Connhex AI

Leveraging state of the art models, Connhex AI forecasts future values based on current and past sensor readings.

It features a variable prediction period - how far into the future readings should be evaluated - and continuous prediction quality monitoring.

Connhex AI is fully integrated with Connhex Cloud: for every timeseries stored, a new virtual sensor can be created. After an initial calibration period, this will start publishing predictions associated to the original connectable: this coupling makes comparing actual and predicted values seamless.

Drastically reduce time-to-market for your new products by developing functionalities on top of an industrial-grade backbone. Sounds familiar, right?

This is a common theme throughout Connhex - and applies to Connhex AI too. You get out of the box automated model selection, optimization and training - with auto-deployment on performance improvement.

And, since we are pessimists by nature, rollback is included too.

Anomaly detection is challenging by definition - anomalies are rare, making it impossible to gather large quantities of data.

This is why Connhex AI automatically detects anomalies in collected data, with an adjustable monitoring window size. It also includes customizable multi-event handling strategies: what should happen if multiple anomalies occur in a given period?