Knowledge base / How the platform works

How the platform works

From a sensor on the line to a number you can defend: this is the path your production data takes through Augos, end to end.

Overview

Augos turns raw machine movement, and the way your team responds to it, into decisions. A sensor counts every cycle, a terminal captures how operators and maintenance handle every stop, the readings sync to the Augos cloud, each stop is decomposed and benchmarked against your line's own best, and it's all presented through a set of reports. Every figure traces back to a measured stop, so it can be checked rather than trusted.

From sensor to insight

There are four stages: the hardware measures, the platform ingests, the engine decomposes and benchmarks, and the reports present. Each stage keeps the underlying event intact, which is what makes the output auditable.

  • Measure, the pulse sensor counts every cycle on the line, and the operator terminal captures the response to every stop.
  • Ingest, readings sync to the Augos cloud, hosted in South Africa. The terminal stores everything on board and syncs when the network returns.
  • Decompose & benchmark, each stop is split into its phases, attributed to an owner, and measured against this line's own proven best.
  • Present, the reports turn the priced loss account into dashboards and analyses, all reconciling to your OEE.

The measurement layer

Measurement is the foundation. One sealed pulse sensor mounts onto the line and counts product as it passes, with nothing wired into your PLC, so it works on any machine, old or new. The operator terminal sits where the operator works: they clear a stop or call maintenance through it, which is how the human side of every stop gets timed. That human response is the part standard OEE never sees.

Because the terminal captures the human side, the same data that scores a line can name who can recover the time. More on the hardware →

The engine: decomposition and benchmarking

This is where raw stops become a priced loss account. Every unplanned stop is split into its five phases, operator evaluation, operator resolution, technician response, active wrench and inactive wrench, and each phase is attributed to the person or function who can recover it. Planned stops are handled differently: they are measured on their allocated time and on operator compliance to that plan, not broken into the five phases. Then each is measured against a benchmark drawn from the line's own best non-outlier stops, never a textbook standard. It's what lets the platform show that a forty-minute breakdown wasn't the repair, and exactly how much of it is recoverable. More on the five phases →

The intelligence layer

On top of the priced account sit the reports: the Executive Summary verdict, Performance Overview, the planned and unplanned stoppage analyses, operator analysis, the records and the tools. Each is a different view of the same measured source, so the numbers reconcile across all of them, down to the individual event. Browse the reports →

Data, export and AI-readiness

Nothing is locked in. Every report exports as raw or structured data, including an AI-ready payload an external agent can read directly, and every figure traces back to the raw machine pulse it came from. The data is retained and remains yours. More on AI-ready data →

Step 1 of 3 · Your details