Which AWS service allows devices to act on their generated data locally and run predictions using ML models?

AWS IoT Greengrass is designed specifically to enable devices to act locally on the data they generate while integrating seamlessly with machine learning models. It extends AWS cloud capabilities to connected devices, allowing them to execute predictions and carry out actions based on local data without always needing to send that data back to the cloud.

With AWS IoT Greengrass, users can deploy machine learning inference capabilities directly onto their devices, which helps reduce latency and dependency on constant internet connectivity. This feature is particularly useful for applications that require real-time decision-making, such as in scenarios with edge computing, where immediate responses are critical.

The other services mentioned serve different purposes. AWS IoT Core is focused on connecting and managing devices, while Amazon SageMaker is primarily a fully managed service for building, training, and deploying machine learning models in the cloud. Amazon Comprehend provides natural language processing capabilities but does not directly relate to running ML predictions on devices.