Loading
MES, SCADA and the PLCs hold what happened. ERP, PLM and quality hold what was planned. We connect them without stopping production, so the gap between plan and actual closes in minutes rather than at month end.
Four systems · one work order
Where it breaks
The floor records what it made, the ERP records what it was told to make, and the two are reconciled by a person with a spreadsheet at the end of the month. Every decision in between is made on the older number.
MES and SCADA hold production reality, ERP and PLM hold the plan, and nothing joins them except a nightly extract and someone's judgement about which number to believe.
We carry plan and actual together on one work order model, so the comparison is a field rather than an exercise.
Machines emit more data than anyone reads, so it lands in a historian and waits. A drift that was visible on Tuesday becomes a scrap batch on Thursday and a root cause meeting the week after.
We stream and evaluate the signals that matter as they arrive, with alerts carrying the machine, the order and the context.
When a customer or an auditor asks which lots went into which batch, the answer is assembled by hand from several systems. Under time pressure the recall scope gets widened, because narrowing it takes longer than the deadline allows.
We record genealogy as production happens, so a recall query returns the affected lots rather than a cautious range.
How it works
We normalise signals at the edge, match them to the order they belong to, evaluate them against the plan and publish them with the genealogy behind them.
Scroll the diagram sideways →
OPC UA and MQTT from the floor, an ISA-95 aligned model in the middle, REST and events for the enterprise. When a machine or a line changes, you change a tag mapping and leave the architecture alone.
The runtime sits at each site so buffering, evaluation and alerting continue when the link to the data centre drops, and synchronise when it returns.
Tag health, missing signals and drift between plan and actual all land on one dashboard, so you spot a silent sensor before the variance report does.
Products
MetricMonitor and Health360 read the same work order and site model, so what the line reports and what the site reports line up.
Throughput, OEE, yield, downtime reasons and integration health across lines and sites, with alerts on the thresholds your production team already argues about.
Explore MetricMonitor →Occupational health records, safety compliance and incident reporting across multi-site operations, with the regulatory returns your HSE team files today.
Explore Health360 →How we work
Integration on a plant floor is a different risk from integration in a data centre. We work the way OT engineers expect: read first, write only where you have agreed it, and never with production depending on our availability.
Floor integration starts read-only. We agree, scope and test every write path back to a controller or an MES separately, and never bundle one into the same release.
The local runtime buffers and evaluates on site. If the link to the data centre drops, or we do, the line does not notice.
Whatever we replaced stays configured until you retire it. Reverting mid-shift means flipping a switch. There is nothing to restore.
Recent work
Anonymised at the client's request
A manufacturer ran MES on the floor and ERP in the office, joined by a nightly extract. Variance was a month-end conversation, and when a customer asked about a lot, three people spent two days assembling the answer.
We put an edge runtime at the site, mapped the tags that mattered rather than all of them, and matched every signal to the work order it belonged to. The floor systems were not modified and no line was stopped to do it.
Talk to us about your plant →Questions
We start read-only from the floor, one line or one cell at a time, with an edge runtime that does not sit in any control path. Nothing is written back to a controller or an MES until you have agreed that specific path. Lines keep running throughout, and there is no shutdown window in the plan.
Get started
Tell us what runs on your floor and what the enterprise never sees. An integration architect reviews it and comes back within two working days.
