What Is an Endpoint?

Endpoints are rapidly increasing, due in part by the proliferation of the internet of things (IoT). In some instances, so much data is collected by an IoT endpoint device that there can be challenges associated with transmitting to the datacenter for processing. To address these issues and maximize data insights, endpoint devices are increasingly empowered to “think” for themselves. Endpoint AI has emerged as a way to make even the smallest endpoints intelligent and capable of performing some degree of real-time analytics.

By processing the data they collect without moving it (or processing the data to the degree to which only the relevant information need be transmitted), allows organizations to save bandwidth and:

• Minimize latency by avoiding the need to wait for responses from the cloud.
• Improve overall system performance by processing data locally.
• Reduce the power budget and increase battery life.
• Reduce reliance on the cloud and increase autonomy.
• Increase privacy and security by avoiding the transmission of data between systems.
• Reduce cost by using less network bandwidth.

For many use cases, intelligence is best distributed across the network, all the way to the endpoints. As AI becomes more mainstream, so does the range of applications and opportunities businesses seek to develop. Endpoint AI use cases often fall into three main categories: Vision, such as object detection, classification and recognition; Voice, such as keyword detection and natural language processing; and Vibration, such as machine monitoring and sensor fusion.