OPC UA & Modbus OEE & downtime Predictive maintenance

Custom industrial IoT software development

Appfront builds custom software that opens up your machine park: retrieving data from controllers and sensors via OPC UA, Modbus or MQTT, turning it into usable management information such as OEE and downtime reasons, and linking it to your maintenance and planning. For manufacturers who want to know why a line stopped, not just that it stopped.

What is industrial IoT software?

Industrial IoT software pulls data from the equipment on your shop floor and turns it into information you can act on. That differs from a general IoT platform, which usually involves standalone sensors measuring over the internet. Here you are talking to controllers that have been running for years, that speak their own protocol and that must not be disrupted. For the broader approach, we have a separate page on IoT platforms.

Most manufacturers already have the data. It sits in the PLC, in the SCADA system, in the historian and in the head of the operator who writes down why the line was down for twenty minutes this morning. What's missing is the layer that turns all of that into one story: why did the line stop, how often does it happen, and what does it cost?

Custom software adds that layer alongside the existing setup rather than inside it. The controller carries on doing what it does, you read along and add context the machine doesn't have: which order was running, which operator, which shift, which batch of raw material. See also our broader approach to custom software development.

Read along, don't intervene

The integration with your control system is read-only and runs through a separate network zone. Your machine keeps running as it always did, and a software fault cannot affect production.

Downtime with a reason

A downtime alert without a cause is just a number. We link every stop to a reason the operator selects in a few taps, so that after a month you can see which cause costs the most.

One definition of OEE

Availability, performance and quality are calculated slightly differently at every company. We define your version explicitly, so that figures remain comparable between lines and sites.

How we build your industrial IoT software

We start on the shop floor, with one line and one machine, not with a platform choice. Production, maintenance, automation and IT are involved early on, because otherwise the discussion about network access will come up halfway through the project anyway.

1
Discovery and scope

We map out which controllers, sensors and systems are in place, which protocols they speak and which data is already being recorded somewhere. We also decide which line will be connected first and why.

2
Design

We design the integration layer, the data model for machines, orders and downtime reasons, and the separation between your control network and your office network. That last point is a security decision, not a network detail.

3
Build & integrate

We build in short iterations and connect one machine first, with real data and real operators. Only when those figures match what the shop floor sees do we expand. A dashboard nobody believes is worthless.

4
Go-live and management

A controlled rollout per line or site, with guidance for operators on recording reasons. After that, ongoing management, monitoring of the integrations and expansion to new machines.

What industrial IoT software actually does

A discrete assembly line raises different questions from a continuous process plant. These are the features we deliver most often.

Opening up machine data

Integrations via OPC UA, Modbus TCP, MQTT or an existing historian, depending on what your control systems can handle. For older machines without an interface, we install a separate measurement instead of replacing the control system.

OEE and performance figures

Availability, performance and quality per machine, line, shift and order, based on your own definition. That way you see not only that a figure is dropping, but also whether the cause is a machine, an order or a shift.

Downtime registration

Automatic detection of a stop, followed by a quick choice from your own list of reasons on a screen at the machine. Short enough that operators actually do it, structured enough that you can act on it.

Maintenance & condition

Running hours, cycle counts and trends in temperature or vibration as the basis for condition-based rather than calendar-based maintenance. See also AI for predictive maintenance.

Integration with planning and ERP

Machine data only becomes meaningful once you add the order. We integrate with your production planning or ERP, so performance becomes visible per order and per item.

Alerting & follow-up

Notifications for deviations that genuinely matter, with an escalation route to the right person. Alerts that fire too often get ignored, so we set thresholds based on your own history.

Who we build industrial IoT software for

The challenge varies greatly with the type of production and the age of the machine park. These are the types of clients we build for most often.

Discrete manufacturing and assembly

Companies with production runs, changeovers and multiple operations per product. For them, changeover times and downtime reasons weigh heaviest, and the integration with the order is essential to make sense of the figures.

Process and packaging industry

Continuous lines where a short stop has knock-on effects further down the chain. See also our page on software for the packaging industry.

Companies with a mixed machine park

Workshops where a machine from the nineties sits next to a new one. They need an approach that chooses, per machine, between reading existing signals, adding sensors, or accepting that nothing useful can be extracted.

Maintenance departments and technical services

Teams that want to move from calendar-based to condition-based maintenance. Their need is fewer dashboards and more reliable running hours and trends per component. See also our IoT app.

Not sure about a large project yet?

Test your idea first: a working prototype in 1 day

With OneDayBuild, we make your idea tangible in a single day for €1,150, so you know whether further development is worth the investment. Decide to go ahead with the full build? Then we deduct the cost in full.

View OneDayBuild →

Technology and integrations

We build with a modern, maintainable stack and connect to the equipment and systems you already have, without replacing your control system. See also our broader AI and IoT solutions.

Node.js / Python / .NET React / Vue front end Time-series database PostgreSQL / SQL database OPC UA Modbus TCP MQTT and Sparkplug Historian integration On-site edge processing OT and IT network segregation ERP and planning integration REST and webhook integrations Role-based authorisation Audit logging Shop-floor screens CI/CD pipelines

Why Appfront for your industrial IoT software?

Appfront builds custom software and starts with your machine park as it actually stands, not with a platform that assumes modern equipment. Almost every factory has a mix of generations, and the question is, machine by machine, what data can reasonably be extracted and what it is worth.

We keep the integration with your control system read-only and separate from your office network. That is not a technical preference but a starting point: software that can touch production is a risk that outweighs the insight it delivers.

We start with one line and only expand once the figures match what the shop floor sees. Dashboards that don't match operators' experience get ignored within two weeks, and then the investment is wasted however good the technology is.

Also take a look at our broader services: custom software development and IoT platform development. Questions about your situation? Get in touch.

  • Read-only integration with your control system only
  • Separation between control network and office network
  • A per-machine approach instead of a single platform choice
  • Your own OEE definition explicitly documented
  • Downtime reasons an operator enters in seconds
  • Integration with orders and planning for figures you can act on
  • Also works with machines that have no modern interface
  • One line first, expanding only once the figures add up
  • Clear documentation that your own technical team can manage
  • Ongoing management and extension to new machines

Security and continuity in industrial IoT

An integration between your shop floor and your office network is one of the most sensitive links you can make. We set it up following the principle of separated zones, with traffic that can only flow one way and without the software being able to send commands to the control system. For organisations covered by the Cybersecurity Act, this is also part of the duty of care.

We build in line with the OWASP security standards and take into account the principles of the IEC 62443 series of standards for industrial automation. If you process data that can be traced to individuals, such as performance per operator, the GDPR applies and it becomes a conversation with your works council.

Continuity works the other way round here compared with office software: if our layer fails, your production must simply keep running. That is why we buffer on site and the connection restores itself. More about our approach: information security policy and CVD policy.

  • Read-only integration, no commands to the control system
  • Separate network zones between OT and IT
  • Principles from the IEC 62443 series
  • On-site buffering when the connection drops
  • Production keeps running if the software fails
  • Encryption in transit (TLS 1.2+) and at rest
  • Role-based access and audit logging
  • Building to the OWASP standards

Frequently asked questions about industrial IoT software

Answers to the questions we get asked most often about custom software for the shop floor.

Industrial IoT software collects data from the equipment on your shop floor, such as control systems and sensors, and turns it into management information: performance per machine and line, downtime reasons, running hours and trends. The difference from a general IoT platform is that here you are talking to existing control systems that speak their own protocol and must not be disrupted.

Almost never. Many control systems can be read via OPC UA or Modbus, and for older machines without an interface we install a separate measurement, for example on power consumption or a signal lamp. That gives less detail than a modern control system, but it does capture the most important fact: is it running, and if not, since when.

Not if it is set up properly. We keep the integration strictly read-only and place it in a separate network zone, so the software cannot send commands to your control system. We also buffer on site, so a fault in our layer has no effect on what happens on the floor.

OPC UA is the standard industrial protocol for exchanging machine data between control systems and higher-level systems, with built-in security and a structured description of what is being measured. If your control system supports it, it is almost always the wisest route. If not, we fall back on Modbus or a separate measurement.

According to your own definition, which we record explicitly. Availability, performance and quality are calculated slightly differently at every company, for example in how you treat planned downtime and changeover time. Until that choice is recorded, figures are not comparable between lines and sites, and discussions end up being about the definition rather than the performance.

Because a stop without a cause is just a number. You know a line was down for twenty minutes, but not whether that was due to a fault, a changeover, a raw material shortage or a break. We have the operator choose a reason from your own list at every stop, in a few taps on a screen at the machine.

That is usually the second step, not the first. Prediction requires a period of reliable measurements and recorded faults, because without history there is nothing to recognise a pattern in. So start with running hours, cycle counts and trends; these already make condition-based maintenance possible before predicting even comes into play.

We build custom software. Machine parks differ widely in age, protocols and what can realistically be extracted from them, and standard platforms assume modern equipment that far from every factory has. After an intake conversation, we decide together which line to start with and in what order to expand.

Ready to build your industrial IoT software?

Tell us which machines you have and which question you can't currently answer: why a line stood still, what an order really cost, or when a component is due for replacement. We're happy to advise on integrations, network segregation and which line to start with. In a no-obligation first conversation, you'll get a clear picture of the possibilities.

Edit content