved summaries become the real payoff. After three or four months, paste several summaries together and ask what has changed across the quarter. That is a question most owners never get around to answering, and it costs nothing but the time of the paste.

When the numbers stay unread anyway

Some owners build this routine on the first try. Others get stuck on the export step, or the prompt, or the follow-through, and the tab gets closed again. That is the kind of working task an AI Concierge call from AZ AI Lab can be spent on: a real person who learns the business holds two 1-hour working calls a month on real tasks, answers voice messages and email between calls with no limit on questions, and keeps the owner accountable for actually running the report. Everything built lives in the owner's own accounts and is theirs to keep.

The same skill can go to a team through AI Training for Business, which runs hands-on sessions of 90 minutes and up, built on approved real tasks, with take-home routines and a follow-up check-in. Most owners need only one AI subscription, such as ChatGPT or Claude, paid directly, and any cost is named before anything is bought.