Industry Insights
& News

From Data to Decisions: IIoT, Traceability and AI on the Shop Floor

Most production lines already generate the data needed to improve them. The problem is that it stays inside the PLC, on paper, or in someone’s memory.

When a line stops, the operator fixes it and moves on. The reason is rarely recorded accurately, and the duration is often guessed at the end of the shift. When a customer raises a quality complaint, the team searches through spreadsheets and paper records to find out which batch, which machine and which settings were involved.

IIoT and traceability close this gap. They capture what happens on the line automatically, link it to each part or batch, and present it to the people who need it. AI then helps those people find patterns faster. This article explains what that means in practice for operators and managers, and where AI genuinely helps.

What IIoT and traceability mean on a production line

IIoT (the Industrial Internet of Things) means connecting machines, sensors and inspection stations so their data is collected automatically. Traceability means linking that data to each part, batch or lot, so you can later prove what happened to it.

On a typical line, the data moves through five steps:

1. Capture. Data is read directly from PLCs, sensors, vision systems and checkweighers, using standard protocols such as OPC UA, Modbus TCP, PROFINET and EtherCAT.
2. Identify. Each part or batch carries an identity, such as a laser mark, a 2D code or a lot number, that ties it to its data.
3. Store. Every count, stop, measurement and reject is time-stamped and recorded.
4. Present. Dashboards show OEE, downtime and quality in real time to the people who act on them.
5. Act. Operators, engineers and managers use that information to fix problems, and quality teams use it to answer audits and complaints.

None of this requires replacing existing controls. In most cases, the data is already in the PLC and simply needs collecting.

How it helps operators

Operators gain the most immediate benefit, because the system removes guesswork and paperwork from their shift.

• Clear alarms that say what to do. Prioritised alarms with first-out logic show the root cause, not twenty knock-on messages.
• Stop reasons recorded at the moment they happen. The duration is logged automatically, and the operator picks a reason on the HMI instead of filling in a sheet later.
• Live targets on the line. A screen shows parts made against target, so the team knows during the shift whether it is on track.
• Immediate quality feedback. When a vision or measurement gate starts rejecting more parts, the operator sees it within minutes rather than at the end of the shift.
• Less manual checking and recording. Inspection and code verification run automatically, with failed parts rejected and logged.

The result is fewer arguments about what happened, and more time spent running the line.

How it helps managers and quality teams

Managers get facts instead of opinions, which changes where money and engineering time are spent.

For quality teams, traceability turns a customer complaint or recall from a broad investigation into a targeted one. If you can identify exactly which parts were affected, you contain less product and protect the customer relationship.

For investment decisions, reliable downtime and OEE data shows whether a new machine is really needed, or whether fixing the top two stop reasons would deliver the same capacity.

Where AI adds value, and where it doesn’t

AI is only as good as the data underneath it. That is why IIoT and traceability come first: they provide the clean, time-stamped, part-linked data that AI needs.

Once that foundation exists, AI helps in four practical ways:

• Vision inspection. AI vision can detect defects that are hard to define with fixed rules, such as surface marks, irregular shapes or variable natural products.
• Pattern finding. AI can spot links people miss, such as rejects rising after a material change, or a fault that recurs at a certain temperature or speed.
• Early warnings. By learning what normal looks like, the system can flag drift or unusual behaviour before it becomes a stop or a quality escape.
• Plain-language summaries. AI can turn a shift’s data into a short summary for a manager, for example: the top losses, what changed, and what to check first.

AI does not replace the engineer, the operator or good process control. It does not fix a poorly maintained machine, and it should not make safety decisions. Its job is to point people to the right problem faster, so they can apply their own expertise.

Worked example: one shift on a demonstration line

The dashboard below uses simulated data from a demonstration line running at 150 parts per minute. It shows how one screen answers the questions in the sections above.

By 14:30, the line is at 88% OEE: 93% availability, 96% performance and 99% quality. That beats the 85% target, but the dashboard shows exactly where the remaining losses are.

• For the operator: five stops have cost 33 minutes. Each one was logged automatically with its time, duration and reason.
• For the maintenance engineer: machine faults account for 13 minutes, or 39% of lost time. That is the first place to look.
• For the quality manager: Gate 1 rejected 310 of the 778 rejected parts, 40% of the total. Its trend shows whether this is normal or a process starting to drift.
• For the production manager: the line is 8,320 good parts behind its target to date. Downtime accounts for about 4,950 of those parts, more than speed losses and rejects combined.

With AI added, the system could summarise this automatically, for example: “Machine faults are the largest loss this shift. Gate 1 rejects rose after the 11:27 material change. Check the infeed and the new material batch.” That summary takes seconds to read, and every claim in it links back to recorded data.

Getting started: a single-line pilot

The lowest-risk way to start is a 60 to 90 day pilot on one line or station. It proves the value on your own equipment before you invest more widely.

1. Discover. We survey the line, on site or virtually, and agree the problems that matter most, such as downtime, rejects or traceability gaps.
2. Pilot. We connect to the existing PLCs and inspection points, then set up dashboards, stop-reason capture and alerts.
3. Measure. Over the pilot, we compare real OEE, downtime and reject data against the starting point.
4. Scale. If the pilot delivers, the same approach rolls out to other lines, products or sites.

A pilot needs only a clear problem to solve, access to the line, and a named contact on site.

About Gomes Technologies

Gomes Technologies is an industrial automation company based in Solihull, West Midlands, and a Beckhoff Solution Partner. We have more than 18 years of experience across aerospace, automotive, food and beverage, battery manufacturing and fulfilment.

We deliver PLC and HMI software, machine vision, robotics and IT/OT integration. Our products include NodeSenseAI for production monitoring and TraceSight AI for inline inspection and traceability.

To discuss a line assessment or pilot, contact us at sales@gomestechnologies.co.uk or visit gomestechnologies.co.uk.