FlowsBook a callSign inCreate free accountSign up
AI employees and automation

An AI employee that watches the news for your market

Stop reading five trade sites every morning. An AI employee reads the feeds you choose, keeps the stories that matter to your market, places them on a map and sends a digest.

Part of the guide AI employees

By the xMap team · · 4 min read

An AI employee that watches the news reads the news feeds and news pages you pick on a timetable, keeps only the stories about what you care about, such as rivals opening, new regulation or big projects in your cities, places each story on a map when it names a place, and sends you a short digest with only stories you have not seen. It replaces the habit of someone skimming the same sites every morning and pasting links into a chat.

This article is for anyone whose work depends on local or trade news: expansion and strategy teams, sales leaders, analysts, PR and communications, property and investment teams. It explains what the employee reads, how it decides what matters, a worked example, the mistakes that flood a digest, and the blocks to build it in xMap Flows.

What it reads and what it keeps

The News block reads up to five sources per run. Each source is either a feed (the RSS or Atom address many newspapers, trade sites and company newsrooms publish) or a news page, whose headlines and links it reads. Only public pages are read: nothing behind a sign-in, and a page whose site asks robots not to read it is left out and said so.

For each story it keeps the headline, the link, when it was published, the source and a short summary of at most 300 characters. It never copies the whole article.

Two filters decide what stays:

  • Words. Keep a story when its title or summary has any of your words, such as "opening", "acquires" or a rival's name.
  • Meaning. Tell xMap AI in your own words what the stories should be about, for example "a retailer opening or closing a store in Spain". It reads each title and summary and keeps the ones that fit, which catches stories that use different words.

When a story names a city or a place, it gets a position, so it can go on a map or be checked against your own sites.

Turning stories into a useful digest

A raw list of matching stories is still too much. Three steps make it worth opening:

  1. Drop stories already sent: Only new knows each story by its link, so a story appears once.
  2. Rank by importance: xMap LLM scores each story from 0 to 100 against a short brief ("how much would this change our plans this quarter?") and writes one line on why.
  3. Keep the top of the list: Sort and limit keeps the ten that matter most.

A worked example

A made-up example: the expansion team of a furniture retailer in Madrid wants to know each morning when rivals or big landlords make moves in Spanish cities.

The employee reads four Spanish business and property feeds and one trade news page. The meaning filter is "furniture or home stores opening, closing or moving, and new retail parks or shopping centres in Spain". On a typical morning it might read 140 stories, keep 9 that fit, find that 3 were already sent, and score the remaining 6. The team gets the top five before nine: a rival's new flagship in Valencia scored 86, a retail park approved near Seville scored 74, and so on, each with its link and a pin on a small map. The figures are invented; the shape of the morning is real.

Mistakes that flood a digest

  • Too many general feeds. A national front page brings hundreds of stories a day. Prefer trade, regional and business sections.
  • Words that are too common. "New" or "store" match half the news. Use specific words, or the meaning filter.
  • No memory. Without Only new, the same story returns every morning until it ages out.
  • No ranking. Ten stories of equal weight are read as noise. A score and a reason let readers stop after three.
  • Expecting full articles. The digest carries a summary and a link, so readers click through to the source for the whole story.

The blocks, in order, and what reaches you

  1. Schedule: weekdays at 7:00 in your time zone.
  2. News: up to five feeds or pages, your words, and what the stories should be about.
  3. Only new: only stories not sent before.
  4. xMap LLM: a score and one line on why it matters.
  5. Sort and limit: the top ten.
  6. Email or a Slack message, with a map of the stories that name a place.

What lands in the inbox each morning is a short digest: each story's headline as a link, its source, when it was published, the score and the reason, and a map of where the stories are. The AI employee also leaves a note in the Team inbox, where anyone can ask it a question such as "which stories this week were about Valencia?". A monthly version with Make a report turns the month's top stories into a PDF for the leadership team. The Business openings in the news, for the business desk, Mergers, lawsuits and fines in the business news and Health news in my markets, on a map templates are ready-made versions for different teams.

Questions

Which news sources can it read?

Public RSS or Atom feeds and public news pages, up to five per run. Pages behind a sign-in, or that ask robots not to read them, are left out.

Does it copy the full articles?

No. It keeps the headline, link, date, source and a short summary of at most 300 characters.

How does it know a story matters to us?

You describe what matters in plain words. xMap AI keeps stories that fit and can score each one against your brief.

Will I get the same story twice?

Not with Only new in the flow: each story is known by its link and sent once.

Templates that do this

Read next

More on ai employees and automation · Every article · Templates

Put it to work

An AI employee in xMap Flows does this on live data, on its own timetable. Start from a template or describe it in plain words.