An AI employee that writes a PDF report for you every month
A monthly report that gathers its own data, writes its summary and arrives as a branded PDF or slides. What goes in it, how to set it up, and how to keep it honest.
Part of the guide AI employees
An AI employee can write a monthly PDF report on its own: on the first of each month it gathers the data you asked for, an AI step writes a short summary of the whole list, and a report step lays it out with your logo and colour (the summary, the headline figures, charts, a map and the list) and emails the PDF or saves it to a Google Drive folder. You set it up once; after that, the report that took someone a day each month simply arrives.
This article is for anyone who owes a recurring report: an analyst sending a market update to directors, an agency sending each client a competition summary, an operations lead sending a network review, a charity sending donors an impact note. It covers what such a report should hold, the order of blocks that produces it, how to make the summary trustworthy, and the mistakes that make automated reports ignored.
What a monthly report should hold
A good monthly report answers three questions in the first minute of reading: what changed, how big it is, and what to look at. Everything else is supporting detail.
- A summary in plain words. Five to eight sentences on what moved this month and why it matters to the reader.
- A few figures. How many results, their average rating, how many scored as strong fits, how many cities. Not twenty tiles.
- Charts that sort, not decorate. The leading kinds and cities, as bars.
- A map, when the results have positions, because location reports are read faster on a map.
- The list, at the back, for the reader who wants to check a name.
Make a report in xMap Flows builds exactly these parts, each only when you include it and the items have it: the summary, the figures, bars of the top eight kinds and cities, a map, and the first 100 items. It can also make short 16:9 slides with one idea a page, or both files at once.
The order of blocks
- When: a Schedule on the 1st of every month, at an hour before the reader starts work in their time zone.
- Where: an Area of interest with the cities or areas the report covers, or My data with your own sites.
- What: the data blocks the report is about, such as Nearby places for competitors, Job postings for hiring, or Team findings for what your other AI employees found.
- Clean up: Remove duplicates, a Filter, and Sort and limit so the list is in the order you want read.
- Write: xMap LLM set to work on the whole list, with instructions for the summary. Its text becomes the report's summary.
- Lay out and send: Make a report, choosing a report, slides or both, what to include, and email or a Drive folder.
Writing instructions the summary can keep
The summary is the part people read, so it is the part to get right. Instructions for a whole-list summary work best when they are specific about the reader and strict about facts.
- Say who reads it: "for the expansion director, who has two minutes".
- Say what to compare: "say how many new competitors appeared in each city and which brands opened most".
- Forbid what you do not want: "use only the figures in the list; never add a number that is not in it; no recommendations about rent".
- Set a length: "no more than 150 words, no bullet points".
A worked example
Here is a made-up case. A coffee chain in Vancouver wants its directors to receive, on the first working morning of each month, a five-page view of the competition around its 30 stores.
Its flow reads the stores from My data, draws the area within 10 minutes by car of each with Travel area, finds cafes in each area with Nearby places, keeps only those it has not reported before with Only new, and has xMap LLM write a summary of the whole list for the directors. Make a report produces slides with the chain's logo: the title, the figures, what stands out, the top eight new competitors, and the leading kind of café. It emails them to the board and saves the A4 version, with the full list, to the strategy team's Drive folder, named after the flow and the date.
A charity in Johannesburg uses the same pattern differently: each month it counts the people living around each of its community sites, has xMap LLM write an impact note for donors, and sends a branded report. A third team, a logistics planner in Mexico City, sends a monthly PDF of road closures near its depots to the operations director.
Mistakes that get automated reports ignored
- Too long. A 40-page PDF every month trains people not to open it. Use slides for leaders and the full report for the team that works the list.
- The same every month. If the list does not change, the report should say so in one line. Use Only new or What changed so the report is about the month.
- No owner. Each report needs a person who reads it first and answers questions about it.
- Sent at the wrong time. Pick the hour in the reader's time zone, not yours.
- Hidden assumptions. When figures are estimates or sample data, the report and its email say so; keep that note in, rather than editing it out.
Once it works, deploy it and the AI employee keeps its appointment every month. If someone asks for a change (another city, a shorter summary), say it in a sentence to Ask xMap and the flow is changed for you.
Questions
Can the report carry my company's logo?
Yes. Make a report uses the logo and colour set for your organization in Settings, on the PDF and the slides.
Can it save the report to Google Drive instead of emailing it?
Yes. Choose To Google Drive and give a folder link or a folder name; without one it uses a folder named after the flow. Each item then carries the report's link for later blocks.
Who writes the summary at the top?
An xMap LLM block set to work on the whole list, placed before Make a report. Without it, the report still has the figures, charts, map and list.
Can it make slides instead of a document?
Yes. Choose Slides for short 16:9 pages with one idea each, A report for A4 pages, or Both for two files.
Templates that do this
Read next
More on ai employees and automation · Every article · Templates