Noukan Industries — homeNoukan IndustriesGmbH · Günsberg CH

Noukan Industries GmbH · Günsberg, Switzerland

A factory that runs in the dark still has to be patched, maintained and proven.

Three things stop manufacturers from going lights-out: security obligations they can show but not evidence, maintenance that depends on somebody standing there, and integration work that cannot be priced twice. Noukan Industries is building those three layers — in that order.

Status
Phase 00 — building
Next gate
Design partners, H1 2027
Documentation deadline
11 Dec 2027, EU/EEA market

The gap

Lights-out manufacturing is a maintenance problem, not a robot problem

The robots already work. What is missing is everything around them: the evidence that the software is safe, the warning before a spindle dies, and a unit of capacity a bank will finance.

01

Security that has to be evidenced

The Cyber Resilience Act makes security a property of the product, not of the marketing site. A manufacturer has to show a vulnerability process, a support period and documentation for every release — and a PDF written once a year does not survive an audit.

02

Maintenance that lives in one person's head

An unattended line fails differently from an attended one: nothing is noticed, so nothing is scheduled. The cells that run today survive on a technician who listens to the machine. That knowledge is not an asset until it is measured.

03

Integration priced as a project

A bespoke automated line is quoted once, delivered late and impossible to price twice. Mid-sized manufacturers are not refusing automation on the technology; they are refusing an unbudgetable capital project with an open-ended tail.

What we are building

Three layers, sequenced so each one pays for the next

Nothing here has shipped except the company itself. Each layer below is dated, and the roadmap states the condition under which we stop.

Phases 01–02

Building now

now → 11 Dec 2027

Aegis — continuous development pipeline
A CI/CD pipeline for products with digital elements. It runs on your real commits and tells you where the product is weak while the change is still cheap to fix. The conformity file the assessor reads is what that leaves behind, not a separate certification you buy.
  • Flaws surface on the merge that introduced them, not in an annual review
  • Self-assessment first: most products with digital elements do not need a notified body
  • Support period, patch policy and SBOM declared per product, per release
  • Details

Phases 03–04

Scheduled

2028 → 2029

Capture kit and fleet software
Retrofit capture hardware and the software on top of it, so maintenance on an existing robot fleet is scheduled against measured degradation instead of a calendar.
  • Retrofits onto cells you already own, without vendor lock on the bus
  • Prediction horizon and false-alarm rate stated, not implied
  • Fleet view with uptime accountability per cell
  • Details

Phase 05

Direction

2030 →

Standard production pods
Lights-out capacity sold as a standard unit. Small and mid-sized manufacturers get an unattended line they can finance as equipment, not survive as a project.
  • One SKU, quoted from a sheet
  • Maintenance included and priced against uptime
  • Repeatable by construction: the second pod costs less than the first
  • Details

Position, stated plainly

Aegis is a development tool. The file is what it leaves behind.

You do not buy Aegis in order to be certified. You buy it because every merge tells you where the product is weak — a dependency nobody is patching, an interface nobody documented, a support period nobody has declared — while the fix costs an hour rather than a recall. Most products with digital elements reach conformity through internal control without a notified body, and what an assessor asks for there is not a logo on a datasheet.

So the conformity file is a byproduct of developing this way, not the product. A certificate records that somebody paid for an assessment in March; a pipeline that regenerates the file on every merge records that the process ran, which is the thing the regulation actually asks for. Where a product genuinely falls into an important or critical category, that same evidence is what makes a notified-body assessment survivable, so the work is never either/or.

The obligation bites when a product is placed on the EU or EEA market, which includes products built in Switzerland and shipped north. Product category then determines the conformity procedure: check Annex III and Annex IV of Regulation (EU) 2024/2847 before assuming self-assessment applies. This is our reading of the regulation as a software vendor, not legal advice.

How we work

Four rules we would rather be judged on

  • 01

    Find the flaw at the merge

    A defect found in the commit that introduced costs an hour; found in the field it costs a recall, and found at an audit it costs a market. We optimise for the first one, and the conformity file is what that habit leaves behind.

  • 02

    Evidence over slides

    Every claim on this site that is not dated has a phase, a price and a stated reason we would stop building it. Nothing is offered as a photograph of a process on the day it was taken.

  • 03

    Standardise or stop

    The first unit is engineering. Anything we cannot quote twice from a sheet we do not sell as a product, however good the project is.

  • 04

    Measure before automating

    Dark factories fail on unplanned downtime, not on robotics. We instrument the cell before we propose removing the person watching it.

The company

Registered, small, and honest about which parts are future

Legal form
GmbH (limited liability company) under Swiss law
Registered
12 June 2026
Seat
4524 Günsberg SO, Switzerland
UID
CHE-130.523.906
Register number
CH-241.4.027.046-6
Share capital
CHF 25,000.00

Building something that has to run unattended?

We are taking a small number of design partners for the compliance pipeline in the first half of 2027. Bring a real product and a real CI system.

Start a conversation