One IoT platform for two plants and 150+ machines

Our client is a mid-sized auto components manufacturer with two plants. They ran a mix of legacy and newer CNC machines, presses and PLCs from several vendors. Machine data was locked in controllers that spoke different protocols, unreliable shop-floor networks left gaps in what was collected, and a custom build had been estimated at more than a year.

What we built

We implemented the platform on our Industrial IoT Accelerator, reusing pre-built modules for connectivity, edge processing, storage, alerting and device management. That cut implementation time, and the effort went into plant-specific configuration instead of core development.

More than 150 machines, PLCs and meters across both plants are connected through a protocol connector layer that supports 15+ industrial protocols, including OPC UA, Modbus TCP/RTU and MQTT.

An edge layer with store-and-forward buffering keeps collecting data during network outages and syncs it automatically once the connection is back. From the edge, more than 400,000 telemetry messages a day (over 12 million a month) stream into a cloud layer that stores the complete historical record for trend analysis and reporting.

A configurable rule engine raises alerts on conditions in the telemetry. Edge devices and connectors get over-the-air software updates, so nobody has to visit the site to update them.

The platform runs on AWS IoT Greengrass, AWS IoT Core, Amazon Kinesis, AWS Lambda and React.

The outcome

Pre-built accelerator components significantly cut implementation time, so the work focused on the plants' own setup.

Edge buffering meant no telemetry was lost during network outages, which gave the client a complete and trustworthy history for every machine.

The serverless, pay-as-you-go cloud model ties running costs to actual usage, with no upfront infrastructure. New machines and plants are added through configuration.