MarginOS

Close the leaks.
See the margin.

MarginOS finds where a company lost money on a job: which jobs ended up unprofitable, where the difference came from, and what can be done about it — using the systems you already have.

Book an assessment call Open the synthetic demo

Thirty minutes, no commitment. The demo is public and uses synthetic data — no figure is a customer result.

What it does

  1. The morning exception queue

    Ordered by where the most money is tied up — not by whatever happened to catch the eye.

  2. The margin bridge

    Shows to the euro where the gap between quote and actual came from: hours, material price, quantity, rework, credits.

  3. Evidence for every claim

    Every figure opens into its source record: which posting, from which system, when. No guesses.

  4. Next step: an approved action, then verified

    A proposal goes for approval, and the outcome is checked afterwards. A number is not proof until it has been verified.

    Not in this version yet. The first version is read-only: it shows the gap and opens every figure to its source rows. The approval flow and the verification of the outcome come next, built with the first customers.

What we don't promise

We don't promise that AI will run your machine shop. We don't promise a savings figure up front, and we don't ship a black box whose numbers nobody can check.

The promise is narrow and measurable: we find the margin leaks and show what they rest on. The calculation is deterministic and traceable to source records, and your finance owner approves the costing rules before a single figure is displayed.

Who it's for

Project-based small and mid-sized companies where work is quoted on certain assumptions and delivered differently — and the difference is noticed only afterwards. The requirement is not the age of the system but whether jobs, hours, material and invoicing can be exported in some form.

The costing rules are fitted to your industry and your practice — that is part of the assessment call.

We are looking for the first 2–3 companies

We are building the first version for job costing in machine shops, and we are looking for two or three shops to define the costing with, walk through historical jobs, and measure the result. We are talking with other industries at the same time. The work is done together, and the source data remains your property.

Get in touch