xMap AI

xMap LLM

xMap LLM is the AI step of a flow. You tell it what to do in plain words, and it reads everything the blocks before it found about each place: name, category, rating, address, details, reviews, scores. Set to Each item, it writes new fields on every place, such as a fit score and a reason. Set to The whole list, it writes one report about all of them for an Email or Slack message after it.

In the editor: Tell the AI what to do in plain words. It reads everything from the blocks before, plus any reference files you add. Each item: it writes new fields on every place. The whole list: it writes one report about all of them, for an Email or Slack message after it.

Settings

Work on

Pick one

One answer per place, written into new fields on it.

Each item
The whole list

Task

Pick one

What kind of answer you want. Each task starts with its own example instructions and field names, which you can change.

Score & rank A score from 0 to 100 for each place and a one-sentence reason. Use Filter or Sort on the score after it.
Match products Which of your products suit each place, and a first order size. Add your catalog as a reference file.
Write summary A short summary of each place for a sales rep or a manager.
Read reviews Reads each place’s latest reviews and says what customers praise, what they complain about and whether it is getting better or worse, with a score of how happy recent customers are. Put POI Intelligence before it with Reviews on.
Classify One of your labels for each place, and why. The first field is always one of the labels you list.
Extract fields Pulls set pieces of information out of each listing and your files, and leaves a field empty when it is not there.

Labels to choose from

Text

Comma-separated.

Needed before the flow can run.

Instructions

Text, several lines

What the AI should do, in your own words. Say what a good answer looks like, for example what makes a place a good customer for you.

Needed before the flow can run.

Reference files

A setting

For example your product catalog or price list. PDF, TXT, CSV, Markdown or JSON, up to 3 MB each. The AI reads their text.

Reference notes (optional)

Text, several lines

Short notes the AI should keep in mind, such as your price bands, the areas you cover or the kind of customer you want.

Fields it adds

Text

Comma-separated field names, up to 12.

Needed before the flow can run.

What comes out

Each item that leaves the block carries these fields, to use in the blocks after it as {{field}}.

Work on: Each item

Every place keeps all its fields and gets the fields you name in Fields it adds (these are the defaults of each task).

fit_score: Score & rank: a number from 0 to 100. Any field whose name has score, fit or rank in it is a number from 0 to 100.
reason: Score & rank and Classify: one short sentence on why.
products: Match products: the products that suit the place.
first_order: Match products: a suggested first order.
summary: Write summary: a short summary for a rep.
review_score: Read reviews: how happy recent customers are, from 0 to 100.
praise, complaints: Read reviews: what customers like most, and what they complain about most.
trend: Read reviews: whether the latest reviews are getting better or worse than the earlier ones.
label: Classify: one of your labels, exactly as you wrote it.
phone, website, opening_hours: Extract fields: what the listing and your files say, or empty.

Work on: The whole list

The places pass on unchanged. The report is a run value for the blocks after this one.

report: One plain-text report about everything that reached the block. Put {{report}} in an Email or Slack message; an Email block shows it at the top on its own.

How to use it

Sales

New places every morning, only new

xMap LLM ranks each new place as a sales prospect, so the email lists the best ones first with a reason for each.

Food and drink

How customers rate competitors

Read reviews: the AI reads the latest 10 reviews of every competitor and says what customers praise and what they complain about. Sorted unhappiest first, the email starts with the places whose customers are looking for somewhere else.

FMCG distribution

New outlets to CRM

Match products: the AI reads each new outlet and your catalog, then suggests the products, a first order and a pitch for the rep.

Banking and payments

Risk label for new merchants

A payments company signs up small merchants and wants a first risk label before the onboarding call.

Task Classify keeps the label to your own list, and Branch sends the high-risk ones to a person first.

Healthcare

A summary of every new competitor

A clinic group in Melbourne wants its area managers to read two sentences about each clinic that opens near their branches.

Before: An analyst collected new clinics by hand and wrote notes a week late.

With the flow: Every Sunday each manager gets the new clinics near them, each with a short summary.

Marketing agencies

A weekly market note for a client

An agency running a gym brand wants one paragraph a week on new gyms in the client’s districts, not a spreadsheet.

Work on: The whole list writes one report; the Email block puts it at the top, above the list.

Recruitment

What each job posting asks for

A staffing agency wants the seniority, contract type and languages of every new posting in its city, in columns it can sort.

Extract fields with out set to seniority, contract_type, languages turns long posting texts into three tidy columns, and leaves a field empty when the posting does not say.

Telecom

Dealer prospects scored on your own rules

A telecom operator scores phone and electronics shops as possible dealers, using its dealer criteria document.

Before: Area managers argued over which shops to approach.

With the flow: Every shop gets the same score from the same written rules, with a reason anyone can check.

Hospitality

A note on what opened near each hotel

The report reads every new place at once and says which hotels have the most new options for guests.

Good to know

Each item: up to 500 places a run, read in batches of 25. Places past that go on without answers, and the step says so.
The whole list reads up to 300 items for one report, and says when there were more. The items pass on unchanged.
It adds up to 8 fields. Fields with score, fit or rank in the name are numbers from 0 to 100; the others are short text. With Classify, an answer that is not one of your labels is left out.
It uses only the data of the flow, your notes and the text of your reference files (PDF, TXT, CSV, Markdown or JSON, up to 3 MB each; about 40,000 characters of text from up to 10 files are read). It is told to say so when something is unknown rather than guess, so check the reasons now and then.
Pictures and internal IDs are not sent to the AI, and long texts are cut at 300 characters, so it reads each place quickly. Reviews are the exception: Each item reads them whole (up to 3,000 characters a place), a whole-list report the first 600 characters of each place’s.
Read reviews sends only the places that have reviews; the others pass on without answers and cost nothing. With no reviews at all (Reviews off in POI Intelligence) the step stops and says so, before any credits are used.
In your account every run uses xMap AI for real, test runs too. The home-page builder and the editor’s preview show sample answers.
AI answers are not live data: a run on sample places still says "Preview data" in its emails.
Credits: 5 for each place it answers (a whole-list report counts each item it reads), plus 2 for every 500 tokens the AI reads and 2 for every 100 it writes. Long texts, many fields and reference files are read, so they cost more; the step says how many tokens it used.

Works well with

Map business scan: Finds businesses on online maps live: name, category, rating, address, phone, website and whether it is open.
POI Intelligence: Adds details to each place found by Map business scan: categories (cuisine), price level, opening hours, phone, website, amenities and a short description, and if you want them its latest reviews.
Filter: Keeps only the items that match a rule.
Sort and limit: Puts the items in order, and can keep only the first ones: the 10 best rated, the 20 nearest, the highest fit scores.
Branch: Splits items in two.
Jev decision: Asks TypeSafe’s Jev model one typed question about each place: yes or no, pick one of your options, or a score on your levels.
Place classifier: Assigns clean categories and merges duplicate places.
Email: Sends one email per run with all the items, or one email per item.

Templates that use it

New outlets to CRM: New restaurants and cafes as leads in your CRM every morning, with a brief for the reps.
Shops that just closed: units to rent: Shops that closed this week, each a unit that may soon be free for your next store.
New places every morning, only new: New places of your type in your city every morning, only the new ones, ranked by xMap AI.
How customers rate competitors: How customers rate your competitors, every week, the weakest first.
Where to open next: Areas of your city scored for a new branch, every week, on a map.
Score new leads from a Google Sheet: Each lead added to a Google Sheet is found on the map and scored by xMap AI.
New homes with no supermarket yet: New apartment buildings with no supermarket nearby, scored as places for your next store.
Decision makers at new cafes: New cafes scored by xMap AI, with the owner and managers from Apollo.
Growing companies and who buys there: Growing companies, sized with Apollo, with their purchasing managers.
New businesses in my territory, researched, into HubSpot: Every new business in your territory, scored, with the decision maker, straight into HubSpot.
Morning call list with decision makers and call notes: The 20 best clinics to call today, with the owner or director, in a sheet.
New businesses routed by size: key accounts to Slack, the rest to a sheet: New businesses scored and routed by size: the established ones to key accounts, the rest to inside sales.
Enrich a lead sheet and push it to HubSpot: Add a business name to a sheet and get it enriched, scored and in HubSpot with an opening line.
Companies hiring buyers, and who to call: Companies hiring buyers this week, sized and scored, with their purchasing managers in HubSpot.
Changes at my key accounts, with what to do: What changed at your key accounts this week: closures, ratings, phones and websites.
More businesses like my best customers: Businesses that look like your best customers, near them, not yet in HubSpot.
Prospects not yet in my CRM: Supermarkets and groceries not in HubSpot yet, scored and added every week.
Score every neighbourhood for a new branch: Every block of your city scored for a new supermarket, the 20 best on a map.
Where my rival is and I am not: Every rival branch with none of yours within 2 km, with who lives around it.
Busy stations with no convenience store: The train and metro stations with the fewest convenience stores around them, scored for a new one.
Stores competing with each other: Your stores that sit within 1 km of each other, with the people around them.
Fair franchise territories for a city: Your city shared into six fair franchise territories by where people live, with a plan for each.
Rival openings and closures every month: Every rival branch across your cities, with openings and closures each month.
Weather-driven promotions per store: A promotion idea for each store when heat or rain is coming.

And 652 more in the template gallery.

Stories with it

The morning a rival slips, she knows (Phone repair)
Better beans for the cafes that need them (Coffee roasting)
A pop-up spot that sells out the drop (Sneakers and streetwear)
Every restaurant her cargo bike can reach (Fresh pasta)
One clean prospect list, every shop once (Organic cosmetics)
A villa priced on real sales, the same afternoon (Estate agency)
The credit committee starts every week on fresh sales (Mortgage lending)
Trouble at a borrower, read the same morning (Commercial banking)
The tenant outgrowing its floor, seen in its hiring (Office leasing)
No guest drags a suitcase past a closed street (Short-term rentals)
The city’s favourite kitchens, on the food floor (Shopping malls)
First broker on the list at every new estate (Mortgage broking)

Help articles

Build your first flow in ten minutes: From a blank page to a flow that emails you a ranked list every morning.
Enrich a list once: A CSV of addresses or companies back with phone, website, rating and an AI score.
Score, rank and label with xMap LLM: A score and a reason for every result, in your words.
A written report about the whole list: One summary of everything a run found.
The Market analyst: a weekly brief: An AI employee that reads what your other employees found and writes the brief, with charts and plain findings.
Write to people with Outreach: A first message and follow-ups to each result, approved first, stopping when they reply.

Guides and terms

How to find new businesses in a city every morning: A daily list of newly listed businesses of a type, ranked, in your inbox.
How to build a B2B lead list from the map: Every business of a type in your territory, with contacts, scored, in a sheet or CRM.
How to score candidate store sites: Compare sites on residents, competitors and neighbours, with an AI score and a reason.
How to find companies hiring for a role near you: Hiring signals by role and city, with the company's details and the person to call.
How to send new leads to your CRM or sheet every day: A record per new lead, with its score and reason.
Location intelligence: Using where things are, and what is around them, to make business decisions.

Industries that use it

B2B sales · Retail & franchise · Restaurants & cafes · Real estate · FMCG & distribution · Marketing & agencies · Recruiting & staffing · Logistics & delivery · Government & cities · Investors & consultants · Healthcare & pharma · Hotels & tourism · Banking & lending · Energy & EV · Construction & facilities · Telecom · Automotive & mobility · Education & training · Agriculture & food supply · Fitness, beauty & wellness · Law & professional services · Media & publishing · Events & venues · Non-profits & humanitarian · Security & emergency services · Environment & climate · Ports & shipping · E-commerce & D2C · Manufacturing & industrial · Software & tech · Sports clubs & leagues · Pet care & veterinary · Home services & trades · Food delivery & quick commerce · Airports & aviation · Mining, oil & gas · Public transport · Water & waste utilities · Insurance · Fishing & aquaculture

Try it

Build a flow with xMap LLM on the home page, no account needed, or book a call. Try it live