predictive maintenance
condition monitoring

The First

scalable, autonomously configuring real-time learning product

for any industry, asset, or environment

single sensor

Step 1: Tuning

Automatically adjust ML parameters to the characteristic of the input signal

Step 2: Training

Uses millions of ML inferences to build a model of the input signal including all normal variation in minutes

Step 3: Running

Autonomously shifts into dtection mode where it looks for all anomalies including transient, drift, and shift anomalies in real-time without downsampling

sensor fusion

What is it?

Sensor Fusion is the process of fusing multiple data sources (vibration, temperature, pressure, etc.) into a single high-dimensional time series vector which is then fed into Amber to learn the normal variation of the data sources in relation to one another.

When is it useful?

Sensor Fusion is useful when there are many (up to hundreds) of sensors connected to the same underlying asset or system.

higher level of analytics

Colored alerts to notify when an asset is changing and needs to be scheduled for maintenance or if there is imminent failure and needs immediate attention

Amber can run at the edge on-prem, and in the cloud




far edge

Cloud-to-cloud access to supplement existing infrastructure with a higher level of analytics with real-time smart alerts.

FPGA deployment at the source of data perfect for environments that produce too much data to send to the cloud or for environments that require the highest level of data security.

True edge deployment on the ARM Cortex M7 allows tuning, training, and detection to occur without downsampling at the device level. A cloud-uplink can optionally occur when an anomaly is detected.