How it works
Install it on your servers, connect the data you already have, and get findings in money.
MARGINT is not a project. It reads what your organization already produces: ERP exports, production logs, procurement, inventory, payroll and attendance, quality. Against all of that it holds the professional knowledge of the role: what the number means, which benchmark it is measured against, and where the money is.
From installation to the first finding
Six steps. The headlines alone tell the story. The details are for anyone who wants to know what happens at each one.
- Step 1Installed in your environment
- Step 2Connect the data you already have
- Step 3Matched to professional knowledge
- Step 4The screens light up
- Step 5Ranked findings
- Step 6Where the money is
- Step 1
Installed in your environment
An on-premise server or a private cloud, inside your network.Nothing runs on our side and no data reaches us, not during the pilot and not after it. The system works without an internet connection.
- Step 2
Connect the data you already have
Export files from your existing systems, or a direct connection (API).ERP, production, procurement, inventory, payroll and attendance, quality, in the format they already come in (CSV, Excel). The pilot uses files only, with no access to source systems.
- Step 3
Matched to professional knowledge
Every field in your data is linked to its professional concept."Cost of goods sold", "overtime hours", "on-time delivery". Clear matches are recognized automatically. Anything that needs judgment is confirmed by a person, once, and the decision carries over to the next files.
- Step 4
The screens light up
The role's topics, and every metric with its formula, source and benchmark.Production and output, customer delivery, quality, inventory and working capital, suppliers and procurement, assets and downtime, workforce and productivity, cost and efficiency. A metric with missing data never shows an invented number. It stays off and says what is missing.
- Step 5
Ranked findings
The system scans by professional rules and raises deviations, ranked by severity.A supplier that is consistently late, a line whose output is eroding, inventory that is aging. Every finding comes with a recommended action and the source it came from.
- Step 6
Where the money is
Every finding is translated into currency.Management sees a short list: what the problem is, what it costs, and what to do this week. Not another dashboard. A list of decisions.
"What was gross margin this quarter?" The answer comes from the formula and the data, with its source. When a question has two meanings, the system asks which one you meant. It does not guess.
The pilot: one week, on your data
The pilot is a proof of concept, not an implementation. One week, on your server, under a mutual NDA.
What you bring
A server or a virtual machine, and export files from your existing systems. We tell you in advance which system and which fields, and what each one enables.
What we do
Install, match your data to the professional knowledge, and run. All on your side, done by us.
What you get at the end of the week
A list of findings ranked by financial impact, each with its source and a recommendation. Findings you can take to management as they are.
And after? If you continue, we move to implementation at the scope and pace that fit your organization. If not, the server is wiped, with everything on it.
The model: you pay for proven savings
Our interest and yours are the same: that your organization saves.
Not a license price. Not a per-seat price. A share of the result.
The pilot shows whether there is money on the floor. If there is not, we part without an invoice.
Every additional system you connect widens the picture, the findings and the savings. That is why we build for years, not for a project.
Questions that come up in every meeting
Do we need to send you data?
No. Everything runs on your side, including the pilot. There is no moment at which your data sits on a server of ours.
Which files do we need to start?
Whatever you have: ERP, production, procurement, inventory, payroll and attendance, quality. You do not need all of them. Each file switches on the metrics it can, and what is missing is stated explicitly.
What happens when data is missing?
The metric stays off and says what is missing and which system to bring it from. The system does not invent a number to fill a gap.
How long does it take?
Installation takes hours. Loading files depends on volume. Matching to professional knowledge is mostly automatic. The whole pilot: one week.
Does it replace our ERP or BI?
No. MARGINT reads from your systems and never writes to them. The ERP remains the system of record and BI remains the display tool. MARGINT answers the question they do not: what the number means in the profession, and where the money is.
How is it different from a dashboard?
A dashboard shows what someone configured it to show. MARGINT arrives with the role's knowledge built in: which metrics matter, how to calculate them and against which benchmark. It raises deviations on its own, and every number comes with evidence.
Who uses it in the organization?
Management. First and foremost the COO and the senior operations team. It is not an analyst tool: the screens speak the language of the role, not the language of data.
Do we need data scientists or analysts to run it?
No. The professional knowledge and the formulas come with the system. What the organization provides is export files, and a person who confirms matches when there is doubt.
Can the system be wrong?
Every number comes with the formula that calculated it, the file it was taken from and the benchmark it is measured against, so it can be checked. And when there is not enough data, the system says so instead of guessing. That is the difference between it and AI tools that answer every question with confidence.
Does it work in Hebrew?
Yes. Hebrew and English, on screen and in questions.
What happens to our data when the pilot ends?
If you do not continue, the server is wiped with everything on it. If you do, the data stays where it was: with you.
Book a session. Start saving.
An intro session, with no commitment and no data needed at this stage.