We present SensorLM, a new family of sensor-language foundation models that bridges the gap between sensor signals and human language. We will discuss the hierarchical pipeline for automatic sensor-text data annotation, the pre-training methods, and the model's zero-shot and few-shot learning, cross-modal retrieval, and caption generation capabilities. We will explore how SensorLM translates complex sensor signals into meaningful insights, opening the door to a new generation of personalized health and wellness applications.