Manufacturing and FMCG
Line intelligence that finds the loss before the shift report does.
On a packaging or process line the losses are small, frequent, and invisible until the month-end number. In-Sight watches every line signal at the rate it is produced, finds the drift, the micro-stop pattern, and the overfill before they become a write-off, and carries the fix through to a work order and a report that proves it held.
The loss that never trips an alarm
A filler that drifts two grams heavy never alarms. A compressor that cycles four percent more often never alarms. A pasteuriser that holds temperature but takes ninety seconds longer to recover after each changeover never alarms. Each of them costs money every hour, and each is visible in the data the line already produces.
In-Sight learns each signal's normal by product and operating mode, so the same filler running a different SKU is not an anomaly, and the same SKU running two grams heavy is. Statistical process control with Western Electric rules catches the slow drift; multivariate detection catches the combinations; trained models catch the patterns that precede a stoppage.
- Filling and net content: fill weight and volume, target and tolerance by SKU, giveaway per hour
- Packaging: line speed, micro-stops, reject counts, changeover duration and recovery
- Process: temperature and pressure holds, CIP cycle time and chemical use, batch consistency
- Utilities: compressed air, steam, refrigeration and electrical load per unit produced
Net content and quality, held inside the limit
Net content control is a legal obligation and a margin lever at the same time. In-Sight treats the target and the tolerance for each SKU as engineering limits with provenance, tracks the fill distribution in real time, and raises a finding when the process drifts toward either edge: the underfill that creates compliance exposure or the overfill that gives product away.
When an excursion happens, the investigation has the timeline ready: the fill trend, the changeover that preceded it, the operator round readings, and the SOP page that specifies the check. The disposition is recorded against the batch, and the monthly report carries the evidence.
Root cause, not just a Pareto
Most line analytics stops at a downtime Pareto. In-Sight starts there and keeps going. An investigation case pulls in the related findings, ranks the likely causes by evidence, and lets the copilot draft the RCA narrative with citations to your own documentation. When the same cause recurs after the fix, the platform says so, because the finding is tracked after closure rather than archived.
The knowledge stays. Every closed case is queryable, so the next engineer to see a similar pattern is shown the precedent and what was done about it.
Reports the plant manager can read, generated on schedule
Shift, daily, and weekly reports assemble themselves from live KPIs, SPC charts, findings, and AI-drafted narrative, and are generated as Word documents on a schedule and emailed to the distribution list. Dashboards bind to live tags for every tile type, from OEE and net content to a Sankey diagram of where the loss went.
Fits the systems on the line today
In-Sight connects to OPC UA, Siemens S7, EtherNet/IP, Modbus TCP, MQTT, OSIsoft PI, SAP OData, and CSV exports, so a line running a mix of PLC generations can be connected without a rip-and-replace. Operator log sheets are captured as structured rounds and join the same pipeline as the sensors. Deployment is a set of containers on one instance, inside the plant network or in a dedicated cloud environment.
What a line team gets on day one
SPC on every signal
Western Electric rules and self-recalibrating baselines by SKU and operating mode.
Net content limits with provenance
Targets and tolerances from the spec sheet, proposed to an engineer, armed on confirmation.
High-rate capture
Up to 2 kHz per tag for fast signals such as vibration, with overshoot detection at ingest.
Investigation cases
Timeline, evidence chain, ranked hypotheses, and a copilot that cites the SOP page.
Scheduled Word reports
Real charts, KPIs, and findings, generated and emailed on the schedule you set.
Work orders and SAP write-back
Dispatch the fix with the evidence attached and track recurrence after closure.
Questions we are asked
- Can In-Sight work with a line that has PLCs from several generations?
- Yes. Connectors cover OPC UA, Siemens S7, EtherNet/IP, Modbus TCP, MQTT, OSIsoft PI, SAP OData, REST APIs, and CSV files, and each source is monitored for connection health and data gaps.
- How does In-Sight handle net content control?
- Each SKU's target and tolerance are treated as engineering limits with provenance. Fill weight or volume is tracked in real time, findings are raised on drift toward either edge, and excursions are investigated and reported against the batch.
- Is this an OEE tool?
- OEE is one of the KPIs In-Sight computes and reports, but the platform is built to find the causes behind the losses rather than only to count them. Every finding can become an investigation, a work order, and a verified fix.
- How much of the setup is manual?
- Connecting a source and backfilling history is configuration, not code. Baselines are learned from the data. Engineering limits are proposed from your spec sheets and SOPs and confirmed by an engineer, which is the one deliberate manual step.
Bring one line. We will show you where the loss is going.
A filler, a packer, or a process unit with a number you do not trust. One session, on your own data.