Entry and follow-up steps
We do not sell a service catalogue. The entry is the automation audit with a written report — your concept foundation. Everything else arises as demand from that: the topics we have delivered and supported most often.
Request an auditEntry
Assessment of production, systems and data. Result: a written report — what pays off, what does not, in which order. The report is your concept foundation and is yours to keep.
Follow-up · Platform
Production, quality, inventory, asset management, energy management and NIS2 evidence on one data foundation. We dissolve data silos instead of adding a seventh system.
Follow-up · Edge
Our edge gateway family in three tiers. Time-series capture is included in every tier — from pure collector to AI inference at the machine.
Follow-up · Integration
Capture machine data without replacing the controller. Vendor-independent over OPC UA, Modbus, MQTT/Sparkplug and BACnet — delivered by a maintainer of the Node-RED OPC UA libraries.
Follow-up · Robotics
Turnkey palletising cells built on DOBOT cobots, validated in 3D beforehand and connected like the rest of your plant.
Extension · when needed
When the audit report points to a larger investment or multi-year modernisation: target state, options and economics as a document you can decide on. For most entries the report is enough.
A typical follow-up from the audit: the order sits in SAP, production happens on the shop floor. In most plants the gap in between is filled with spreadsheets, paper travellers and end-of-shift catch-up bookings. iniationOS closes it — in both directions and at the pace of production.
iniationOS in detail →iniationOS picks up the production order with quantity, due date and routing from your SAP — through the interfaces it already provides. SAP stays the system of record: no second place to maintain, no export file someone distributes each morning.
The order is broken down into individual jobs and assigned to the lines, cells and machines that actually run it. Every station gets exactly its part, in the sequence it has to be worked — not the whole order to sort out for itself.
Via OPC UA the job travels all the way to the machine: setpoints, variant and quantity are present where production happens, instead of being typed in from a printout. Machines without their own connectivity are reached through the SWIFT-BOX.
Good parts, scrap, downtime and run time travel the same path back into SAP — while production is running, not at the end of the shift. Anyone looking at the order in SAP sees the state of the shop floor: what is running, what has stopped, what is done.
We work vendor-independently and strictly on standards. Data modelling follows ISA-95 and the work of the IDTA, so your structure still holds when systems are replaced. The AI side runs on that same basis: models and agents work from the same described data — run locally wherever it must not leave the building. Which of this comes first at your site is in the audit report.
What management and IT ask before compute moves onto the shop floor — typically as a follow-up after the audit.
Computing happens where the data is created. The SWIFT-BOX sits at the machine, captures, historises and condenses before anything travels upwards. What arrives in iniationOS, in the ERP or in the cloud is already prepared — that saves bandwidth and recurring cloud cost, and the line keeps working when the link goes down.
At the machine. In the Compute stage, anomaly detection, quality prediction and image processing run locally — without cloud latency and without process data leaving the plant. Training happens where compute is cheap; execution happens on the line, at the pace of the process.
Down stays what arrives fast and in volume: millisecond readings, control decisions, raw images. Up go condensed KPIs, order confirmations, faults and events. We set that boundary during the audit — it later decides both your cloud bill and how fast the machine can react.
Three reasons speak against it. Latency — a control decision at the machine cannot wait for an answer from the data centre. Cost — permanently transferring and storing raw data from every machine is the most expensive route imaginable. And data sovereignty. The cloud stays useful for what it does well: cross-plant analysis and model training.
Describe your starting point in a few lines. You get an assessment of whether an audit is the right next step — and what the report would cover.
Request an audit