Who is losing money in the neighborhood
Lists businesses from open sources (company registry, health and tourism registries, OpenStreetMap), reads each website and scores who has demand and is losing money. Writes the outreach message.
Margin X-Ray area · sell results, not AI
Show in hard money where the business leaks. Fix it and prove what came back.
The question stops being "do you need AI?" and becomes "where is the money you already earned going?". The Margin X-Ray kit covers a consultant's whole path: find the client with open data, run the diagnosis with the owner in 40 minutes, put the fix in place and measure month by month what came back to the till. Restaurant, hotel, clinic and salon, in Brazil, Latin America and worldwide. All in the browser, nothing to install.
Open, free kit (MIT). Guide and app in PT, EN and ES. Tested in a real pilot.
01 · The kit
One tool for each stage of the sale.
It doesn't sell AI. It doesn't sell an app. It sells less loss, more margin, more control and more recurring revenue.
Money per month, in margin
Filled in with the owner in 40 minutes: computes each leak per month (in margin, not revenue), orders them from quickest to hardest and prints the report.
Fix and prove
For each leak, a guide with a ready-made (SaaS) or open path, a checklist and before × after measurement in the Recovery Dashboard — what backs the monthly fee.
Hunter→Outreach→X-Ray with the owner→Proposal→Playbooks→Measurement→Monthly fee
Starting with the quickest leak — usually card fees — pays for the setup and buys trust for the rest.
02 · Leaks
Each sector is a package: questions, formulas, how much can be recovered and how to measure it.
There are 15 packages — 5 sectors × 3 markets. Another language is another market, with its own platforms, payments and rules (iFood and Pix in Brazil, Rappi and CoDi in Latin America, DoorDash and cards globally).
11 leaks
Marketplaces charging again for customers who are already yours, card fees, advance payments, interest, customers who don't return, low ticket, waste, unquoted purchases, menu, staff and ads.
The OTA's cut
Commission on returning guests, visibility programs that don't pay off, empty rooms in low season, guests who don't return, extras, no-shows, cards and cash.
Appointment-based services
No-shows, slots never booked, clients who don't return on time, per-client platform fees, add-ons and cards — and in clinics, claim denials and treatments that never start.
03 · From the neighborhood to the report
Eight steps, from lead to recurring revenue.
1 · List
The neighborhood's businesses from open sources — on the Collect screen or with one command.
2 · Enrich
By reading each website (respecting robots.txt); AI through your subscription reviews the signals.
3 · Check
In the Hunter, by hand, what the terms don't allow you to collect: Google rating, Instagram, marketplace.
4 · Reach out
With the generated message, which only cites what was checked and invites the owner to a 40-minute diagnosis.
5 · X-Ray
With the owner, marketplace and card-machine statements open. Save the diagnosis: it's the "before" baseline.
6 · Proposal
At most 3 playbooks. Setup spread over 6 months + monthly fee ≤ 1/3 of what's recoverable. Contract and data-protection consent in the kit.
7 · Implement
The playbook, with a ready-made or open path and what not to do.
8 · Measure and bill
In the Recovery Dashboard: before × after per leak, % of target, running total and a bonus only on what can be attributed.
# app in the browser (online or local) https://inematds.github.io/raio-x-margem/app/ # collect leads for a neighborhood python3 coletor/servidor.py # screen at http://127.0.0.1:8790/app/coletar.html bash coletor/rodar.sh --setor restaurante --uf PR --cidade Curitiba --bairro Batel --limite 30 --ia claude
04 · Within the rules
It could scrape — and why the kit doesn't.
There are open tools for scraping Google Maps and Instagram. The kit doesn't use them: the terms forbid it and the risk falls on you and your client. For every source there is a path that brings the same information without breaking the rules.
1 · OwnerThey export it from their own dashboard.
2 · Official registryCompany, health and tourism registries, OpenStreetMap.
3 · Indirect signalThe business website links its app, Instagram, booking.
4 · Official APIWith authorization.
5 · Manual check~1 min per lead, only for the best ones.
Data protection and professional councils. The kit doesn't use customer contacts from marketplaces and only stores business data. Clinics follow their council: dentists don't advertise prices or run referral rewards. Client data never leaves your machine.
05 · What came out of the pilot
Run in October 2026, in Curitiba and in the Serra Gaúcha.
417 leads
394 active companies in the registry + OpenStreetMap. Of the 30 enriched, half had no own ordering channel and only 2 had a loyalty program.
60 of 69 with a booking engine
197 in Cadastur, 383 lodging companies. The leak is the OTA's cut, not a missing channel.
318 health businesses
In the CNES registry: 174 practices, 81 clinics and 63 polyclinics, companies only.
Restaurant example: R$ 19,238/month leaking across 9 points, ordered by priority.
06 · Related areas
Other Events areas for people who sell and implement AI in businesses.
07 · Projects and courses
Find the client, sell the service and deliver results: all of INEMA's work in one place.
Projects and kits
Kit · PT · EN · ES
Margin X-Ray
Hunter, X-Ray, playbooks, Recovery Dashboard and sales kit.
Open the guide → ProjectProspector Agent
A pipeline for prospecting and selling websites.
Open the guide → Projectwacrm
Self-hostable WhatsApp CRM: follow up and win back customers.
Open the guide → ProjectAIV 2026
AI Visibility as a service (AEO/GEO).
Open the guide →Courses (in Portuguese)
Course
An AI Services Business
How to set one up, in 8 lessons, no coding.
Open the course → CourseAI Consulting Playbook
Selling with AI, part 2.
Open the course → CourseFrom Zero to the First Client
Selling with AI, part 4.
Open the course → CourseAI Consultant
From label to results.
Open the course → CourseProfessional 2027
AI implementer for small businesses.
Open the course → CourseCustomer Map
The DICA training.
Open the course → CourseCommercial Voice Agents
Sell the service.
Open the course → CourseSites and Apps
Build and resell.
Open the course →08 · Get started
The best first step: an X-Ray of a business you know.
Open the app, pick the sector and fill it in with the owner. Any field they don't know stays empty — it becomes a list of data to gather. Print the report and save the diagnosis.
Kit · open
Margin X-Ray
Hunter, X-Ray, playbooks, Recovery Dashboard, proposal, contract and data-protection consent.
Open the guide →
App · in the browser
X-Ray online
Works offline, nothing to install. Client data never leaves your machine.
Open the app →
Course
An AI Services Business
How to set up the business that sells this diagnosis, in 8 lessons.
Open the course →Not sure where to start? Talk to us on Telegram.